Home

Background Material for SLURP . TOPAZ

1. V r smarty C J C A Federer and A L Schloss 1998 Potential evaporation functions compared on US watersheds possible implications for global scale water balance and terrestrial ecosystem modeling Journal of Hydrology 207 147 169 Watermark Numerical Computing 2000 PEST model independent parameter estimation jdoherty gil com au 55 WMO 1964 Guide to hydrometeorological practices Report No 168 World Meteorological Organization Geneva WMO 1986 Intercomparison of models of snowmelt runoff Operational Hydrology Report No 23 World Meteorological Organization Geneva Wyss J E R Williams and R L Bras 1990 Hydrologic modelling of New England river basins using radar rainfall data Journal of Geophysical Research 95 D3 2143 2152 56
2. Granger R J 1995 A feedback approach for the estimation of evapotranspiration using remotely sensed data In Applications of Remote Sensing in Hydrology Proceedings of the Second International Workshop 18 20 October 1994 by G W Kite A Pietroniro and T Pultz eds Symposium No 14 NHRI Saskatoon Saskatchewan 211 222 Grayson R B I D Moore and T A McMahon 1992a Physically based hydrologic modelling 1 A terrain based model for investigative purposes Water Resources Research 28 10 2639 2658 Green W H and G A Ampt 1911 Studies in soil physics 1 Flow of air and water through soils J Agric Science 4 1 24 Institute of Hydrology 1995 Assessment of global water resources preliminary report Report to the Overseas Development Administration ODA Report 95 2 Institute of Hydrology Wallingford 38pp Jain S K B Storm J C Bathurst J C Refsgaard and R D Singh 1992 Application of the SHE to catchments in India Part 2 Field experiments and simulation studies with the SHE on the Kolar subcatchment of the Narmada River J Hydrology 140 25 47 Jakeman A J and G M Hornberger 1993 How much complexity is warranted in a rainfall runoff model Water Resources Research 29 8 2637 2649 Jenson S K and J O Domingue 1988 Extracting Topographic Structure from Digital Elevation Data for Geographic Information System Analysis Photogrammetric Engineering and Remote Sensing 54 11 1593 16
3. With the Granger 1995 method which uses surface temperature in a feedback relationship with vapour deficit In this method surface temperature may be obtained from remotely sensed data A net radiation algorithm may also use data from channel 1 of the NOAA advanced very high resolution radiometer AVHRR as an indicator of albedo Daily evapotranspiration at each 1 km pixel is computed from net radiation and vapour pressure deficit and averaged over each type of land cover Daily wind data are optional with this method Using the Spittlehouse 1989 technique in which available energy is calculated using the Priestley and Taylor 1972 approach and the available soil moisture is calculated as a func tion of the field capacity and root zone depth Evaporation from the canopy interception and the soil and transpiration from the vegetation are computed separately in this method Using the FAO version of the Penman Monteith Verhoef and Feddes 1991 algorithm In SLURP a version of the implementation in the SWAP model van Dam et al 1997 is used This implementation computes bare soil and dry crop potential evapotranspiration and requires information on crop height and canopy resistance Daily wind data are required for this option Using the Linacre 1977 approximation to the Penman equation This method uses only temperature data Clemence and Schultze 1982 found that the Linacre method performed better than other temperature based met
4. and A Henderson Sellers 1988 Modelling tropical deforestation A case study of GCM land surface parametrizations Quarterly Journal of the Royal Meteorological Society 114 439 462 50 Douglas D H 1986 Experiments to Locate Ridges and Channels to Create a New Type of Digital Elevation Models Cartographica 23 4 29 61 Droogers P G W Kite H Korkmaz and O Acar 1998 The role of remote sensing in integrated basin modelling In Applications of Remote Sensing in Hydrology Proceedings of the Fourth International Workshop Santa Fe 4 6 November 1998 by G W Kite A Pietroniro and T Pultz eds NWRI Saskatoon Saskatchewan Droogers P G W Kite amp W G M Bastiaanssen 1998 Land cover classification using public domain datasets an example for the Gediz Basin Turkey Proc Int Symposium on Arid Region Soils Izmir Turkey 21 24 September 34 40 Duan Q Sorooshian S Soroosh and V K Gupta 1994 Optimal use of the SCE UA global optimization method for calibrating watershed models J Hydrology 158 265 284 Fairfield J and P Leymarie 1991 Drainage Networks from Grid Digital Elevation Models Water Resources Research 30 6 1681 1692 Fread D L 1985 Channel routing Chapter 14 in Hydrological Forecasting M G Anderson and T P Burt eds John Wiley NewYork Garbrecht J 1988 Determination of the Execution Sequence of Channel Flow for Cascade Routing in a Drainage Network Hydrosoft Softwa
5. such as time series of precipitation and air temperature areally distributed data such as topography and land cover and finally model parameters and coefficients such as albedos and infiltration rates To include modifications to the natural regime additional data will be needed Of course the model uses digital data but the first step in any basin study is to collect paper Maps at an appropriate scale usually 1 100 000 or better are needed showing the topography the river distribution the locations of important cities river diversions and water input extraction points and the locations and sizes of irrigation schemes lakes reservoirs and any other features to be modelled The topography and river network will be used as checks on the automatic river network derived in the model all digital elevation data contain errors and the locations of cities irrigation schemes lakes dams diversions and water input extraction points will be used to site these features in the hydrological model Maps showing the locations of climate stations meteorological and hydrometric stations may be needed to check the positions and elevations of these stations contained in the corresponding digital data not all information in data files is correct Finally information on the operating rules of dams and reservoirs the crop pattern and watering rules of irrigation schemes and any diversions or consumptive uses will be needed The next step is to collect cli
6. 0 3 So 0 6 ah 31 c 1 27p 0 0 where depends on the cross section shape and is assumed to be 5 3 for a rectangular channel qo is a unit width reference discharge and n is the Manning roughness factor The weighting constant X 0 lt X lt 1 is calculated as X 0 5x 32 cS Ax where So is the channel bottom slope The data needed to rout flows using the Muskingum Cunge method are input manually on screens such as Channel Routing Data and Flowpaths Diversions amp Interventions input screens 2 3 3 Lake Reservoir Routing Regulation Many river basins contain large lakes which modify the hydrograph of the river If the stage storage and stage discharge relationships are known then these may be used as an additional routing at the end of a sub basin Similarly if a sub basin contains a dam and reservoir from which the outflows are controlled by a regulation plan then this effect may also be simulated in SLURP Dam locations may be stored in file STRCTURS INP and then converted to an IDRISI Vector Export Format using option Tools Convert INP file to VXP file 25 The Vector Export file may then be imported to IDRISI and used as a vector overlay on a basin image Figure 8 Figure 8 STRCTURS overlay showing proposed Beydagi reservoir site K k Menderes Basin Turkey Routing may be carried out either internally within the SLURP program or externally in a user written program T
7. 100 81 92 Rango A And J Martinec 1995 Revisiting the degree day method for snowmelt conditions Wat Res Bull 31 4 657 669 Refsgaard J C S M Seth J C Bathurst M Erlich B Storm G H Jorgensen and S Chandra 1992 Application of the SHE to catchments in India Part 1 General results Journal of Hydrology 140 1 23 Richards L A 1931 Capillary conduction of liquids through porous mediums Physics 1 318 333 Sabourin J F amp Associates 1996 Implementation of a distributed hydrologic model Using SLURP on the Carp watershed CCRS Ottawa 43pp 54 Schultz R E et al 1989 ACRU Background concepts and theory Report 35 Agricultural Catchments Research Unit Department of Agricultural Engineering University of Natal Pietermaritzburg 3201 South Africa Shuttleworth W J 1993 Evaporation In Handbook of Hydrology D R Maidment ed McGraw Hill New York 4 1 4 53 Slough K and G W Kite 1992 Remote sensing estimates of snow water equivalent for hydrological modelling applications Canadian Water Resources Journal 17 4 1 8 Smith M 1992 CROPWAT a computer program for irrigation planning and management FAO Irrigation and Drainage Paper 46 FAO Rome Spittlehouse D L 1989 Estimating evapotranspiration from land surfaces in British Columbia In Estimation of Areal Evapotranspiration ZAHS Publication No 177 245 253 Stuttart M J J B Hayball G Narcisco M Suppo an
8. IDRISI images READPET converts limate dat SLURP distributed SLURP bonne ee i f outputs to IDRISI lt hydrological 4 shat calibration or images model verification only Figure 14 Sequence of topographic analysis routines Model Parameters and Coefficients A hydrological model attempts to simulate the physical processes involved in the transformation of precipitation into evapotranspiration and runoff Model parameters control the way in which this transformation is carried out through such processes as interception infiltration and snowmelt Parameters include maximum capacities for the canopy detention store and slow store initial contents of the snowpack and fast and slow stores initial contents of the snowpack and fast and slow stores interception coefficients and maximum and minimum leaf area index LAI surface albedo and maximum soil heat flux snowmelt rates and energy conversion factors saturated infiltration rate conductivity roughness coefficient for each land cover and for the channel within the ASA lapse rate for temperature and precipitation elevation adjustment rate parameters for specific evapotranspiration methods such as windspeed function wilting point field capacity and Priestley Taylor coefficient 43 Table 11 summarizes typical values of some of the parameters used in SLURP Model parameters may be obtained from field measurements although
9. channel paths unless both happen to coincide Also the geometric characteristics of a subcatchment cannot be accurately estimated if the subcatchment size is small and the effects of the finite raster representation significantly impacts the geometric measure under consideration Assumption b The drainage direction at a single cell can be in error by a much as 22 5 degrees However in dissected landscapes with convergent drainage patterns the general drainage direction over a distance of several cells is relatively well reproduced because drainage paths are defined by topography and not by flow directions at a single cell Drainage direction for plane undissected hillslopes are the most susceptible to larger approximation errors Assumption c The drainage over the digital landscape is only suitable for flow analysis using forward marching routing schemes i e kinematic cascade approach and runoff conditions without significant backwater or tidal effects Assumption c also allows for only one downstream path out of each cell Therefore divergent and braided drainage patterns cannot be modelled Assumption d The critical source area CSA defines the beginning of source channels 1st order channels However the latter often display random and systematic variations due to spatial variations in topography soil geology and vegetation among others The CSA concept cannot account for random variations However systematic spatial
10. comparison of eleven evapotranspiration methods that errors in simulating streamflow were caused much more by problems with precipitation and snowmelt than by differences in estimation of potential evapotranspiration For all methods except Linacre the net radiation is computed within the model from recorded global radiation or from recorded hours of bright sunshine and computed extraterrestrial radiation as follows e The direct beam clear sky radiation Qaro is calculated Granger and Gray 1990 from On p cos XAS dh o where J is a solar constant 1 35 kW m r is the radius vector of earth s orbit p is the mean transmissivity of the atmosphere along the zenith path M is the optical air mass cos XAS is the cosine of the angle of incidence of the sun s rays on the surface and A is the hour angle of the sun The integral is taken over the duration of sunlight In the Penman Monteith method direct beam clear sky radiation Qaro is computed using a simpler approximation e Incoming radiation is then reduced to account for cloud cover Qars using Qin _ 9 9244 0 974 35 2 dro where n is the actual number of hours of bright sunshine in the day and N is the maximum possible number of hours of bright sunshine in the day The clear sky diffuse radiation Q is obtained from 2m 172 day P 3 5 cos XAS 0 45 sin Qao G 365 3 where P is the pressure at the surface Po is the standard press
11. data needed for routing regulation SLURP12_ RUT should carry out the reservoir routing or regulation and write the results to data files SLURP will then read the results of the routing regulation and continue with the rest of the simulation External lake or reservoir routing regulation for an ASA is specified in the option Edit Flow paths diversions interventions on the main SLURP menu 29 2 4 Diversions and Interventions 2 4 1 Diversions In many basins regulatory structures exist to divert water from the river for hydropower water supply or irrigation see Figure 10 To specify in SLURP that an ASA has diversions use the input screen in the Flowpaths Diversions amp Interventions input window and prepare a lt Command_name gt DIV file with the diversion specifications In this file specify the name of the ASA at which the diversion takes place there may be an unlimited number of diversions at any ASA the names of the file s to contain the diverted water s and the details of each diversion Table 7 below gives a simple example Table 7 Example ofa DIV diversion file for the Sel uk sub basin K k Menderes Basin selcuk 1 35 selcuk selcuk exr selcuk 1994 4 1 1994 4 30 0 104 0 0 selcuk 1994 5 1 1994 5 31 1 772 0 0 selcuk 1994 6 1 1994 6 30 4 956 0 0 selcuk 1994 7 1 1994 7 316 552 0 0 selcuk 1994 8 1 1994 8 31 6 520 0 0 selcuk 1994 9 1 19949 309 60 0 0 selcuk 1994 10 1 1994 10 31 0 652 0 0 Diversions fr
12. evaporation Table 1 Comparison of evapotranspiration methods Method Advantage Disadvantage Penman Monteith Widely accepted Data intensive Priestley Taylor Separate E and T Uses coefficients Morton CRAE Better logic Not widely accepted Granger Uses satellite data Data intensive Linacre Uses only temperature data Approximation only 2 2 1 DESCRIPTION OF THE SLURP MODEL The Vertical Water Budget 2 1 1 Evapotranspiration At a particular time step the first operation within the vertical water balance for a particular land cover and sub basin ASA is to compute the evapotranspirative demand As noted earlier potential evapotranspiration may be computed in any of five different ways a b c d e Using the complementary relationship areal evaporation model CRAE Morton 1983 CRAE computes a potential evapotranspiration by solving the energy balance and aerodynamic equations at equilibrium temperature using a modification of the Penman 1948 equation in which the wind function is replaced with a vapour transfer coefficient Daily net radiation is computed from either recorded hours of bright sunshine or recorded global radiation A wet environment evapotranspiration is computed from the slope of the saturation vapour pressure temperature curve and the net radiation The actual areal evapotranspiration is then computed as the difference between twice the potential and the wet evapotranspiration
13. gov cgi bin res40 pl GDS is also available from the IWMI Climate Atlas and from the USGS Hydrolab website although the format of the latter is uncertain and GSOD is also available from the University of Miami website GDS contains daily maximum and minimum temperatures and daily precipitation from over 10 000 stations for the period 1977 1991 in metric units GSOD contains 13 parameters for over 8000 stations for the period 1994 to date is updated on the web every month and is in imperial units The SLURP CD ROM contains programs to extract and process GDS and GSOD data sets GDS has neither radiation nor humidity data and GSOD does not have radiation data Approximations for humidity and radiation described in the manual and included in programs included on the CD ROM may be used to complete missing data sets Observed daily streamflow data FLO files are optionally used in the SLURP model for calibration These may be obtained from hydrometric station records in the country concerned or sometimes from websites such as ORNL http www eosdis ornl gov Long term mean daily 40 streamflow LTM files are optionally used to compute the Garrick performance statistic Additional streamflow data FLI files may be used to include inflows from sources such as inter basin transfers or groundwater originating from another basin These inflows can also be used in cases where SLURP is used to model only part of a basin and the inflows are
14. gt observed i Computed flow mm 9 580E 02 Figure 9 Hydrograph after passing through the reservoir routing procedure 27 The SLURP basin model simulates the effect of routing and regulation for a lake or reservoir using the following steps for each day 1 Compute the initial lake surface area and volume from an lt ASA_name gt LVL file containing recorded daily levels 2 Add the reservoir inflow and over lake precipitation to the existing volume 3 Subtract the over lake evaporation from the volume 4 Compute the new reservoir level and check for spill if the level exceeds the allowable maximum 5 Check the regulation rules in file lt command_ name gt RUT to see if a release should be made today 6 Make the release if appropriate and if the reservoir level is above the minimum allowed 7 Recompute the volume and new level The example above is for a reservoir in which the outflow follows a series of specified releases If the water body is a naturally regulated lake then the outflow will not be specified but will depend on the lake level In this case the first record of the RUT file would appear as for example beydagi N 0 indicating that Beydagi is a lake with natural routing Following the stage area and stage volume parameters the user would include in the same way the parameters of the applicable stage discharge curve Discharge a level b 35 in which discharge is in m s the volume is in
15. is added to the slow store which then generates groundwater flow RG as RG S 23 where kz is the slow store retention constant and S gt is the current contents of the slow store If overflow from the slow store occurs it will be added to the interflow component RZ Transpiration is extracted from the slow store The concepts used in the vertical water balance are simple and use parameters which may be readily estimated Many studies e g Naef 1981 Jain et al 1992 Grayson et al 1992a Grayson et al 1992b Jakeman and Hornberger 1993 have demonstrated that the use of more sophisticated models making use of field and laboratory measurements of infiltration and groundwater parameters does not provide improved results 2 1 3 Vertical Water Balance for a Water Land Cover SLURP can handle a water land cover within a sub basin ASA in two different ways If the water is a lake or reservoir for which regulation or routing is specified on Figure 5 then the vertical water balance for the water within the ASA will be carried out as part of the regulation or routing as described in section 2 3 3 19 If on the other hand the water in an ASA represents only a small unregulated lake or even a collection of small water bodies then the vertical water balance is carried out as for any other land cover except that some of the parameters are changed to eliminate the functions of the canopy store and the fast store and to ens
16. m x 10 and a b and c are specified constants g P The SLURP model uses a modified Goodrich procedure Goodrich 1931 to solve the storage equation TEL 0 0 36 in which Z and 0 are inflows and outflows in m s respectively S is the reservoir volume in m x 10 f is the time interval in seconds and 1 and 2 indicate successive time steps Equation 36 is solved by developing an equation relating 2S t 0 to 0 from the specified stage discharge and stage volume equations Flows are then routed at successive time steps by computing the outflow from the initial reservoir level finding 2S t 0 from 0 converting to 2S t and deriving the left hand side of Equation 36 The new outflow is then found from the right hand side of Equation 36 If output for the ASA being routed is specified in the Output Options window and if the output is daily then the lake reservoir level area volume and output will be printed or saved to file lt command_ name gt PRN Table 6 28 Table 6 Example of routing output Date Level Area Volume Outflow dd mm yyyy m m2 10 6 m3 10 6 m3 s 1 9 1988 2 20065E 02 1 97054E 01 2 21355E 02 1 04000E 01 2 9 1988 2 20037E 02 1 96735E 01 2 20793E 02 1 04000E 01 3 91988 2 20018E 02 1 96529E 01 2 20431E 02 1 04000E 01 4 9 1988 2 20005E 02 1 96378E 01 2 20166E 02 1 04000E 01 5 91988 2 19993E 02 1 96247E 01 2 19935E 02 1 04000E 01 The evaporation from a lake or reservoir may b
17. network because it always has two upstream inflows per junction The binary channel network is believed to be a better approximation of natural channel networks because it overcomes the effects of systematic lumping due to limited raster resolution At the user s choice properties and parameters of the binary channel network and corresponding channel links and sub catchments are computed and tabulated in a manner similar to that for the raster channel network channel links and sub catchments In addition to computing the parameters of each channel link and subcatchment TOPAZ evaluates the mean and standard deviation of these parameters over the entire channel network as well as by Strahler order Topaz also computes the drainage network composition parameters The entire statistical evaluation is performed for both the raster and binary channel networks 4 2 5 Assumptions and Limitations General assumptions and limitations of the D8 method the downslope flow routing concept and the CSA concept as described in O Callaghan and Mark 1984 Douglas 1986 Fairfield and Leymarie 1991 and Costa Cabral and Burges 1994 apply Also applicable are assumptions and limitations pertaining to DEM resolution vertical and horizontal and finite raster representation of landscape elevations Of particular significance for TOPAZ are the following four assumptions limitations a the DEM must be of sufficient vertical and horizontal resolution to a
18. of a channel link or subcatchment are stored in tabular format Examples of channel network parameters include number of channel links total channel length subcatchment area drainage density and network composition parameters Examples of channel link parameters include link length slope elevation upstream and lateral drainage area and topologic connectivity information Examples of subcatchment parameters include representative drainage area length slope aspect distribution and overland flow distance A complete list of computed properties and parameters can be found in the documentation of individual TOPAZ programs 47 Raster derived channel networks often contain junction nodes with more than two inflows Such junction nodes are referred to as complex junction nodes Complex junction nodes very seldom occur in nature Tributaries joining a main stem generally are spaced along the main stem though the spacing between junctions can occasionally be small Complex junctions in generated raster channel networks can be common They are the result of a low horizontal DEM resolution that cannot resolve the spacing between two or more consecutive tributary junctions thus lumping multiple junctions together into a single cell A TOPAZ feature allows the decomposition of complex junction nodes into a sequence of simple junction nodes only two inflows per junction at a sub raster resolution The resulting channel network is called a binary channel
19. roughness or wind Estimated Low func Measured Estimated Medium Field capacity wilting point porosity Parameter can be included in the calibration procedure Parameter is optional 33 Many of the parameters in Table 9 together with the between ASA routing coefficients may be calibrated Other parameters are more easily measured or estimated and need not be included in any automated optimization The precipitation factor P 9 and the snowmelt temperature P 10 are redundant in most cases and should be set to 1 0 and 0 0 respectively although as noted earlier P 9 may be used to compensate for gauge undercatch and P 10 may be used to bias the precipitation distribution to snow or rain The effect of elevation on precipitation and temperature are accounted for by non calibrated rates of change per 100 meters elevation If the model run is started during the fall season when there is no snowpack and soil moistures are low then the initial contents P 1 and P 2 may be set to zero or low values and do not need to be optimized For the water land cover the only parameters needed are an initial snow water equivalent if there is a snowpack an optional precipitation factor and an optional snowmelt temperature If a snowpack is a possibility then snowmelt rates and snow albedos will also be needed The goodness of fit of the SLURP model may be judged both graphically and from statistical criteria After each year of mo
20. s manual Geographica Bernensia Report P29 University of Bern Switzerland Martz L W and E De Jong 1988 Catch a FORTRAN Program for Measuring Catchment area from Digital Elevation Models Computers amp Geosciences 14 5 627 640 Martz L W and J Garbrecht 1992 Numerical definition of drainage network and subcatchent areas from digital elevation models Computers amp Geosciences 18 747 761 Martz L W and J Garbrecht 1998 The Treatment of Flat Areas and Depressions in Automated Drainage Analysis of Raster Digital Elevation Models Hydrologic processing 12 843 855 Martz L W and J Garbrecht 1993 Automated extraction of drainage network and watershed data from digital elevation models Water Resources Bulletin 29 6 901 908 Monro J C 1971 Direct search optimization in mathematical modelling and a watershed application NOAA Technical Memorandum National Oceanic and Atmospheric Administration U S Department of Commerce Washington D C 52p 53 Monteith J L 1981 Evaporation and surface temperature Quarterly Journal of the Royal Meteorological Society 107 1 27 Morris D G and R G Heerdegen 1988 Automatically Derived Catchment Boundary and Channel Networks and Their Hydrological Applications Geomorphology 1 2 131 141 Morton F I 1983 Operational estimates of areal evapotranspiration and their significance to the science and practice of hydrology J Hydrology 66 77 100 Morto
21. the land cover distribution into percentages for each sub basin ASA The basic requirements for an ASA are that the distributions of land covers and elevations for elements pixels within the ASA are known and that the ASA contributes runoff to a definable stream channel The latter requirement is also an operational consideration since it means that the stream system within the basin must be at a level of detail such that each ASA contains a defined stream connected to the basin outlet This is achieved by using digital topographic maps with a stream layer such as the National Topographic Service 1 50000 and 1 250000 series by digitizing the stream network as a series of vectors off a topographic map or preferably by using an automated delineation of the channel network from digital elevation data with a program such as TOPAZ Martz and Garbrecht 1992 Garbrecht and Martz 1993 Martz and Garbrecht 1993 Garbrecht and Martz 1997 Garbrecht and Campbell 1997 The stream network may then be stored as a vector layer in a geographic information system GIS or saved by the topographic analysis package as a file for use in further analyses The number of sub basins ASAs used in modelling a basin will depend on the size of the basin and the scales of data available The SLURP model may use an unlimited number of ASAs and an unlimited number of land covers per ASA potentially dividing a basin into very many sub areas To be sure of stability whe
22. then the following is needed e daily wind speed Optionally the following data from satellite may be used e daily cloud cover e daily snow cover e snow water equivalent from satellite or from snowcourses If the basin contains large lakes or reservoirs which must be simulated in the model then the following will be needed e observed lake or reservoir levels All of the point or station data are stored in separate files for each data type Each file is given a name made up of the station name and a three character extension as lt Station Name xxx gt where the extensions xxx are listed in the following table for each type of data The most important data types needed are the PCP TAV TDP or RHU and GLO or SUN All station data files are sequential and unformatted Data files for different basin studies should be stored in separate directories Any data needed specifically to control the evapotranspiration computations are held in separate files for each evapotranspiration method For Morton the mean annual precipitation and the mean lake depth when needed for computation of lake evaporation are stored in file lt Command_ name MOR gt For Spittlehouse Black the Priestley Taylor coefficient beta and the field capacity and wilting point for each land cover type is held in the file lt Command_name SB gt Granger evapotranspiration uses a file lt Command_name GRA gt that contains an aerodynamic resistance or a windspeed fun
23. watersheds Research over the past decade has demonstrated the feasibility of extracting topographic information directly from raster Digital Elevation Models DEM In the field of water resources and hydrology automated evaluation of DEMs has focused on watershed segmentation definition of drainage divides and identification of the drainage networks This automated extraction of network and sub watershed properties from DEMs represents a convenient and rapid way to parameterize a watershed The increasing commercial availability of DEM coverage for many areas of the United States and advances in DEM processing makes an automated approach to watershed parameterization a practical and useful alternative to manual techniques Automated techniques also have the advantage of generating digital data that can be readily imported and analyzed by Geographic Information Systems GIS 44 The TOPAZ software TOpographic PArameteriZation is a digital landscape analysis tool that provides comprehensive processing and evaluation of raster DEMs to identify topographic features measure topographic parameters define surface drainage subdivide watersheds along drainage divides quantify the drainage network and calculate representative subcatchment parameters TOPAZ emphasizes consistency between all derived data the initial input topography and the physics underlying energy and water flux processes at the landscape topography The primary objective of TOPAZ
24. 0 28 Open wetlands 2 88 0 84 0 31 Snowpack in terms of snow water equivalent SWE may also be input directly to the model from snowcourse data or computed from passive microwave data Slough amp Kite 1992 It was found that for a mountain basin data from individual snowcourses are of little or no use in the model point SWE measurements are not good estimates of ASA averages SWE values computed from passive microwave data on the other hand provide better estimates because they themselves are average values from large areas e g 25 km x 25 km for data from the DMSP satellite for south eastern British Columbia In addition the microwave data are available in all weathers and at frequent sampling intervals Snowmelt may optionally be modified by the areal extent of snowcover from NOAA AVHRR images Kite 1989a The subsurface flow processes are simulated using two linear reservoirs the fast store which may be considered as an unsaturated soil layer contributing surface runoff interflow and percolation to a deeper slow store and allowing soil evaporation and crop transpiration The slow store is a source of interflow groundwater flow and transpiration Figure 4 Rainfall and any snowmelt infiltrates through the soil surface into the fast store depending on the current infiltration rate Si URUS Inf rax 19 1 max where S is the current contents of the fast store S7 max is the maximum capacity of the store and Inf
25. 00 Kite G W 1975 Performance of two deterministic models In Application of Mathematical Models in Hydrology and Water Resources Systems Proceedings of the Bratislava Symposium September 1975 JASH Publication No 115 136 142 Kite G W 1978 Development of a hydrological model for a Canadian watershed Canadian Journal of Civil Engineering 5 1 126 134 Kite G W 1989a Using NOAA data for hydrological modelling In Quantitative Remote Sensing An Economic Tool for the Nineties IGARSS 89 12th Canadian Symposium on Remote Sensing IEEE 10 14 July 1989 Vancouver British Columbia 553 557 Kite G W 1989b Hydrological modelling with remotely sensed data In Proceedings of the 57 Annual Western Snow Conference 18 20 April 1989 Fort Collins Colorado 1 8 Kite G W 1991 A watershed model using satellite data applied to a mountain basin in Canada J Hydrology 128 157 169 Kite G W 1993 Application of a land class hydrological model to climatic change Water Resources Research 29 7 2377 2384 Kite G W 1995a Scaling of input data for macroscale hydrologic modeling Water Resources Research 31 11 2769 2781 52 Kite G W 1995b The SLURP model Chapter 15 in Computer Models of Watershed Hydrology ed V J Singh Water Resources Publications Colorado pp 521 562 Kite G W A Dalton and K Dion 1994 Simulation of streamflow in a macro scale watershed using GCM data Water Resources Res
26. 85 198 The routing equations for the reservoir are Area 036165 level 173 179 33 Volume 019644 level 173 74 34 In these equations the area is in m x 10 the volume is in m x 10 and the level is in meters Figure 9 shows the SLURP simulated regulated outflow from the proposed Beydagi Reservoir Such equations may be fitted to level area and volume data by using program STADIS provided in directory PROGRAMS SLURP12 UTILITY on the SLURP CD ROM If a sub basin contains a water land cover and lake or reservoir routing regulation is specified then the model compares the area of water in km within the basin from the command file with the maximum surface area of the lake or reservoir If the water area is greater than the lake area then this means that not all the water in the sub basin is subject to routing The water balance from the excess area is computed in the normal way and then the water within the lake or reservoir is routed or regulated If the water area is less than the lake area then no vertical water balance is needed only routing or regulation Slurp model output for sub basin beydagi for period 1 Oct 1994 to 30 Sep 1998 Rain amp Snow mm 23 11 200 400 600 1000 1200 1400 aoo Days from 1 Oct 1994 Rain snow i Precipitation mm 3 262E 03 Evapotranspiration mm 2 493E 03 Observed flow gt computed i There is no observed flow Rain snow temp C 0 000E 00 Computed flow
27. A concept defines the channels draining the landscape as those raster cells that have an upstream drainage area greater than a threshold drainage area called the critical source area CSA The CSA value defines a minimum drainage area below which a permanent channel is defined Mark 1984 Martz and Garbrecht 1992 The CSA concept controls the watershed segmentation and all resulting spatial and topologic drainage network and subcatchment characteristics These methods and concepts have been applied previously in DEM processing algorithms above references and Jenson and Domingue 1988 Martz and DeJong 1988 and others TOPAZ builds upon and extends that work A number of the limitations of previous work with respect to drainage identification in depressions and over flat surfaces have been overcome in TOPAZ and a number of new features have been added to address specific hydrologic and hydraulic needs Garbrecht and Martz 1996 TOPAZ is designed to provide interrelated landscape analysis functions and to rely on an external user selected GIS for image display as well as for additional data manipulation and raster algebra operations Within this general framework the digital elevation data is processed in TOPAZ by a system of interdependent computational programs The analytical operations performed by these programs achieve three broad functions 45 1 elevation data pre processing 2 hydrographic segmentation and 3 topographic par
28. Background Material for SLURP TOPAZ MAGS Model Cross Training Workshop 5 6 September 2002 York University Prepared by Dr Lawrence Martz Department of Geography University of Saskatchewan Based on Manual for the SLURP Hydrological Model Version 12 2 Geoff Kite HydroLogic Solutions http www hydrologic solutions com January 2002 Overview of TOPAZ an automated digital landscape analysis tool for topographic evaluation drainage identification watershed segmentation and subcatchment parameterization Jurgen Garbrecht and Lawrence W Martz Rep GRL 99 1 Grazinglands Research Laboratory USDA Agricultural Research Service April 1999 il TABLE OF CONTENTS Introduction sos Kasi sor arene aash unio a reibiacd moa asoce a a aea Description of the SLURP Model ccccc ccccsssscrsscssssncecsesenssesessssnenesess 2 1 The Vertical Water Budgets Gocicaaicacca hed ceetetiasescrs ees eteseed 2 1 1 EVADOIMAnSPIRAtlONs orig beeen ele acne eae 2 1 2 Vertical Water Balance for all Land Covers except Water 2 1 3 Vertical Water Balance for a Water Land Cover 2 2 Routing within a Sub basin ies caccs alent isacenisdes setacditeasisassncdoieenecs 2 3 Routing between Sub basins eee eceeeccceeseeeeeeeeeeneeeeeneeeeeeeees 2 3 1 Muskingum Channel Routing 0 cccccceeeeseceeeeteeeeeeeees 2 3 2 Muskingum Cunge Channel Routing 0 ceeceeeeeteees 2 3 3 Lake Reservoir Routing Regul
29. K KX 4At 2 C C 0 5 q q AxAt C The values of x and K are specified for an ASA and will then be used to route flows from other ASAs through that ASA The parameters are input on a screen in the SLURP sub basin routing input window K is measured in days and has an upper limit set to 9999 and x may vary from 0 to 0 5 In this implementation of the Muskingum method C local inflow is not used The effects of some typical values of x and K are as follows x 0 0 K 0 outflow hydrograph is the same as the inflow hydrograph x 0 0 K 1 outflow hydrograph is lagged by 1 day but flows are the same x 0 25 K 1 outflow is lagged by 1 day and the peak is reduced by about 10 If the TOPAZ SLURPAZ programs are used to derive SLURP inputs then K will default to 0 25 and K will be computed from the change in elevation down the channel D as Institute of Hydrology 1995 aloe 29 24 2 3 2 Muskingum Cunge Channel Routing The more exact Muskingum Cunge channel routing method may be used if channel characteristics data such as length change in elevation average width depth and roughness are available This method assumes a single valued depth discharge relationship and uses the classical kinematic wave equation Fead 1986 In this implementation the storage constant K is computed as K 4x c 30 where c is the kinematic wave speed and Ax is the channel length The kinematic wave speed c is computed from
30. a part of such an unformatted space delimited ASCII data file listing daily precipitation in mm at a particular climate station for the period 8 14 October 1994 1994 10 8 0 00 1994 10 9 0 00 1994 10 10 0 11 1994 10 11 2 62 1994 10 12 12 81 1994 10 13 0 00 1994 10 14 0 96 All data files must be for the complete period of the intended model run including days with missing data specified as 99999 99 For each climate station obtain and retain the station location either as latitude and longitude or as UTM coordinates and obtain the station elevation The sets of data observed at climate stations or simulated by atmospheric models will be converted by SLURP to climate data sets for each sub basin or ASA The model uses the Thiessen polygon method with weights computed from the DEM in the topographic analysis routines The differences in elevation between the climate stations and the sub basins are also used to correct the climate station data Traditionally hydrological models have depended on locally collected climate data However in many countries such data are becoming difficult to obtain either because of lengthy processing times or high data costs Instead we can obtain many data from databases held on the Internet Currently there are two main global datasets of daily climate data GDS Global Daily Summary and GSOD Global Surface Summary of the Day both provided by NOAA from NCDC data collections at website http www ncdc noaa
31. actually outflows from an upper part of the basin either observed or computed with some other model 3 2 Topographic Data and Land Cover Data SLURP requires information on slopes and distances to move water from the top to the bottom of the basin Raw elevation datasets digital elevation models DEMs such as ETOPOS which provides 5 minute resolution data and GTOPO30 HYDROIK which provide 30 second gridded elevation data for the world are manipulated by SLURP s ancillary topographic analysis routines to yield the derived inputs required by the SLURP model The easiest way to obtain the DEM is to download the tile for the basin area from an Internet site http edcwww cr usgs gov landdaac gtopo30 gtopo30 html such as that for the USGS GTOPO30 DEM After uncompressing the file may be imported into IDRISI or a similar GIS After extracting a window covering the basin the DEM may be exported from IDRISI in the ASCII flat file format required by the SLURP topographic analysis routines The full procedure is described in detail in the SLURP manual all the user needs to know at this stage is how to download the raw data The following figure shows an example of a DEM for the K k Menderes Basin in western Turkey Metres 2000 1750 1500 1250 1000 750 N h 500 250 20 40 km 0 0 Figure 12 DEM for the K k Menderes Basin in western Turkey At the same time the topographic analysis routines use land cover raste
32. ameterization 4 2 2 Elevation Data Pre processing DEMs commonly contain localized depressions and flat surfaces most of which are artifacts of the horizontal and vertical DEM resolution DEM generation method and elevation data noise These features are problematic for the downslope flow routing concept on which many DEM processing models including TOPAZ are based Depressions are sinks at the bottom of which drainage terminates and flat surfaces have indeterminate drainage Therefore TOPAZ pre processes the input DEM to rectify these features and allow the unambiguous determination of the drainage over the entire digital landscape Rectifications made during DEM pre processing are strictly limited to cells of depressions and flat surfaces so as to minimize the impact on the overall information content of the elevation data Among the innovative capabilities provided by TOPAZ is the ability to distinguish between two types of depressions sink depressions and impoundment depressions Sink depressions are caused by a group of raster cells at lower elevation than the surrounding landscape and impoundment depressions are caused by a narrow band of raster cells of higher elevation across drainage paths similar to an obstruction or dam across a stream In the latter situations TOPAZ can lower selected DEM elevation values to simulate breaching of the obstruction or dam across the drainage path This removes or reduces the size of the impoundment d
33. asin transfers groundwater originating from another basin or outflows from an upper part of the basin either observed or computed with some other model are accumulated with the runoffs from all the land covers within the ASA and with any return flows from upstream irrigation projects see section 2 4 The resulting flow must be routed from the outlet of this ASA to the outlet of the next ASA down the stream system and so on until all the flows arrive at the outlet of the basin The user 23 may choose to simply accumulate the flows from each ASA down the basin with no delay or attenuation select N for no routing on the input screen for sub basin routing diversions and interventions or may choose between Muskingum routing select M or Muskingum Cunge routing C The linkages between ASAs are also specified on the input screen 2 3 1 Muskingum Channel Routing The Muskingum method developed by McCarthy in 1938 quoted in Linsley et al 1949 is a simple method of hydrological storage routing which makes use of two constants the relative weight of inflow and outflow in computing channel storage x and the time of travel along the channel reach K to solve the storage equation The outflow at time 2 O2 is computed from OC ACL COFO 27 where J I2 and O are inflows at times 1 and 2 and outflow at time 1 respectively and C7 C2 C3 and C4 are constants defined by C KX Vt 2 C KX At 2 C C KX A12 C 28 C
34. ation cccceeeseeeeeeeees 2 3 4 External Routing Regulation ccccccceseesseceeeeeteeeeeeeaes 2 4 Diversions and Interventions 0 cccccecesccceeseteeeeeeeeneeeeseeeeeeneeaes DA lt DIV CESIOUS AE EE TTE lea aaa Ponees 2 4 2 Interventions in the Vertical Water Balance cce 2 5 Calibrating the parameters 4 5 lt ccccsaccshetsnssuceeheseatendiaeantuauctonshiss Data Requirements for the SLURP Hydrological Model Version 12 2 3 1 Point Station VALE ses a ag tose wo caso eae aa ooo 3 2 Topographic Data and Land Cover Data eeeeeseeeeesteeeenneeeeees 3 3 Model Parameters and Coefficients 0 ccceccecceeeeseeeeesteeeeeeeeaes Overview of TOPAZ 3 12 Digital Terrain Analysis Model 0 00 4 1 IFAC OTOUNG Je up ciha ae Siva o wie NA atone Ted ata Me Ah aies Sth 4 2 Fundamentals and Program Capabilities ccccccccessecceeeetteeeeees 4 2 1 Fundamental Concepts cccccccccceessseceeeeeneeeeeseeneeeeeeeeaes 4 2 2 Elevation Data Pre processing cccccccceeeesteeeeeeeteeeeeeeeaes 4 2 3 Hydrographic Segmentation cceccecceeseseeceeeeeteeeeeeeenes 4 2 4 Topographic Parameterization eeseeeeseeeeeteeeeeneeeeeneees 4 2 5 Assumptions and Limitations 00 ccccccceeeseeeeeeeeteeeeeeees IRS EST CIIO RS innean vst tana gested tewGirs cg cate a a a tease ange eae ete iii iv 1 INTRODUCTION A hydrological model is an attempt
35. b basin ASA This use of satellite data is particularly helpful in apply ing the model to macro scale basins where sufficient land based data may not be available Applications of SLURP have been published for basins varying in size from prairie sloughs measured in a few hectares Su et al 1997 through small basins of the order of 250 square kilometers Sabourin 1996 and basins in the range of 7 500 35 000 km Kite 1993 to a macroscale basin of 1 8 million square kilometers Kite et al 1994 but the model may perform best when applied to meso and macro scale basins Hydrologic models to be successful must be able to represent the significant processes at the scale of interest The SLURP model allows a choice between five different methods of calculating evaporation from soils and transpiration from vegetation The first method is the complementary relationship areal evapotranspiration CRAE model Morton 1983 In most hydrological models and GCMs e g Dickinson amp Henderson Sellers 1988 actual evapotranspiration is a positive linear function of potential evapotranspiration In CRAE on the other hand the actual evapotranspiration is the complement of potential evapotranspiration the argument being that a high potential means a low actual and conversely a low potential means that the actual evapotranspiration is already high In a desert there will be a high potential but no actual evapotranspiration In an irrigated area surr
36. b basins known in the model as aggregated simulation areas or ASAs e Percentages of each sub basin ASA area occupied by different land covers e Main channel length and change in elevation in the sub basins ASAs e Means and standard deviations for each land cover of differences in elevation to the nearest point on a stream for each sub basin ASA e Means and standard deviations for each land cover of differences in elevation along the stream to the outlet of each sub basin ASA e Means and standard deviations for each land cover of distances to the nearest point on a stream for each sub basin ASA e Means and standard deviations for each land cover of distances along the stream to the sub basin ASA outlets Mean latitude and altitude of each sub basin ASA Areas of influence for climate stations Mean annual precipitation for each sub basin ASA Sequence of sub basins ASAs within the basin The routines automatically convert the data into the format used by the model The sequence of events is shown in Figure 14 Details are in the SLURP manual but at the data collection stage are not needed 42 3 3 DEDNM INP Derive DEM m input DEM for TOPRIVER to in IDRISI TOPAZ modify DEM TOPAZ topographic analysis igi se ie SLURPAZ Land cover map aed an 4 prepares SLURP 6N and climate SLU outputs to input files station locations
37. be aware that measurements in the field are usually point measurements while SLURP uses areal averages the two may not be the same or from previous experience in a similar basin parameter sets for several basins are included on the SLURP CD ROM The model parameters are put into a data file known as lt Command_ name CMD gt where Command name is usually the name of the basin being modelled by filling in fields on the SLURP screens using EDIT Table 11 Some typical parameter values for different land cover types on Field f Ave E capacity Maximum on Bin elt Wind Priestley Land Cover E p 3s fraction Albedo infiltration fate speed Taylor Beta of soil mm day fetn Alpha soil water Saer mm day Deciduous 0 05 0 25 0 15 100 2 25 1 26 10 Forest Coniferous 0 05 0 25 0 12 100 2 25 1 26 10 Forest Mixed Forest 0 05 0 25 0 14 100 2 25 1 26 10 Crop amp grass 0 05 0 25 0 23 40 gt 17 1 26 15 Bare Soil 0 05 0 25 0 25 20 5 12 1 26 15 4 TOPAZ 3 12 DIGITAL TERRAIN ANALYSIS MODEL 4 1 Background Topography is important to the description quantification and interpretation of many biosphere processes Examples of such processes in the field of hydrology include surface runoff and water storage energy fluxes evapotranspiration soil erosion and snow metamorphosis Extracting topographic information for a watershed by traditional manual techniques can be a tedious time consuming subjective and error prone task particularly for large
38. c data for selected land covers in Beydagi sub basin K k Menderes Basin Non Coniferous Land cover Maki irrigated i forest agriculture Average change in elevation m to nearest 287 0 51 0 657 0 stream Mean distance km to nearest stream 7 6 2 6 1 2 Standard deviation of distances to nearest 33 18 33 stream Ave change in elevation m down stream 78 0 66 0 84 0 Mean distance km down stream 11 3 9 6 12 1 Standard deviation of distances down stream 3 5 3 8 3 1 The total travel times are used in a lognormal smoothing filter to distribute the runoff from each land cover over time The results are weighted by the percentages of the ASA covered by each land cover converted to m s and added to the total flow of the ASA The distances and changes in elevation to stream and down stream are most easily computed using the TOPAZ TOpographic PArameteriZation system described in more detail in Section 4 1 TOPAZ processes digital elevation data to define the stream courses and sub basins and produce a series of raster images and output tables The SLURPAZ program then processes the topographic outputs from TOPAZ together with land cover data to produce input files for the SLURP model Alternatively the user could use GIS directly to compute the distances and changes in elevation and the land cover percentages for each ASA 2 3 Routing between Sub basins ASAs Any imported streamflow FLI files from for example inter b
39. ction for each land cover depending on whether wind speed data are available or not The Penman Monteith method uses file lt Command_name PM gt while the Linacre method uses lt Command_name LIN gt These files will all be created automatically as the user fills in the on screen forms in SLURP and the user does not need to be concerned with formats The expression lt Command_ name gt is usually the name of the basin being modelled 38 Table 10 Data types used in the SLURP basin model Data Type command file precipitation air temperature dewpoint temperature relative humidity average wind speed incident global radiation hours of bright sunshine cloud cover snow covered area snow water equivalent recorded streamflow streamflow imported to an ASA long term mean streamflow inputs for Morton CRAE evapotranspiration option inputs for Granger evapotranspiration option inputs for Spittlehouse Black evapotranspiration option inputs for Penman Monteith evapotranspiration option inputs for Linacre evapotranspiration option specifications for reservoir operation specifications for diversions operation daily lake or reservoir levels used for routing specifications for interventions in vertical water balance of a land cover specifications for changes to land cover specifications for PET outputs calculated streamflow log of optimization activity complete daily water balance ASA hydrographs precip temperature st
40. d A Orada 1994 Use of GIS to assist hydrological modelling in the Kenyan Rift Valley Agi 94 Birmingham 13 4 1 8 Su M W J Stolte and G van der Kamp 1997 Modelling wetland hydrology using SLURP Proc Scientific Meeting of the Canadian Geophysical Union Banff Alberta p 198 Tarboten D G M J Al Adhami and D S Bowles 1991 A preliminary comparison of snowmelt models for erosion prediction In Proceedings of the 59th Annual Western Snow Conference 12 15 April 1991 Juneau Alaska 79 90 Vanclooster M P Viaene J Diels and K Christiaens 1994 WAVE Reference and User Manual release 2 Institute for Land and Water Management Katholieke Universiteit Leuven Belgium van Dam J C J Huygen J G Wesseling R A Feddes P Kabat P E V van Walsum P Groenendijk and C A van Diepen 1997 Theory of SWAP version 2 0 Simulation of water flow solute transport and plant growth in the Soil Water Atmosphere Plant environment Report 71 Department of Water Resources Wageningen Agricultural University The Netherlands Vehvil inen B 1991 A physically based snowcover model In Recent Advances in the Modelling of Hydrologic Systems by D S Bowles and P E O Connel eds NATO ASI Series Vol C345 Kluwer Academic Publishers Dordrecht Netherlands 113 136 Verhoef A and R A Feddes 1991 Preliminary review of revised FAO radiation and temperature methods Report 16 Landbouwuniversiteit Wageningen Wageningen
41. del simulation the observed and computed ASA hydrographs may be displayed on screen The model generates statistics for the recorded and simulated streamflows for each ASA The statistics include four measures of efficiency which may be used to measure the effectiveness of the calibration or simulation a The Nash Sutcliffe efficiency Nash amp Sutcliffe 1970 OF F F ae 37 where F Da E 38 and Fi e 39 n is the number of days q is the observed flow on day i c is the simulated flow on day 7 and q is the average measured flow Martinec amp Rango 1989 have discussed the implications of defining q as either the average flow over the period in question or as the long term mean flow b The Garrick efficiency measure Garrick et al 1978 in which F in 37 above is replace by 1d lt p PACI q 40 i 34 where q is the long term mean flow on day i This is a more stringent criterion as it compares the model simulation on a particular day to the long term mean for that day This criterion is also known as the Coefficient of Gain from Daily Means WMO 1986 c The Previous Day criterion Kite 1991 Given no other information one possible estimate of flow on day i might be the flow on the previous day i 1 taking advantage of the persistence component of river and lake time series A suitable measure of efficiency replaces F in 37 above with 1 n F 4 qa 41 i l The sensitivity
42. e P E O Connell and J Rasmussen 1986 An introduction to the European Hydrological System Syst me Hydrologique Europ en SHE 1 History and philosophy of a physically based distributed modelling system Journal of Hydrology 87 45 59 Allen R G M E Wright and R D Burman 1989 Operational estimates of evapotranspiration Agron J 81 650 662 Barr A G G W Kite R Granger and C Smith 1997 Evaluating three evapotranspiration methods in the SLURP macroscale hydrological model Hydrological Processes Vol 11 p 1685 1705 Brubaker K A Rango and W Kustas 1996 Incorporating radiation inputs into the Snowmelt Runoff Model Hydrological Processes 464 Calder I R 1996 Dependence of rainfall interception on drop size 1 Development of the two layer stochastic model Journal of Hydrology 185 363 378 Costa Cabral M C and S J Burges 1994 Digital Elevation Model Networks DEMON A Model of Flow Over Hillslopes for Computations of Contributing and Dispersal Areas Water Resources Research 30 6 1681 1692 Clemence B S E and R E Schultze 1982 An assessment of temperature based equations for estimating daily crop water loss to the atmosphere in South Africa Crop Production 11 21 25 Dent M C Schultze R E and G R Angus 1988 Crop water requirements deficits and water yield for irrigation planning in southern Africa Report 118 1 88 Water Research Commission Pretoria Dickinson R E
43. e computed in two ways If the area of the lake or reservoir is relatively small i e is only a fraction of the ASA area then the method used will be one of the Morton CRAE Granger Spittlehouse Black Penman Monteith or Linacre methods just as is used for other land covers in the ASA However for very large lakes the heat storage in the lake causes a lag in the annual evaporation cycle To account for this the user can specify the use of the Morton lake model Morton et al 1985 This evaporation model first computes monthly evaporations then daily and should be used for at least a complete year to allow for simulation of the seasonal changes in heat storage of the lake It is this heat storage which results for example in Great Slave Lake having its maximum evaporation in September and October long after the July August peak air temperatures At present this option operates for complete calendar years of data only ie January to December 2 3 4 External Routing Regulation In some cases the simple lake or reservoir routing and regulation built in to the SLURP model and described above may not be adequate For such cases the user can prepare a computer program in any language to describe the transformation of inflow to outflow which may also involve other data files compile the program link in any other routines needed and name the resulting executable program SLURP1I2 RUT EXE The SLURP model will provide SLURP12_RUT EXE with the
44. e upstream drainage area at each raster cell is then determined using the downslope flow routing concept and the channel network is defined as those cells with an upstream drainage area greater than a user defined critical source area CSA value The critical source area is the drainage area at which a permanent channel 46 begins The network identified in this way is a fully connected convergent and uni directional downslope channel network One of several innovations in TOPAZ is the capability to generate a hydrographic segmentation and channel network with spatially varying characteristics that is the network structure drainage density and sub catchments properties can be different in different parts of the watershed This capability is used to account for spatial variation in hydrologic controls such as geology soil type vegetation and or climate TOPAZ can also prune very short and likely spurious exterior channel links from the generated channel network The above capabilities allow the generation of channel networks of different density and resolution to meet the scale needs and purpose of a particular application For example the same drainage network can look quite different when examined on topographic maps of different scales This different network density resolution can be reproduced by TOPAZ by either generating all large and small channels within the landscape only the largest channels or a network between these two extr
45. earch 30 5 1546 1559 Kite G W M Danard and B Li 1998 Simulating long series of streamflow using data from an atmospheric model Hydrological Sciences Journal 43 3 391 408 Kouwen N F Seglenieks and E D Soulis 1995 The use of distributed rainfall data and distributed hydrologic models for the estimation of peak flows for the Columbia River Basin Progress Report No 2 Waterloo Research Institute Waterloo Krysanova V Miller Wohlfeil D I and A Becker 1996 Integrated modelling of hydrology and water quality in mesoscale watersheds Proc Third Int Conf on Integrating GIS and Environmental Modeling Santa Fe New Mexico January 21 25 Kustas W P A Rango and R Uijlenhoet 1994 A simple energy budget algorithm for the snowmelt runoff model Water Resources Research 30 5 1515 1527 Linacre E T 1977 A simple formula for estimating evaporation rates in various climates using temperature data alone Agricultural Meteorology 18 409 424 Linsley R K M A Kohler and J L H Paulhus 1949 Applied hydrology McGraw Hill MacCracken M C A D Hecht M I Budyko andY A Izrael eds 1990 Prospects for Future Climate A special US USSR report on climate and climate change Lewis Publishers New York 270pp Mark D M 1984 Automatic Detection of Drainage Networks from Digital Elevation Models Cartographica 21 2 3 168 178 Martinec J A Rango and R Roberts 1994 Snowmelt Runoff Model SRM user
46. eat flux for each ASA For each land cover it also requires the coefficients alpha and beta the field capacity and the wilting point The underlying Priestley Taylor relationship has been found Verhoef and Feddes 1991 to slightly underestimate in the low evapotranspiration range lt 2 mm day and to seriously underestimate in semi arid and arid regions where evapotranspiration exceeds 6 mm day 12 Computing Potential Evapotranspiration using the FAO Penman Monteith Method The FAO combination method was developed Verhoef and Feddes 1991 from the Penman Monteith algorithm E i R 0 x U x Qe e 13 S Y A where E is the potential evapotranspiration s is the slope of the vapour pressure curve y is the adiabatic psychrometric constant R is the net radiation G is the soil heat flux U gt is the windspeed at 2 meters and e eg is the vapour pressure deficit with e as the saturation vapour pressure and e as the saturation vapour pressure at dewpoint In the SWAP version van Dam et al 1997 of the FAO combination method used in SLURP the radiation term uses separate albedos for crop and soil further modified in SLURP for ageing snowpack and the aerodynamic resistance depends on the crop height and canopy resistance The crop height on a particular day is computed in SLURP from the maximum crop height and the stage of plant growth The start and end of the growing season are specified on an input screen and sto
47. ed as the soil water content reached when water has been allowed to percolate naturally from the soil until drainage ceases and the remaining water is held by capillary forces great enough to resist gravity The permanent wilting point is the lower limit of soil water available to plants At the wilting point the hydraulic conductivity is so low that water cannot move to the roots and no water is available for transpiration Field capacity wilting point and porosity may be estimated from soil types using pedo transfer functions Potential Evapotranspiration Wet Environment Areal Evapotranspiration Evaporation Areal Evapotranspiration Water Supply to Soil Plant Surface Figure 3 Relationship between actual potential and wet environment evaporations in the CRAE model Morton 1983 10 Computing Potential Evapotranspiration using the Granger Method Evapotranspiration E mm is calculated for each land cover as _ VGO yGE 11 7 VG Y E where Vis the slope of the vapour pressure curve kPa C G is the relative evaporation dimensionless Oy is the net radiation mm eq d y is the psychrometric constant 0 066 kPa C and is the drying power mm eq d G is computed from D the relative drying power dimensionless which in turn is derived from 1 0 905 0 095e 2 0 2 1 D Ea the drying power mm eq d fu a wind speed function mm d kPa ea the vapour pressure kPa and e
48. edium or High Table 9 Typical parameters for the SLURP model Parameter name How derived Sensitivity Initial contents of snow store Measured Estimated Medium Initial contents of slow store Estimated Low Maximum infiltration rate Estimated Low Manning roughness surface Estimated Low Retention constant fast store Estimated High Max capacity fast store Measured Estimated High Retention constant slow store Estimated High Max capacity slow store Measured Estimated Medium Precipitation factor Estimated High Snow melt temperature Estimated Medium Temperature lapse rate Measured Estimated Medium Precipitation lapse rate Measured Estimated Low Initial contents of canopy store Null Low Initial contents of fast store Null Low Max capacity of canopy store Estimated Medium Albedos surface snow Measured Estimated Medium Leaf area index Measured Estimated Medium Interception coefficients a b Estimated Medium Soil heat flux Estimated Low Areas of the land covers Measured GIS Medium Elevations of the land covers Measured GIS Medium Distances to stream Measured GIS Medium Distances downstream Measured GIS Medium Changes in elevation to stream Measured GIS Medium Changes in elevation downstream Measured GIS Medium Snow melt rates and radiation coeff Estimated High Muskingum routing coeff x K Estimated Low River geometry Measured Estimated Low Priestley Taylor coefficients Estimated Medium Aerodynamic
49. emes Once the channel network has been fully defined it is further processed to determine the Strahler order of each channel link and to assign an identification number to each network node and channel link Garbrecht and Martz 1997b These identification numbers are used to determine the optimal routing sequence for cascade type flow routing through the channel network Garbrecht 1988 Sub catchments represent the direct contributing drainage areas of the left and right side of each channel link and of the source node or upstream end of each exterior link or 1st order channel Sub catchments and corresponding divides are determined from the previously defined drainage pattern and channel network Also subcatchment identification numbers based on network node and channel link identification numbers are assigned to provide the topological relationship between network nodes channel links and sub catchments 4 2 4 Topographic Parameterization Topographic parameterization involves measuring a variety of properties and parameters of the DEM raster and of the derived channel network channel links and sub catchments Spatial parameters such as rectified elevations landscape slope and aspect drainage pattern channel network and subcatchment boundaries are generally stored in raster format and are available for visual display using an external GIS On the other hand parameters that describe specific features of the landscape such as the geometry
50. epression Any remaining depression after the breaching is considered to be a sink depression Sink depressions are treated in the conventional manner by raising the elevations within the depression to the elevation of its lowest breached overflow Martz and Garbrecht 1998 The flat surfaces produced by this depression filling as well as those inherent to the DEM such as level valley floors or plateaus at drainage divides require further rectification to ensure unambiguous downslope drainage at every location in the DEM In TOPAZ this is achieved through a relief imposition algorithm which takes into consideration the rising and falling topography surrounding the flat surfaces to generate a realistic topographically consistent and convergent drainage over those surfaces Arbitrary drainage direction assignment is minimized because the topography surrounding the flat surface controls the relief imposition and drainage determination The DEM that has been rectified for depressions and flat surfaces is the basis for all subsequent topographic evaluations Garbrecht and Martz 1997a 4 2 3 Hydrographic Segmentation Hydrographic segmentation identifies the channel network the upstream and lateral sub catchments contributing flow to each channel link in that network and corresponding drainage divides The drainage over the digital landscape is first determined as the steepest downslope path from each raster cell to one of its 8 adjacent cells Th
51. erature in degrees above the critical temperature Terit Sn R T T 17 In SLURP the snowmelt rate R on any day is calculated using a parabolic interpolation from values specified for each land cover on January Ist and July Ist WMO 1974 provides some estimates of the range of melt rates Table 2 16 Table 2 Some typical values of snowmelt rates Month Forested Non forested April 2 0 4 0 May 3 0 6 0 June 4 0 7 0 Many studies e g Tarboten et al 1991 Vehvil inen 1991 Rango and Martinec 1995 have found that despite its simplicity the degree day method of snowmelt is not easily improved upon even by a full energy balance approach Second SLURP may use the simplified energy budget method proposed by Kustas et al 1994 Brubaker et al 1996 describe the application of this restricted degree day algorithm to the Snowmelt Runoff Model SRM Martinec et al 1994 In this algorithm the melt rate is a constant different for each land cover in SLURP and does not vary with season The increased melting later in the season is simulated using net radiation data so that Equation 17 becomes Sin R T T 4 a Q 18 where R3 is the restricted degree day snowmelt rate Q is the net radiation in MJ m day and a is a conversion factor Brubaker et al 1996 used a value of 2 0 mm day C for R2 The theoretical value of the radiation conversion factor is Allen et al 1998 the inverse of the la
52. from the slow store of the appropriate ASA and land cover For water use within the basin only the actual consumptive use should be withdrawn For water exports from the basin the full water demand should be used The intervention file must specify the file to which water withdrawals will be written as lt anyname EXG gt 2 5 Calibrating the Parameters In an ideal model all parameters would be physically based and easily measured and calibration would be unnecessary However given the extreme variability in hydrological conditions across a basin this ideal state is unlikely ever to be reached The SHE model is probably one of the most physically realistic of current models requiring vast numbers of distributed data yet even SHE needs to be calibrated for practical applications Refsgaard et al 1992 SLURP uses parameters which may be estimated directly for different land classes e g Manning s n infiltration rate hydraulic conductivity soil depths etc but because the model is distributed and the land class parameters are applied over large areas some form of calibration is often necessary Once parameters have been derived for a particular set of land classes the model may be applied to other basins which contain the same land classes without any further calibration 32 Table 9 lists the parameters of the model and shows which are measured estimated or calibrated The sensitivities of the various parameters are classed as Low M
53. he Morton CRAE Method Areal evapotranspiration is calculated over an entire sub basin ASA from potential and wet environment evaporation and is taken first from the canopy then the snowpack if it exists then from the fast store and finally from the slow store More conventional evaporation models such as the Spittlehouse Black described below compute a point potential evapotranspiration E using the Penman 1948 or similar method and then calculate a point actual evapotranspiration E as a positive linear function of the potential E QE 7 a p where is an arbitrary coefficient A variation on this method replaces with resistance factors such as those of Monteith 1981 The CRAE model however first computes an areal wet environment evaporation Ew and then calculates the areal actual evapotranspiration E as a complementary function of the wet environment value Bouchet 1963 E 2E E 8 The CRAE model computes E by solving the energy balance and aerodynamic equations at equilibrium temperature using a modification of the Penman equation to replace the wind function with a vapour transfer coefficient Ew is computed from an empirical equation using the slope of the saturation vapour pressure temperature curve and global radiation The version of the CRAE model built into SLURP uses mean daily temperature dew point temperature and either hours of bright sunshine the ratio of actual hours to theoretical s
54. hods The version used in SLURP includes the daylight hours correction factor proposed by Dent et al 1988 The Morton method has the advantage of not requiring many parameters but it does not take into account the properties of different land cover types The Granger method takes into account vegetation type and requires either wind data or a windspeed function The Spittlehouse method computes transpiration and evaporation separately and requires three extra parameters for each land cover type It is well known that the Priestley Taylor algorithm on which Spittlehouse is based underestimates potential evapotranspiration in arid and semi arid environments A critical review of the Morton Granger and Spittlehouse methods is given in Barr et al 1997 The implementation of the Penman Monteith method is the most data intensive but has the advantage of being the FAO standard For most purposes and particularly if irrigation is included in the basin the Penman Monteith method is the most appropriate as it allows differentiation by crop For each of the first four methods SLURP may use either hours of bright sunshine or the ratio of recorded hours of sunshine to the maximum possible hours of sunshine or global radiation as input When sunshine data are used as input the ASA average daily global radiation in MJ m day will be computed by SLURP and written to file and may be used in subsequent model runs V r smarty et al 1998 have noted in a
55. in river v Irrigation diversion Urban withdrawal at structure at structure v Urban withdrawal from groundwater Accumulated streamflow from ASA1 and runoffs from land covers in ASA2 Figure 10 Diversions and interventions in the vertical water balance of the SLURP basin model 2 4 2 Interventions in the Vertical Water Balance Section 2 4 1 has discussed how water may be diverted from the river for irrigation or consumption using a regulatory structure at an ASA outlet The use of this water for irrigating a particular land cover crop in a particular ASA is treated as an intervention in the vertical water balance of that land cover Extraction of groundwater for consumption by urban or industrial users or for export from the basin is also treated as an intervention in the vertical water balance An intervention is specified by putting Y for each affected ASA on the Flowpaths Diversion amp Interventions data input screen As the model simulates the vertical water balance it checks this parameter and if applicable the model will read an intervention file lt Command name gt INT and follow the rules for each land cover 31 File lt Command_name gt INT should be prepared as shown in Table 8 below The program MAKE INT F is provided in directory PROGRAMS SLURP UTILITY on the CD ROM as an example of an automated procedure to prepare a INT file Each record of the file contains information on the ASA land c
56. is to support hydrologic modelling and analysis but it can also be used to address a variety of geomorphological environmental and remote sensing applications TOPAZ is not a GIS in the traditional sense but is a system of software programs that performs numerical processing of raster DEMs and produces a number of data layers and attribute tables Data layer algebra and monitor display capabilities are not included in TOPAZ For such functions TOPAZ relies on a user selected GIS The interface to a GIS is provided through generated raster files The purpose of this section is to provide an overview of the fundamental concepts capabilities and limitations of the software underlying methods and models organization and operation of the software input requirements and output and available documentation 4 2 Fundamentals and Program Capabilities 4 2 1 Fundamental Concepts The DEM processing in TOPAZ is based on the D8 method the downslope flow routing concept and the critical source area CSA concept The D8 method Douglas 1986 Fairfield and Leymarie 1991 defines landscape properties for each individual raster cell by the evaluation of itself and its 8 immediately adjacent cells The downslope flow routing concept defines the drainage and flow direction on the landscape surface as the steepest downslope path from the cell of interest to one of its 8 adjacent cells Mark 1984 O Callaghan and Mark 1984 Morris and Heerdeger 1988 The CS
57. llow the identification and parameterization of the channel network sub catchments and other landscape features of relevance b local drainage and flow direction can adequately be defined by 8 directions 4 diagonal to and 4 in the principle raster directions c upstream drainage areas and drainage paths are adequately described by downstream flow routing from one cell spilling onto another adjacent cell in the direction of steepest descent d the channel network is adequately described by all cells that have an upstream drainage area above a user specified threshold drainage area the CSA parameter e representative values of distributed subcatchment properties are quantified based on models that define the reduction of distributed values of the properties into single representative values 48 The above five assumptions limitations do not inhibit the numerical processing of DEMs by software TOPAZ they only impact the quality and usefulness of the TOPAZ produced data The major implications of the assumptions are discussed in the following Assumption a Landscape features of relevance must be several times the size of a raster cell to be uniquely identifiable and lend themselves to parameterization For example DEM elevation cannot represent thalweg elevation of incised channels that are narrower that the horizontal DEM resolution As a result computed drainage reflects general terrain disposition and not necessarily unresolved
58. ly e_ water Q water x x R G S Y e_soil a x x R G 12 S y store max x wilt max x field wilt e swl p x where e_water and e_soil are the energy limited evapotranspiration rates mm from water and soil respectively is the empirical Priestley Taylor coefficient s is the slope of the saturation vapour pressure curve kPa C y is the psychrometric constant 0 066 kPa C R is the net radiation and G is the soil heat flux Barr et al 1997 found that varies from 1 26 for crop or grass to 0 80 for coniferous forest Shuttleworth 1993 recommends 1 26 if relative humidity exceeds 60 and 1 74 otherwise e_swi is the soil water limited transpiration rate mm and f is an empirical coefficient Spittlehouse 1989 recommended estimating 2 using the assumption that soil moisture begins to limit evapotranspiration at about 30 extractable soil moisture and E a 4 0 mm day This assumption may be expressed as 8 a 13 mm day and gives values of of 10 mm day for coniferous forest and 16 mm day for pasture Store and max are the current and maximum possible soil water contents mm and field and wilt are the field capacity and wilting point as decimals 0 0 1 0 The average evapotranspiration demand from the Spittlehouse Black method for the whole land cover is determined from the relative areas of soil and crop cover The Spittlehouse Black method requires air temperature and soil h
59. mate data for as many stations as possible within and close to the study basin and flow data for as many stations as possible on the main river tributaries and distributaries The third step is to obtain a digital elevation model and a land cover dataset Finally data for the model parameters should be obtained The next sections of this report give further information on each of these steps 3 1 Point Station Data The vertical water balance within the SLURP model converts precipitation data into evapotranspiration runoff and changes in soil and groundwater storages To do this the model needs climate data either as recorded at climate stations or as output by an atmospheric model The following time series of data are required e daily mean air temperature e daily dewpoint temperature or relative humidity e daily total precipitation e daily global radiation or hours of bright sunshine or ratio of observed to maximum possible hours of bright sunshine 37 Other optional time series that may be used include For calibration e observed daily discharge For calculation of the Garrick performance criterion e long term mean daily discharges Any of four methods of computing evapotranspiration may be used in the SLURP model If the user selects the Morton method then the following is needed e long term mean annual precipitation If the user selects either the Granger or the Penman Monteith method of computing evapotranspiration
60. max is the maximum possible infiltration rate If the water supply is higher than Inf or the fast store is full the water excess is spilt as surface runoff RO Since the model operates on a daily time step it is difficult to include any better representation of the physics of infiltration Other authors e g Krysanova et al 1996 divided daily rainfall into 4 mm slugs which allowed use of a smaller infiltration rate but this is equally artificial Still other models use various approximations to the Richardson equation Richards 1931 but this is not suitable for a daily time step Previous versions of SLURP 18 Kite 1995 used the Philip 1954 refinement of the early Green amp Ampt 1911 equation based on Darcy s equation for flow in porous media but again this was found to be ineffective at the model s daily time interval If the current content of the fast store S is higher than the field capacity FC then the excess is spilled as interflow RZ Percolation from the fast store to the slow store takes place at a rate depending on the current store contents S the wilting point WP and the retention constant k PERC S1 WP k 20 Evaporation and transpiration are taken from the fast store in proportion to the percentages of soil and vegetation cover derived from the LAI Vanclooster et al 1994 E soilcover E p cropcover E arer 21 where 0 6 LAI soilcover e 22 The percolation
61. n F I F Ricard and S Fogarasi 1985 Operational estimates of areal evapotranspiration and lake evaporation Program WREVAP NHRI Paper 24 National Hydrology Research Institute Ottawa Naden P S 1993 A routing model for continental scale hydrology In Macroscale Modelling of the Hydrosphere Proceedings of the Yokahama Symposium July 1993 JASH Publication No 214 67 79 Naef F 1981 Can we model the rainfall runoff process today Hydrol Sci Bull 26 3 281 9 Nash J E and J V Sutcliffe 1970 River flow forecasting through conceptual models Part 1 A discussion of principles J Hydrology 10 3 282 290 O Callaghan J F and D M Mark 1984 The Extraction of Drainage Networks from Digital Elevation Data Computer Vision Graphics and Image Processing 28 323 344 Penman H L 1948 Natural evaporation from open water bare soil and grass Proceedings of the Royal Meteorological Society London 193 120 145 Philip J R 1954 An infiltration equation with physical significance Soil Sci 77 1 153 157 Pietroniro A L Hamlin T Prowse E Soulis and N Kouwen 1997 Application of a radiation temperature index snowmelt model to the lower Liard River valley Proc 2nd Scientific Workshop for the Mackenzie GEWEX Study MAGS NHRI Saskatoon 89 90 Priestley C H B and R J Taylor 1972 On the assessment of surface heat flux and evaporation using large scale parameters Monthly Weather Review
62. n mm Z is the interception mm and 4 and B are coefficients There are other models for interception e g Calder 1996 which take into account drop size and multiple canopy layers but since in most cases evaporation from interception is relatively small the above equation seems adequate The model initially sets A to 1 LAImax and sets B to 1 0 so that interception is proportional to the ratio of actual LAI to maximum LAI The user may reset these initial settings Interception is limited by the maximum canopy capacity The model interpolates daily values of LAI from beginning of month values poe l lv Fez Withdrawds Figure 4 Simplified flow chart of the vertical water balance applied to each land class within each sub basin ASA Intercepted water is evaporated from the canopy at rate e_water Land cover albedo is input by the user In the Penman Monteith method soil albedo is assumed to be 0 15 and canopy resistance is set to 0 Canopy resistance for other crops varies between 30 s m for arable crops to 150 s m for forest van Dam et al 1997 These parameters are not used in the Morton Granger or Spittlehouse methods 15 The mean daily air temperatures T C are derived for each land cover in each sub basin ASA by adjusting for elevation using a vertical lapse rate in two stages First during the process of deriving ASA average temperatures from recorded and second when converting from ASA average tempera
63. n a number of distinct GRUs In the SLURP model the basin is divided into sub basins on the basis of topography Figure 1 Each sub basin is known as an aggregated simulation area ASA since it in turn is subdivided into sub areas of different land use The sub basin ASA used in the SLURP model is therefore similar to a collection of Kouwen s GRUs or Krysanova s hydrotops It is a grouping of smaller areas each of which has known land cover properties For example land cover may be measured from satellite for pixels as small as 10 m but it would be impracticable for a hydrological model to operate at such a pixel dimension for a macro scale basin Instead the pixels are aggregated into areas which are more convenient for modelling Tahtali Torbali Beyindir Burgaz Aktas Beydagi e a g Selcuk M S Menderes Tire Odemis Figure 1 Dividing the basin into sub basins ASAs by topographic analysis Although the sub basins ASAs used in the SLURP model could be squares rectangles or other regularly shaped areas they are more properly based on stream network shapes derived from the topography since nature does not operate on grid squares Those models which reduce a basin to a series of grid squares as in the GRU concept are attempting to simulate an artificial rather than a natural environment Hydraulic routing in a grid square model reduces river basins to a series of plates and waterfalls rather like the terraced rice fields of Bali o
64. n calibrating it is better if the number of ASAs equals or exceeds the number of land cover classes If the model is to be calibrated then at least one of the sub basins ASAs must have a streamflow gauge at the outlet At each time increment the model is applied sequentially to the matrix of sub basins ASAs and land covers Each element of the ASA x land cover matrix is simulated by four nonlinear reservoirs representing canopy interception snowpack a store for fast runoff may be considered as a combined surface storage and top soil layer and a store for slow runoff may be considered as groundwater The model routes precipitation through the appropriate processes and generates outputs evaporation transpiration surface runoff interflow and groundwater discharge and changes in storages canopy interception snowpack fast store and slow store Runoffs are accumulated from each land cover within a sub basin ASA using a time contributing area relationship for each land class and the combined runoff is converted to streamflow and routed between each sub basin ASA A key concept in the development of the SLURP model is that it was designed to make maximum use of remotely sensed data As well as using land cover information from Landsat or NOAA satellites the SLURP model may also use NOAA AVHRR visible and infrared data to augment the calculation of snow extent and DMSP satellite data to compute average snow water equivalent over each su
65. ng ASA average precipitation from climate station data and second when converting from ASA average precipitation to land cover precipitation in the daily water balance There is a cap on the maximum percent increase in precipitation of 50 at both stages Because each land cover occupies a band of elevation there is also a precipitation correction factor This should normally be set to 1 0 but may be used to compensate for known gauge undercatch One of the biggest problems facing hydrological modellers is that of obtaining distributed precipitation data Without good distributed precipitation data the most physically sophisti cated distributed model will give no better results than the simplest lumped model The vertical water balance in SLURP assumes that each sub basin ASA either has precipitation data derived from an external distribution scheme such as radar or an atmospheric model or uses Thiessen weights and lapse rates in Option 4 to distribute data from climate stations Optionally the precipitation data may be modified by the percentage of the ASA covered by cloud as derived from NOAA AVHRR visible and near infrared satellite data Kite 1989a e Precipitation is intercepted by the vegetation canopy The maximum canopy capacity is set as the product of the canopy capacity and the leaf area index LAI The canopy capacity is filled by intercepted precipitation as Spittlehouse 1989 14 AP 15 where P is the precipitatio
66. o specify that an ASA should use internal lake or reservoir routing set the appropriate variable in the SLURP sub basin routing diversions and interventions input window to I and prepare a lt Command_name gt RUT file with the routing specifications In this file the user specifies the name of the ASA at which the reservoir routing takes place there may be an unlimited number of ASAs with reservoir routing the type of routing specified outflows or natural the parameters of the stage area relationship the stage volume relationship and the stage discharge relationship if appropriate for the lake or reservoir the maximum and minimum allowable reservoir levels and the starting and ending dates of each diversion period Table 5 shows an example for Beydagi Reservoir in the K k Menderes Basin Turkey Table 5 Example of part of a RUT reservoir routing file for the Beydagi Reservoir in the K k Menderes Basin beydagi R 60 beydagi 036165 173 1 52763 beydagi 019644 173 2 44276 beydagi 240 173 beydagi 1994 1 1 1994 1 310 beydagi 1994 2 1 1994 2 28 0 beydagi 1994 3 1 1994 3 31 0 beydagi 1994 4 1 1994 4 30 0 156 beydagi 1994 5 1 1994 5 31 2 658 beydagi 1994 6 1 1994 6 30 7 434 beydagi 1994 7 1 1994 7 31 9 828 beydagi 1994 8 1 1994 8 31 9 780 beydagi 1994 9 1 1994 9 30 9 60 beydagi 1994 10 1 1994 10 31 0 978 26 In this example the Beyagi sub basin contains a reservoir with outflows specified for the years 19
67. of the first three criteria to changes in the parameter values is very different It is easy to calibrate the model to give a high value of the Nash Sutcliffe statistic less easy to improve the Garrick statistic and very difficult to obtain a calibration with a high value of the Previous Day criterion Figure 11 shows an example of the variation in from the top Nash Sutcliffe Garrick and Previous Day as the values of two parameters are stepped across a grid The three criteria are all displayed to the same scale and clearly show the relative sensitivity Figure 11 Relative sensitivities of three performance criteria used in the SLURP basin model as two parameters are changed in a regular grid pattern 35 d The fourth criterion computed in the model is the deviation of runoff volumes D WMO 1986 D 100 a 42 m where Vm is the measured volume of runoff and V is the computed volume of runoff over the period of interest This criterion is simply a transformation of the mean computed and observed flows for the simulation period A negative value of D means that the mean computed flow is too high This is somewhat counterintuitive and it might have made more sense for WMO to reverse D so that a positive criterion would indicate an excess of simulated water The Student s t test is often used to test the hypothesis that observed and computed values are not significantly different from each other and to determine at
68. om the river may be made to two types of file SIR and EXR In SLURP water diverted may be used as irrigation supply for any land cover in any downstream sub basin ASA Diversions for irrigation water must be written to files labelled lt ASAname gt SIR where lt ASAname gt refers to the ASA containing the irrigated area and SIR stands for Surface IRrigation Any water remaining in an SIR file after use for irrigation will be added to the outflow from the irrigated ASA Water being consumed in the basin or exported from the basin must be written to files with the name lt ASAname gt EXR where lt ASAname gt is the name of the ASA at which the EXtraction from the basin takes place In this way they will be recognized by SLURP to get the basin water balance correct There is no provision for return flows from an EXR file and so water withdrawals for urban and industrial use within the basin should be specified as only that portion of the total water use which is consumed For water exported from the basin the full withdrawal should be specified If the river does not contain sufficient flow to meet the total diversion demand the diversions will be reduced proportionately to equal the river flow In the DIV file the user may also specify a minimum allowable flow in the river for environmental or dilution purposes for example In 30 this case the requested diversions will be reduced as necessary to maintain this minimum flow in the ma
69. ore contents surface water diversions from an ASA for irrigation water diverted from a river for consumption or export water extracted from groundwater for consumption or export input file for lake evaporation output file for lake evaporation output file with annual PET data output file with PET data for specific dates output file with daily PET data at specific ASA land cover combinations File Extension CMD PCP TAV TDP RHU WND GLO SUN CLD SNO SWE FLO FLI LTM MOR SB PM LIN RUT DIV LVL LCC MAP FLC OPT PRN PLT STO SIR EXR EXG WRI WRO PET DAY TIM Input or Output I I I I I It I It I I I I I I It I It It It I I It So OOOO OOoO OO OOO O Units various Mm Ce On decimal m s w m day hours or percentage percentage mm m s m s m s various various various various various various various m various various mm m s text various mm amp m s mm m s m s m s various various mm mm mm signifies data which are optional depending on choice of analysis and model output requirements 39 All the climate station and streamflow data used in SLURP are contained in unformatted space delimited ASCII data files Each record of these files contains the data for one day arranged in the order year 4 digit integer month 2 digit integer day 2 digit integer and data real including decimal point The following shows
70. ounded by desert there will be a high actual evapotranspiration and a correspondingly low potential evapotranspiration The second method for estimating evapotranspiration which is available in the model is also based on a feedback mechanism and takes into account land cover and vegetation Granger 1991 This method is adaptable to use remotely sensed data The third option is to use the Spittlehouse and Black S B method Spittlehouse 1989 which uses the more conventional Priestley and Taylor 1972 method and introduces a soil moisture limited evapotranspiration Fourthly the widely accepted FAO version of Penman Monteith Verhoef and Feddes 1991 is used in the implementation used in the SWAP model van Dam et al 1997 Finally the Linacre 1977 approximation to the Penman equation is available for data sparse areas In all five methods the overall evapotranspiration from a land cover is computed as evaporation from the canopy and soil surface and transpiration from the vegetation in proportion to the ratio of soil covered and vegetation covered area The soil and vegetation areas are determined by the leaf area index which may be derived from remotely sensed data In some cases such as the Great Lakes of the St Lawrence and Mackenzie Basins it may be necessary to use the lake evaporation method of Morton et al 1985 which accounts for seasonal changes in sub surface heat storage to simulate the lag between air temperature and
71. over type of intervention start and end dates of the intervention and the rate of intervention The file may contain any number of records but a particular date for a particular land cover within a particular ASA may occur in only one record Table 8 Example of part of a INT intervention file for the Gediz Basin medar non irrig X 19861 1 1998 12 31 0 0060 N N medar exg gordes non irrig X 19861 1 1998 12 31 0 0015 N N gol exg gordes coniferous X 19861 1 1998 12 31 0 0013 N N gordes exg gordes maki X 19861 1 1998 1231 0 0839 N N izmir 1 exg kum maki X 19861 1 1998 1231 0 0026 N N surahan exg kum non irrig X 19861 1 1998 12 31 0 1305 N N izmir 2 exg selendi coniferous X 19861 1 1998 12 31 0 0003 N N selendi exg If the intervention is for irrigation then the INT record will specify the source of the water surface or groundwater and the type of irrigation sprinkler or furrow If the irrigation source is specified as surface water then SLURP will read the appropriate SIR file generated by diversion from the river see previous section and check that enough water is available for the requested irrigation rate The actual rate used will be the smaller of that requested and that available Excess water will be returned to the ASA outlet Groundwater for irrigation is taken from the slow store Interventions may also be used to withdraw groundwater for urban or industrial water consumption or for export from the basin This will be taken
72. r data such as the USGS global dataset derived from satellite sensors such as NOAA AVHRR or Landsat TM MSS There are many different global land cover classification schemes at many different scales For many basins the U S Geological Survey s USGS Earth Resources Observation System EROS Data Center 1 km resolution global land cover characteristics data base available from the USGS website http edcwww cr usgs gov landdaac glcc glcc html will be suitable Just as with the DEM the land cover image may be imported into IDRISI after unpacking The image must now 41 be made to correspond to the DEM image by converting from the Lambert Azimuthal Equal Area projection to the UTM projection and by forcing the image to the same number of rows and columns Finally the land cover image must be converted to an integer ASCII file named LCLASS INP and exported for use by the SLURP ancillary topographic analysis routines Again the full procedure is described in detail in the SLURP manual The following figure shows an example of a land cover dataset for the K k Menderes Basin in western Turkey GR Irrigated F Non irrigated GE Coniferous E Maki E Barren GM Shrubland _ k Water 0 20 40 km Figure 13 A land cover dataset for the K k Menderes Basin in western Turkey The topographic analysis routines derive the following information from the DEM and the land cover dataset e A division of the basin into su
73. r the Philippines When we can model quite well the way in which water flows through real rivers and channels there can be no advantage in simulating such an artificial environment While grid based hydrological models can usually vary the grid size to some extent using topographically based sub basins ASAs allows the size of the sub basin to vary over the full range of possibilities A further advantage of modelling the basin with sub basins ASAs rather than a grid is that lakes and reservoirs can be more easily simulated Kite 1998 Land cover data are commonly derived from satellite images and are used in SLURP as an indicator of vegetation type soil characteristics and physiography The very close relationship between land cover and soil characteristics has been demonstrated by Droogers et al 1998 This embedded use of land cover information makes the SLURP model particularly useful for studies in which land cover is expected to change for example in climatic change studies Kite 1993 SLURP has the ability to work with land covers that may change any number of times throughout a model run Figure 2 shows the concept of aggregating land cover data from the pixels of the original map or image into percentages of each land cover for each sub basin ASA I Irrigated __ Non irrigated B Coniferous E Maki E Barren B Shrubland E Water 0 20 40 km 7 15 31 1 46 Figure 2 Converting
74. re for Hydraulics Hydrology and Hydrodynamics Computational Mechanics Publications 1 3 129 138 Garbrecht J and L W Martz 1993 Network and subwatershed parameters extracted from digital elevation models the Bill s Creek experience Water Resources Research 29 909 916 Garbrecht J and L W Martz 1997a The Assignment of Drainage Direction over Flat Surfaces in Raster Digital Elevation Models Journal of Hydrology 193 204 213 Garbrecht J and L W Martz 1997b Automated channel ordering and Node Indexing for Raster Channel Networks Computers and Geosciences 23 a 961 966 Garbrecht J and J Campbell 1997 TOPAZ Version 1 20 An automated digital landscape analysis tool for topographic evaluation drainage identification watershed segmentation and subcatchment parameterization User Manual Rep GRL 97 4 Grazinglands Research Laboratory USDA Agricultural Research Service El Reno Oklahoma Garrick M C Cunnane and J E Nash 1978 A criterion for efficiency of rainfall runoff models J Hydrology 36 375 381 Goodrich R D 1931 Rapid calculation of reservoir discharge Civil Eng 1 417 418 Granger R J and D M Gray 1990 A net radiation model for calculating daily snowmelt in open environments Nordic Hydrology 21 217 234 Granger R J 1991 Evaporation from natural nonsaturated surfaces Ph D Thesis Department of Agricultural Engineering University of Saskatchewan Saskatoon 140p 51
75. red in file lt command name gt pm For land covers such as forests and shrubs the growing season is year round Computing Potential Evapotranspiration using the Linacre Method The Linacre 1977 pan equivalent potential evaporation in mm day is computed as Schultze 1989 as p 100T 100 4 u T T 80 T a 14 where Tm Ta 0 006 Am and Ta is the mean air temperature Am is the elevation m is the latitude in degrees u is a wind factor often defaulted to 15 and T T4 is the difference between air and dewpoint temperatures approximated by Ta Ta 0 0023 Am 0 37 Ta 0 53 Rm 0 35 Race 10 9 with R as the mean daily or monthly range in temperature and Rro as the difference between the mean temperatures of the hottest and coldest months of the year Dent et al 1988 adjusted equation 14 by adding a correction for the length of daylight to g 2 1007 100 9 u T T4 E 80 T where D is the number of daylight hours divided by 12 13 Computing Actual Evapotranspiration from Potential For all five methods the soil evaporation and crop transpiration are computed and recorded separately in the SLURP model depending on the daily ratio of soil and crop cover within each land cover The ratio of actual to potential evaporation and transpiration varies with the current contents of the soil water stores If the current content is greater than the field capacity of the soil then the ratio va
76. resentative minimum and maximum distances to stream are computed from the mean u and standard deviation as L nin u z 20 26 Laax H 20 max The minimum and maximum to stream travel times are then computed for each land cover using the velocity from Equation 25 and the minimum and maximum distances from Equation 26 Similarly Manning s equation is used with measured stream depth width and estimated stream roughness to calculate a stream velocity Travel time down the stream to the ASA outlet for each land cover is computed from the mean distance down stream the change in elevation down stream and the stream velocity Total travel times are the sums of the to stream and the down stream travel times 21 Rontmg from n r ASAs either non linear reservoir or Mnekingnro Cunge TN Distance to stream External ronting Distance down stream Figure 6 Concept of within ASA routing using distributions of to stream and down stream distances for each land cover 600 500 400 No of pixels 300 200 100 3 Distance km Figure 7 Distribution of the to stream distances for ASA Menderes Basin 22 Water Shrub Irrig Crop Barren Non irrig Crop Maki Con Forest each land cover for a typical sub basin Table 4 summarizes the to stream and down stream data for selected land covers in Beydagi sub basin K k Menderes Basin Table 4 Physiographi
77. ries from 0 3 to 1 0 If the current content is between the field capacity and a point f half way between the field capacity and the wilting point then the ratio is 1 0 If the current content is between f and the wilting point then the ratio varies from 1 0 to 0 0 Finally if the current content is less than the wilting point then the ratio is 0 0 2 1 2 The Vertical Water Balance for all Land Covers except Water Figure 4 shows a simplified flow chart of the vertical water balance that the SLURP model applies to each element of the matrix of ASAs and land covers except for a water land cover The structure contains four non linear reservoirs or stores one for canopy storage one for snowpack one for a rapid runoff may be considered as a combined surface storage and top soil layer storage and one for a slow runoff may be considered as a groundwater response Model parameters include the initial contents of snow storage mm the initial water content of the slow store as a percent of the maximum capacity and a maximum infiltration capacity mm day There are retention constants and maximum capacities for both the fast store and the slow store The vertical water balance operates at a daily time step The sequence of operations is as follows e The daily precipitation mm for an ASA is adjusted to the mean elevation of the land cover using an estimated percent change per 100 m in two stages First during the process of derivi
78. tent heat of vapourization ie 0 408 mm per MJ m which is equivalent to 0 035 mm per W m Pietroniro et al 1997 compared the simple degree day and the restricted degree approaches for a northern basin and found that the addition of radiation data made no significant difference The snowmelt rates and conversion factors calibrated in their model are given in Table 3 In the implementation of this method in SLURP a different value of is used for each land cover to allow for the different proportion of radiation intercepted by the various heights of vegetation When SLURP is run with either hours of bright sunshine or global radiation the net radiation needed for snowmelt will be computed internally Climate change scenarios such as those generated by GCMs for increased atmospheric greenhouse gases change not only air temperatures but also cloud cover and hence net radiation If SLURP is used to evaluate the effects of such a climate change scenario on hydrology and water resources the advantage of the restricted degree day approach is that it can include separate changes for temperature and net radiation 17 Table 3 Values of snowmelt rates and radiation conversion factors Pietroniro et al 1997 Land cover Degree day melt Restricted degree Radiation rate day melt rate conversion factor mm day C mm day C mm w m7 day Deciduous mixed forest 1 80 0 12 0 27 Coniferous forest 1 80 0 84 0 17 Transitional forest 3 36 1 08
79. the saturated vapour pressure at the air temperature kPa E a 7 E Oy E f ea e The wind speed function f is defined as f a bU a 8 19 0 22Z b 1 16 0 08Z where U is the wind speed m s and Zo is an aerodynamic resistance s m for the particular land cover In SLURP two options are available for Granger evaporation If daily average wind speed m s data are available then an aerodynamic resistance Zo is input for each land cover in the Granger Feedback Evapotranspiration data input window If wind data are not available then the input window is used to input the wind speed function f directly Z usually varies between 30 for grass and 150 for trees Allen et al 1989 and fa is between 10 and 30 If the mean daily wind speed is used then this will be multiplied by 1 33 to obtain the average daytime wind speed Smith 1992 11 The Granger potential evapotranspiration a is divided into a soil evaporation demand e_soil and a crop transpiration demand e water using Equation 9 and the actual evaporation and transpiration are limited by the relationship of the current fast or slow store contents store the soil field capacity fe and the wilting point w using Equation 10 Computing Potential Evapotranspiration using the Spittlehouse Black Method The Spittlehouse Black Spittlehouse 1989 method of computing evapotranspiration uses the Priestley and Taylor 1972 equation for water and soil separate
80. ting would be done for each cell or pixel of the basin In the semi distributed approach taken in SLURP we compute the distances for each pixel and then calculate the moments of normal distributions fitted to the to stream and down stream distances Figure 6 shows this concept and Figure 7 gives an example of the distributions of these distances for land classes used in one ASA of the K k Menderes Basin The statistics required are the mean and standard deviation of the distances to the nearest stream for each land cover the mean and standard deviation of the distances along the stream to the ASA outlet for each land cover and the average changes in elevation to the nearest stream and down the stream for each land cover To compute travel times for each land cover it is necessary to estimate a velocity for travel to stream and a velocity for travel down stream Naden 1993 used a velocity of 0 6 m s for networks of the order of 10 000 km and Wyss et al 1990 used a subsurface velocity of 2 x 10 m s and a channel velocity of 0 6 m s for small New England catchments SLURP computes an average velocity V m s for each land cover for flow to the stream using Manning s equation with a different coefficient of roughness for each land cover n and assuming that the hydraulic radius R is a small number for wide shallow flow If H is the average change in elevation over the distance L to the stream then V 1 00 mR H L 25 Rep
81. to describe the physical processes controlling the transformation of precipitation to runoff and generally includes the hydraulics of water transfer to the ocean or other sink The detail of the model design will depend on the purpose of the modelling For simple simulation or forecasting of a streamflow hydrograph one of the many lumped models will often suffice but for scientific research it may be necessary to use a more detailed physically based fully distributed model such as the Syst me Hydrologique Europ en SHE model Abbott et al 1986 SLURP Semi distributed Land Use based Runoff Processes is a conceptual model which although normally used in semi distributed form is capable of use as a fully distributed hydrological model The first 1975 version was developed for use in meso scale Canadian basins as an alternative to the use of larger and more complicated models such as the U S Army Corps of Engineers SSARR Streamflow Synthesis and Reservoir Routing model Kite 1975 1978 The major advantage of semi distributed models such as SLURP is that they can incorporate the necessary physics while retaining comparative simplicity of operation Using SLURP in a semi distributed form the model is able to simulate the behaviour of a basin at many points and in many variables while avoiding the data and computation hungry excesses of the fully distributed models In practical applications the users of both simple lumped models and ph
82. tures to land cover temperatures If the mean air temperature for the land cover is above a critical temperature then the precipitation is assumed to be rainfall The critical temperature should be set initially to 0 0 but as average lapse rates are used and as any particular land cover is not all at the same elevation some minor adjustment above or below 0 0 may be necessary Setting the critical temperature to a small positive value will increase the percentage of precipitation which occurs as snow Dew point temperatures are lapsed at a rate of 20 that used for air temperature Rainfall left over from the canopy is added to the fast store A second store represents snow storage If the daily mean temperature is below or equal to the critical temperature then precipitation is assumed to be snowfall and is accumulated to the snowpack Sp If the temperature is above the critical temperature precipitation is assumed to be rain Land cover albedos can be adjusted for the effects of snow on the land cover The snowpack albedo is set to 0 80 initially and reduced as a 0 086 0 798 4 16 each day to a minimum of 0 40 or until the snowpack is melted or sublimated If fresh snow falls the albedo is reset to 0 80 US Army 1966 The snowpack may then be depleted by snowmelt Sm This may be done in two ways in SLURP First a simple degree day approach may be used in which melt depends only on a melt rate R mm day C and the temp
83. unshine hours or global radiation Figure 3 shows the relationships between Ea Ew and E The abscissa represents water supply and the ordinate indicates the magnitude of the various evaporation terms The left hand side of the diagram shows that with no water supply there can be no actual evaporation resulting in a very high potential evaporation Moving to the right along the x axis as the water supply increases the actual evaporation increases and the potential evaporation decreases until when the water supply reaches its limiting value both values meet at the wet environment evaporation In the SLURP model the CRAE areal potential evapotranspiration Ea is divided into a soil evaporation demand e soil and a crop transpiration demand e_water using the relative areas of soil and vegetation cover within a given land cover as e_soil E x soilcover 9 e_water E x crop cover store max x W gt X potential F max x e W p e e 10 actual where soilcover and cropcover are fractions of total cover as computed from LAI in Equation 22 The actual evaporation and transpiration are limited by the relationship of the current fast or slow store contents store the soil field capacity fe and the wilting point w In SLURP the field capacity and wilting point are input as decimals of the maximum water capacity max of the fast and slow stores and are converted to mm in the program The field capacity is defin
84. ure at sea level day is the Julian day of the year and the rest of the variables are the same as for Equation 1 The diffuse radiation is then increased due to cloud cover as Oxi dfo 2 68 2 2 n N 3 85 n N 4 The total global shortwave radiation is the sum of the direct Qars and diffuse Quy radiation The net shortwave radiation is then found by taking into account surface albedo Net shortwave radiation Q is finally Q Qa Ox l a 5 where alpha is the surface albedo Outgoing net longwave radiation Qm is then obtained using QO oT 0 34 0 139 fe 0 1 0 91 6 Verhoef and Feddes 1991 where T is the absolute temperature in degrees Kelvin e4 is the vapour pressure in kPa and o is the Stefan Boltzmann constant The net radiation available for evapotranspiration Qn is Qs Qm Maximum soil heat flux G in MJ m day and the date on which this occurs is specified by the user and SLURP then computes the soil heat flux for a particular day using a sine curve Soil heat flux is positive when the soil is warming and subtracted from net radiation Soil heat flux for a month may be estimated as Allen et al 1998 G 0 14 T nonni T nonthi1 where Tmonth is the mean monthly air temperature In many cases soil heat flux is small and may be ignored Potential evapotranspiration is computed from the net radiation differently by each method Computing Potential Evapotranspiration using t
85. ure that evapotranspiration from the slow store is assigned to evaporation and not transpiration For a water land cover the parameters and associated variables should be set as follows LAI set to 0 0 canopy capacity set to 0 0 interception coefficients set to 0 0 wilting point set to 0 0 field capacity set to 1 0 porosity set to 1 0 max crop height set to 0 0 start and end of crop growing season set to 0 Parameter 1 Initial contents of snow store mm set as normal Parameter 2 Initial contents of slow store set to 50 Parameter 3 Maximum infiltration rate mm day set to maximum Parameter 4 Manning roughness n set to 0 0 Parameter 5 Retention constant for fast store set to 1 0 Parameter 6 Maximum capacity for fast store mm set to 0 0 Parameter 7 Retention constant for slow store set to maximum Parameter 8 Maximum capacity for slow store mm set as estimated maximum lake depth Parameter 9 Precipitation factor set to 1 0 Parameter 10 Rain snow division temperature deg C set to 0 0 Precipitation Evaporation Snow ice Runoff Figure 5 Simplified flow chart of the vertical water balance applied to a water land cover 20 2 2 Routing within a Sub basin ASA Runoff from each land cover is first routed to the nearest point on a stream and then routed down the stream to the sub basin ASA outlet In a fully distributed model this rou
86. variations can be modelled in TOPAZ by use of a spatially variable CSA parameter Assumption e For a particular distributed subcatchment property there can exist several different representative values depending on the model selected for reducing the distributed values of the property into a representative value The user must select the model that is consistent and valid for his her particular application The user should also be aware of the magnitude of the difference in representative values due to alternative property reduction models 49 An additional assumption is that depressions and flat surfaces in the digital landscape are spurious and can be rectified to allow an unambiguous determination of downslope drainage assumption c Therefore landscape drainage properties associated with true depressions or lakes at the bottom of which downslope drainage terminates cannot be generated directly As a result of assumptions a b and c the landscape under consideration must be dissected by channels and display sufficient relief to define downstream flow paths Also the DEM must be of sufficient resolution to adequately define these drainage features In general the data produced by TOPAZ adequately represents most natural landscapes given sufficient DEM resolution Man made drainage features in natural landscapes can be evaluated if they are incorporated into the DEM elevations 5 REFERENCES Abbott M B J C Bathurst J A Cung
87. what level of significance this hypothesis may be true t 43 where r is the simple correlation coefficient between q and c It is common to compare observed and computed streamflows in graphical form and to fit a simple linear regression such as g atbe 44 to the two sets of data In a good simulation the intercept of such a regression a would be zero and the slope b would be 1 0 The actual values may be computed as Gc POPAL b i l i l 24 E n 45 Dye b 4 a n While the four criteria may be output to screen printer or data file Student s t statistic and the regression coefficients may only be output to the printer or a data file Changes to the model parameters may be made to improve model results if necessary in three ways 36 i By changing individual parameter values making a series of simulations and comparing the criteria values and generated hydrographs ii By using the built in Shuffled Complex Evolution SCE UA optimization method Duan et al 1994 The SCE UA method is very suited to hydrological models as it avoids the many local minima that may exist see for example the peaks and valleys in Figure 11 ili By preparing files for the PEST Watermark Numerical Computing 2000 This method is model independent 3 DATA REQUIREMENTS FOR THE SLURP HYDROLOGICAL MODEL 12 2 The SLURP model requires three types of data to model a natural system point station data
88. ysically based fully distributed models may also tend towards this semi distributed middle ground For example users of lumped models may simulate the use of a distributed model by successively applying a lumped model to sub basins and from the other end of the scale Jain et al 1992 describe the need to use areally averaged variables in practical applications of the fully distributed SHE model Various concepts have been used in distributed or semi distributed hydrological modelling to subdivide basins into hydrologically consistent subareas of the basin Leavesley and Stannard 1984 defined a hydrological response unit HRU as a homogeneous area having a distinct hydrological response and defined by land cover by soil type by slope or by aspect In practice the HRU may be coincident with sub basins defined by the availability of streamflow gauges In models for mountain regions the HRU may also be represented by elevation bands as in the SRM model of Rango 1995 The SWIM model Krysanova et al 1996 disaggregates a basin into hydrotops where a hydrotop is a set of disconnected units in a sub basin which have a unique land use and soil type Stuttart et al 1994 used a similar construct in which areas of common land cover are linked Kouwen et al 1995 described a grouped response unit GRU for a grid square model in which a GRU is a grouping of all areas with a similar land cover so that a grid square will contai

Background Material for SLURP . TOPAZ

Contents

Download Pdf Manuals

Related Search

Related Contents