- Figshare
Contents
1. Possible saturated area Example of a digital hemispherical photograph Saturation should only be found near the zenith far from the foliage Alternative exposure settings exposure setting based on the sky reference is not possible If you have time at hand and want to be as accurate as possible and can read out data on a large memory capacity laptop every other evening Take 2 4 photos per sampling point and evaluate visually right in the field and at night of each field work The photographs with the highest foliage to sky contrast are the best thus you may want to delete all others keeping only one photograph per sampling point Unless you want to evaluate the uncertainty in LAI estimation per sampling point accounting for exposure errors Fix the F stop to constant best at 5 3 Then take picture at 1 2 3 and 2 exposure relative to the automatic For example if the automatic exposure is 1 125 then the four exposures would respectably be 1 250 1 1 500 1 1000 and 1 30 2 Set up used for Canon EOS 5D Mark II wit Sigma 6mm f 3 5 EX DG Circular Fisheye at SAFE Experimental Forest Site Sabah Malaysia http www safeproject net At each sampling point three images have been taken in auto bracketing mode the camera automatically takes all three at each go set via the camera menu The aperture has been set to F 9 which can be adjusted if needed The three images are from left 0 standard exposure 1 underexposure 12. that Leave sample and plot names unchanged When finished go back to Settings and highlight Constants set Absorption to 0 85 EAADP Ellipsoidal leaf Angle Distribution Parameter 1 Done only once for specific fieldwork person Menu use the arrow tab to the right to change in Menu with arrow tab to File Data Storage Enter Your name e g Samuel Disk internal no memory cards inserted thus no A B C existing in this Psion Type comma separated In Menu File you can save your configuration Menu use the arrow tab to the right to change to Settings Highlight Site and press enter Enter new name and with arrows pointing downwards enter new geographic coordinates Then at plot go to site far from forest edge or big opening in canopy or hold Sunscan probe levelled above the canopy e g low vegetation such as short grass there should be no shadow of vegetation on Sunscan Ideally there is no direct sunlight or interspaced white clouds but slightly overcast sky conditions Measure beam fraction Highlight BFrac on screen of Psion Hold a white sheet of paper up so that ca 1 3 or the 1m long probe is covered by its shade and then Press the button on the Sunscan Don t hold the shade too close to the probe otherwise it will cut out some of the diffuse light as well SunData looks at the readings from the photodiodes and uses the lowest value to calculate the Diffuse component of the incident light It use
3. A Disney M Pellikka P Marchant R 2012 Leaf area index for biomes of the Eastern Arc Mountains Landsat and SPOTobservations along precipitation and altitude gradients Remote Sensing of Environment 118 103 115 Ridler W amp Calvard S 1978 Picture thresholding using an iterative selection method EEE Transactions on Systems Man and Cybernetics 6 260 263 Weiss M amp Baret F 2010 CAN EYE V6 1 User Manual EMMAH laboratory Mediterranean environment and agro hydro system modelisation French National Institute of Agricultural Research INRA Weiss M Baret F Myneni R B Pragn re A amp Knyazikhin Y 2000 Investigation of a model inversion technique to estimate canopy biophysical variables from spectral and directional reflectance data Agronomie 20 3 22 Zhang Y Chen J M amp Miller J R 2005 Determining digital hemispherical photograph exposure for leaf area index estimation Agricultural and Forest Meteorology 133 166 181 12
4. and elevation in WGS 1984 or UTM of large field markers like road crossing tall single trees tall buildings to provide data for satellite image ortho rectification Measurement checklist e Plot picture showing biome in its beauty e GPS of plot corners and in the case of transect of sampling points e 3 hemispherical images upward facing per sampling point gt 12 per subplot following Zhang method e Optional hemispherical images that are facing downward e 3 Sunscan readings per sampling point 36 readings per subplot e Slope Elevation of plot e Estimated size of forest patch distance to road e Notes on soil undergrowth cover density height and leaf cover on ground 3 Hemispherical Images camera settings 3 1 Camera and lens Different camera systems can be used but each camera needs to be calibrated to define optical centre and projection function Calibration may be done following steps outlined in the CANEYE image analysis software Weiss amp Baret 2010 download software and manuals from https www4 paca inra fr can eye Suggestions for camera systems e 8 mega pixels Nikon Coolpix 8800 VR digital camera equipped with a fish eye Nikon FC E9 lens adapter full frame sensor e 12 3 megapixels Nikon D5000 SLR digital camera equipped with a Sigma 4 5mm F2 8 fisheye lens adapter full frame sensor e 12 2 megapixels Canon EOS 450D equipped with a Sigma 4 5mm F2 8 fisheye lens adapter not a ful
5. dissertation Publicationes Instituti Geographici Universitatis Helsingiensis p A147 Gonsamo A amp Pellikka P 2008 Methodology comparison for slope correction in canopy leaf area index estimation using hemispherical photography Forest Ecology and Management 256 749 759 Gonsamo A amp Pellikka P 2009 The computation of foliage clumping index using hemispherical photography Agricultural and Forest Meteorology 149 1781 1787 Jacquemoud S amp Baret F 1990 Prospect A model of leaf optical properties spectra Remote Sensing of Environment 34 75 91 Jonckheere B Muys B amp Coppin P 2005 a Allometry and optical in situ LAI determination A case study in Belgium Tree Physiology 25 723 732 Jonckheere I G C Muys B amp Coppin P R 2005 b Derivative Analysis for In Situ High Dynamic Range Hemispherical Photography and Its Application in Forest Stands IEEE Geoscience and Remote Sensing Letters 2 296 300 Jonckheere I G C Fleck S Nackaerts K Muys B Coppin P Weiss M amp Baret F 2004 Review of methods for in situ leaf area index determination part I Theories sensors and hemispherical photography Agricultural and Forest Meteorology 121 19 35 Lang A R G amp Yueqin X 1986 Estimation of leaf area index from transmission of direct sunlight in discontinuous canopies Agricultural and Forest Meteorology 37 229 243 Pfeifer M Gonsamo
6. Manual to measure and model leaf area index and its spatial variability on local and landscape scale Dr Marion Pfeifer amp Dr Alemu Gonsamo February 7 2014 Ecology and Evolution Faculty of Natural Sciences Imperial College of Science Technology and Medicine UK marion pfeifer gmail com Geography Department University of Toronto 100 St George Street Toronto M5S 3G3 Canada gonsamoaQ geog utoronto ca To be used as guide in the projects ICIPE CHIESA Climate Change Impacts on Ecosystem Services and Food Security in Eastern Africa http chiesa icipe org funded by Ministry for Foreign Affairs of Finland and TZ REDD in Tanzania funded by the Ministry of Foreign Affairs Norway Key points Landscape scale Stratified random sampling design to capture heterogeneity within biomesFriday 07 February 2014Friday 07 February 2014 Equipment hemispherical camera tripod compass camera lens lens adapter if needed and Sunscan instrument batteries for Psion handheld incl backup battery markers level GPS more batteries Measure GPS of land markers points observable from space such as trees big houses Measure GPS of plots and biome types Take GPS of plot corners V ALERI plots and sampling points Transects Hemispherical measure sky exposure decrease shutter speed by two stops or employ auto bracketing Sunscan 3 SunScan readings per sampling point taken from different ang
7. TATION If there is slope then upscaling using high resolution satellite to coarse resolution may not work To fasten the 1mage processing step the user should always keep the same position when acquiring the image This will allow the operator to apply the same mask to all the images at once However this should only become relevant when taking downward looking pictures In the upward looking images best use remote control to set off camera or use automatic countdown function Shadows in the image will make the class allocation more difficult A fish eye field of view is large Therefore make sure that there is no one or anything else bag instrument etc present on the image 3 5 Image extraction processing and analysis After downloading from the camera system each image has to be pre processed The pixel brightness values for the blue band are extracted from each RGB image to achieve maximum contrast between leaf and sky because absorption of leafy materials is maximal and sky scattering tends to be highest in that band Jacquemoud amp Baret 1990 Jonckheere et al 2005 b Marker amp GPS Image processing Ria Image download i Extraction of blue band K l Thresholding CAN EYE v6 e Vegetation versus non vegetation LAI based on gap fraction i ca South Marker amp GPS e Plotestimates gt 8 images Thresholding procedures have been developed in order to avoid subjective decisi
8. ands in Tanzania are more heterogeneous Otherwise it would be very good to have as many plots randomly distributed in forest stratified patches so that one may be able to use them to validate global LAI or other biophysical products The more variable the biome the more sampling plots or the larger the plot size 2 Sampling Design on local scale To capture heterogeneity of vegetation canopy structure and biomass within a stratum e g Vegetation plot as specified for RAINFOR carbon measurement protocols five subplots SP should be sampled The subplots are distributed randomly within the plot e g red SP1 to SP5 or grey SP1 to SP5 Subplot size may vary with plant functional type sampled but ideally is 20 m x 20 m unless sampling follows a linear transect Mark plot corners with red yellow tape and record the coordinates of the plot comers Transect plots are very good to save time on field data collection For example you can run four transects for 1 km To avoid sampling bias chose a starting point and then follow the linear transect running through the stratum along the south north line with measurements every 10 m to 20 m Ideally there 1s a land marker at the or close to the starting point which is obvious to detect from space for locating the transect on the satellite image Take a GPS reading at each measurement point Marker amp GPS M2 North Marker amp GPS MI It is essential to record GPS coordinates
9. egetation Compute difference between LAI 80cm per plot and LAI ground level to get LAI of low vegetation 5 References Baret Morissette T J Fernandes R A Champeaus J L Myneni R B Chen J Plummer S et al 2006 Evaluation of the representativeness of network of sites for the global validation and intercomparison of land biophysical products Proposition of the CEOS BELMANIP IEEE Transactions on Geoscience and Remote Sensing 44 1794 1803 Breda N J J 2003 Ground based measurements of leaf area index a review of methods instruments and current controversies Journal of Experimental Botany 54 2403 2417 Garrigues S Allard D Weiss M amp Baret F 2002 Comparing VALERI sampling schemes to better represent high spatial resolution satellite pixel from ground measurements How to characterize an ESU Available for download at http w3 avignon inra fr valeri methodology samplingschemes pdf accessed 17 08 2011 11 Garrigues S Lacaze R Baret F Morisette J T Weiss M Nickeson J E Fernandes R Plummer S Shabanov N V Myneni R B Knyazikhin Y amp Yang W 2008 Validation and intercomparison of global Leaf Area Index products derived from remote sensing data Journal of Geophysical Research 113 doi 10 1029 2007JG000635 Gonsamo A 2009 Remote sensing of leaf area index enhanced retrieval from close range and remotely sensed optical observations Academic
10. l frame sensor resulting in images being cropped at the top and at the bottom e Personal favourite 21 1 megapixels Canon EOS 5D Mark II Digital SLR camera body equipped with SIGMA 8mm f 3 5 EX DG Circular Fisheye using a VANGUARD Aluminium Alloy Tripod Alta 263AGH GPR System Grip Position Release 3 2 Camera settings exposure and resolution Make sure the camera is in fine mode Jpeg or Hi mode Tiff with the highest resolution The camera needs to be set to manual and fisheyel modes if applicable Check your camera manual Exposure settings The correct exposure is to make the sky appear white Using automated exposure may result in unreliable LAI estimates Zhang et al 2005 Thus use the hereafter called Zhang method Two stops of overexposure relative to the sky reference This sky reference reading should be made in a large opening outside the forest stand which provides an unobstructed sky view in all azimuthal directions With experience in angular variations of sky radiance this reference reading can also be made in small openings When the camera s aperture size is fixed to ensure a consistent field of view a decrease of the camera s shutter speed by 2 3 stops will provide the desired exposure Meaning if the sky reference exposure time were determined to be 1 1000 s F5 3 a series of photographs would be taken with the aperture fixed at F5 3 and the shutter speed decreasing to 1 250
11. les Contents 1 Sampling Design on landscape scale Sampling follows a stratified sampling protocol with random point selection within strata disproportionate stratification Pre selection of strata takes place according to following criteria e Rainfall classes e g lt 500 500 800 800 1100 1100 1500 1500 1800 gt 1800 mm e Disturbance levels no low intensity moderate high intensity optional disturbance categories fire logging charcoal e Plant functional types Bushland Woodland Forest Plantation Cropland e WDPA World Database on Protected Areas http www wdpa org categories forest reserves national parks nature reserves unprotected game management parks e Age classes e g forests depending on previous disturbances if known e Crop types Corn Rice Sisal Woody savannah vegetation and croplands are highly seasonal in tropical East Africa driven by fire rainfall and human land use Thus ideally measurements will be repeated within the season to measure intra annual variability of vegetation canopy structure as a base for correlations with environmental parameters later on Field plots should be placed in uniform vegetation of a specific biome e g forest at a distance of gt 100 m from the edge to allow for coarse pixel validation in subsequent satellite image analyses This will apply mostly to forests woodlands and woody savannah systems since many cropl
12. nckheere et al 2004 The aim is to avoid sun glares or sunflecks when taking upward hemispherical images to avoid very over exposed parts in the image 6 which may make the class allocation more difficult in subsequent image processing If there are some it is still possible to mask then during the post processing Be sure that the camera is the most horizontal camera body levelled best done with a level as possible when acquiring a hemispherical image If the image is taken in the upward direction tall canopy use a tripod The upper picture margin should be facing exactly northward at each image take a compass at 1m above ground or another height as long as you always use the same Ideally you got a full frame sensor camera body and a nice fisheye lens on it Don t forget to take the adapter linking camera body and tripod into the field with you Slope Slope affects LAI estimation using hemispherical images Gonsamo amp Pellika 2009 Take take pictures on the flat ground so that the point spread contribution of the surrounding landscape to remote sensing pixel is minimal if the field is large and uniform When working on slopes always take picture levelled to horizontal ground Gonsamo amp Pellikka 2008 and measure the slope and its orientation And again the forest patch should at least be more than 1 pixel 1 km2 on uniform slope for upscaling to moderate spatial resolution sensors such as MODIS and SPOT VEGE
13. ons by the user and to identify the optimal brightness threshold to distinguish vegetation from sky Jonckheere et al 2005 b We use the global Ridler method Ridler amp Calvard 1978 for image thresholding see Pfeifer et al 2012 because of reliable results in hemispherical photography studies Jonckheere et al 2005 a Gonsamo amp Pellikka 2008 The resulting binary images are analysed using the canopy analysis software CAN EYE Weiss amp Baret 2010 http www paca inra fr can eye limiting the field of view of the lens to values between 0 and 60 to avoid mixed pixels and thus misclassifications Certain sky conditions can complicate classification of image pixels into vegetation and non vegetation e g sunflecks and dark clouds Images can be edited in processing software to mask such areas CAN EYE estimates mean effective LAI for each plot LAIgpr LAI assuming randomly located foliage based on a series of at least 8 images from measured gap fraction Weiss amp Baret 2010 LAlgpr is derived by inversion of the Poisson model using look up tables and assuming an ellipsoidal distribution of leaf inclination Weiss et al 2000 A clumping index is computed using the Lang amp Yueqin 1986 logarithm gap fraction averaging method and used in a modified Poisson model which is then inverted to estimate LAIrygug the LAI taking into account canopy clumping Note that estimated LAI rug includes not foliage materials such as stem
14. overexposure e AF Autofocus On Switch on side of lens e Av mode e In Menu settings One Shot File Numbering Continuous Quality Raw Large JPEG Picture Style Standard C Fn I Exposure 4 Bracketing Autocancel OFF Expo comp AEB enter and mark the bracketing points to 0 1 Underexposure 1 Overexposure e Set F to 9 0 Start switch to On and check F Then start switch to line Little wheel on top of camera body behind photo button turn to change F to desired one Then change start switch back to On Resolution The image resolution and the compressing rate if there is will have a significant impact on the image analysis Results will be better with a high resolution and a low compressing rate For example when the leaves are small coniferous forest it is recommended to increase the resolution of the image so that less mixed pixels are present in the image 3 3 Tips and tricks in different vegetation structure Very Short Canopy lt 30 cm height For very short canopies such as young crops the digital camera must be placed looking downward not too close from the foliage so that one leaf does not cover the whole field of view On the other hand the camera should not be placed too high so that the user does not observe a sky ring in the image Therefore an average height of 60 to 80 cm above the last leaf appears reasonable Note that the advantage of downward looking photographs is the pos
15. s trunks branches twigs and plant reproductive parts and is more accurately called plant area index Breda et al 2003 Datasheet LAI using hemispherical images in SAFE vegetation monitoring plots Nama Name Block Plot e g Fragment C Plot 621 Vegetation Plot SAFE Experiment 25 m x 25 m Pictures 3 each e g when reading 432 from camera 430 431 432 lt N A ue un 7 i LAI Pictures m P d e Centre 6 o N ma 7 I ITI GPS coordinates LAI GPS Point 1 Lat Long 4 Lat Long 10 Lat Long 7 Lat Long 4 Sunscan Instrument Measuring in the field Fieldwork step by step 1 2 3 4 5 6 Switch the Psion on after connecting the probe the thing with the 1 m long probe to the psion with the accompanying cable Hold both symbols V E and set remote link to off when finished with measuring in field set remote link on for later readout into PC Note pressing both N X will exit from any application that is running Highlight Sunscan icon and enter the program Continue by pressing enter Done once when doing fieldwork in a specific region Go to Menu use the arrow tab to change to Settings Highlight Time and Date e g Tanzania e g GMT 3 depending on time in year Enter Back to Settings and Highlight Display LAI as display check for
16. s the highest photodiode values to 10 calculate the Total incident and uses these two values to calculate and display the Beam Fraction Thus if the background light changes strongly e g sudden clouds in front of sun causing different light measure beam fraction again Sometimes you may encounter errors and you will have to re measure the beam fraction anyway This may happen more often when you think so don t lose your spirit 7 Go QUICKLY to S1 of plot see plot setup above at each sampling point take three Sunscan readings from three different directions at about 80 to 100 cm above the ground Make sure your shadow does not appear on the 1 m probe Hold the Probe levelled no tilting or fancy movements All readings get appended one after the other Go to S2 and take another three readings from three different directions holding the Probe levelled go to 53 etc Check in between finishing at sample 6 you should have taken 18 readings and be ready for reading LAI number 19 f you went wrong best start from beginning setting up new plot name You will get very confused otherwise believe me At the end in case you have 12 sampling points you should have done 36 LAI readings If you are interested in getting the ground vegetation LAI you may want to run again through the plot taking at each Sample point 3 readings from three different directions but holding the probe levelled just at ground level into the v
17. sibility to get a better spatial representation by increasing the distance between the camera and the canopy while keeping it not too far away to be able to get a clear image of vegetation elements minimizing the mixed pixels problem Short canopy between 30 and 70 cm The digital camera can be placed looking downward with the Same recommendations as for very short canopies It can also be placed horizontally on the ground looking upward In this case be sure that the presence of the camera on the ground does not change the canopy structure Tall canopy without understorey gt 70 cm The camera should be placed at ground level looking upward It 1s necessary also to take care of not modifying the canopy structure when positioning the camera on the ground Tall canopy with understorey Two series of hemispherical images can be acquired one looking downward to characterize understorey the other looking upward to estimate tree characteristics The two kind of images must be processed as two separate series The resulting characteristics can then be recombined to represent the whole canopy 3 4 Measurements in the plot Sky conditions All images sampled in a subplot plot must be acquired in similar illumination conditions so that the colour dynamics does not vary a lot between all the images that are processed together Ideally measurements should be performed under overcast conditions to minimize anisotropy of the sky radiance Jo
Download Pdf Manuals
Related Search