Home

Post Processing Software Manual

Contents

1. 0 8 16 24 32 40 48 56 64 72 80 88 96 104 112 120 128 diode 40 2 R Zoomed Showing Horizontal channel from 12 55 57 488 083 894 to 12 55 57 491 973 147 D 12 55 57 490 583 147 Speed in Secs Length L1 um 40 qq 4 p gt gt Height L2 um 30 F Generate_images_png_pages NS T NOA fad ac nnn 1 cnn Figure A 3 Shows an example of noisy diode data intermixed with good particle data The images highlighted in yellow are rejected Labels in red point to images centered on diodes determined to be bad too noisy by the criteria described in the text and exemplified in the particle center location distribution shown above 1 This final step is another loop through the data but is applied only if the current precipitation status is yes The current precipitation status Is yes if there is one or more particle s with L7 gt 100 microns in the 10000 particles accepted by the previous 4 cleaning steps centered approximately on the current particle It is approximately centered because the precipitation status is updated only every 100 particles instead of for every particle to improve speed Another parameter the current mean inter particle time aveW8 is similarly updated every 100 particles and calculated as the average inter particle time between the 10000 images accepted by the previous 4 cleaning steps centered approximately on the current particle All the particles in t
2. Ymax 1000 00 0 2 15 43 08 000 15 43 10 500 15 43 13 000 15 43 15 500 15 43 18 000 15 43 20 500 15 43 23 000 15 43 25 500 15 43 28 Manual scaling blue M1 H conc mean 7 04E 003 red M1 Viconce mean 5 68E 003 DeadT not Adjusted for Mean yet 30 2 compare Time Series df extinction Trace 1 E 20 MM1 7 CIH 7 Plextn lt Trace 2 S10 MM1 CIV v Plextn Yimin 0 00000 Ymax 10 0000 7 Manual scaling 15 43 08 000 15 43 10 500 15 43 13 000 15 43 15 500 15 43 18 000 15 43 20 500 15 43 23 000 15 43 25 500 15 43 28 blue M1 H extn Mean 1 96E 000 red M1 Vrextn mean 1 50E 000 DeadT not Adjusted for Mean yet 0 08 3 Compare Time Series df mass Trace 1 MIMI 7 CIH 7 Pilwe Y gt gt o Trace 2 MIMI 7 CIV Yi Pilwe ae Nap i 0 00000 3 00000 i js ta fa tA on anette Nica tah NE nari ted aan Ymin Ymax 15 43 08 000 15 43 10 500 15 43 13 000 15 43 15 500 15 43 18 000 15 43 20 500 15 43 23 000 15 43 25 500 15 43 28 e PUE M1 H Lwc mean 6 10E 003 red W1 ViLwo meon 4 36E 003 DeadT nat Adjusted for Wean yet F Manua scaling critenia None g m3 g m3 Cc o A ag gt Mo Figure 12 2 5 3 Comparing size distributions Use the CompPSD button to compare particle size distributions In this window different sizing methods arrays and parameters can be chosen plotted and compare Sizing methods are described in Appendix A Furthermore average values of counts concentration extinction and
3. i TA 3022 Sterling Circle Suite 200 Boulder CO 80301 Phone 303 449 1105 Fax 303 449 0132 www specinc com HVPS Post Processing Using HVPSview Software User Manual September 2010 Using HVPSview Software The HVPS data processing software is constantly being updated We urge users to consult with SPEC on the current status of the software before beginning processing and periodically 1 0 HVPS Pre Processing 2 1 1 Masked diode filling 5 1 2 Dead diode replacement 5 2 0 HVPSview 7 2 1 HVPSview installation 7 2 2 More pre processing 8 2 3 Viewing Images 11 2 4 Artifact Removal 11 2 5 Time Series and Particle Size Distribution PSD Plots 12 2 5 1 Zooming into a time interval 12 2 5 2 Comparing time series 14 2 5 3 Comparing size distributions 15 2 5 4 Viewing HVPS images from a selected time interval 16 2 5 5 Boolean Filtering 17 3 HVPS data Batch processing Archiving 19 3 1 Liquid mode Batch processing 19 3 1 1 Archiving ASCII Files 19 3 1 2 Archiving Images 21 3 1 3 Combining the archived liquid ASCII files 22 3 1 4 Additional Notes 22 3 2 Ice mode Batch processing 23 3 2 1 Archiving Ice ASCII Files 23 3 2 2 Archiving Images 25 3 2 3Combining the archived Ice ASCII files 26 3 2 4 Additional Notes 26 Appendix A Derived Parameters Methods Equations in HVPSview 28 Appendix B Artifact removal algorithms a a 37 1 0 HVPS Pre Processing Processing HVPS data first r
4. Unsorted True air Tas m aa i 2x 10r 4x10 w10 axic 1x10 Unsorted W particle area Arey i 2x10 4x10 Axio axi1c 1x10 heriz08 1031123842 _124124_124149_6007 bin Figure 5 4 Again a warning messages window pops up showing warnings of idiosyncratic data events This window is rarely empty but usually has several to a few dozen events An example is shown in Fig 8 qj Messages generated while loading the data file ol x File Waring there are 18 beg 0 and area gt 0 0 Indes and 3 Area 104 0 4 0 4 0 Indes and 3 Malice 0 0 20 20 Indes and 3 Ticks 1 16822198536 1682219939 1682220518 Figure 8 Example of the warning messages window that pops up after loading the bin files Once this is completed three time series are displayed for concentration extinction and ice water content If the file contains more than 300 seconds the data is coarsely binned into 300 segments for quick look using the 1 Hz data file that was generated upon opening the BIN file If the file is less than 300 seconds then the 1 Hz data is used directly These 1 Hz data are for quick look only The processing uses a crude length parameter Lj see appendix A and no artifact rejection H Showing data from SORTEDS1031123842_ 124124 124149_0007 2D5 zz Joj x File G CleanAll CleanNaise fos UselnFocusOnly Archive_this_liq_ascii Archive_this_png Make_N_asci_help Vertical Par
5. 1 Copy all sorted HVPS and the associated HK_bin HVPS HK_TXT HVPS and Log_TXT HVPS files into the same directory 2 Get IDL started and follow the steps described in section 1 1 1 3 The initial GUI screen will appear Select archive multi Liq asci from the drop down list A dialog window will appear for the user to select an input file of type Sorted HVPS as displayed in Figure 20 Navigate to the directory containing the SORTED HVPS data that requires processing Click on one of the sorted files and then click open on the lower right corner That please select a file window will disappear S iew Yersion 0 S SPEC 2DS i ersion 0 59 oj x File IE H y Q CleanAll CleanNaise 0 9 M UselnfocusOnly Archive Multi_Liq ascii x Archive_this_png v Make_N_asci_hep 7 Please Select a File Look in E lan_June_1_2009 amp ex E Fe J SORTEDOS1031123842_123921_123941_0001 2D5 21 3 SORTED081031123842_123939_124004_0002 2D5 E SORTEDOS1031123842_124002_124022_0003 2D5 Documents 2 SORTEDOS1031 123842_124020_124049_0004 2D5 D own Li St png Drop 3 SORTEDO81031123842_124047_124110_0005 2D5 3 SORTEDO81031123842_124108_124132_0006 2D5 Desktop 2 SORTEDO81031123842_124130_124156_0007 2D5 3 SORTEDO81031123842_124154_124223_0008 2D5 S 2 SOBFEBRSTIG 123842_124221_124246_0009 2D5 8 SORTEDOS1031123842_124244_124307_0010 2D5 My Documents 8 SORTEDO81031123842
6. 225694 00 59 02 202 125 169 L27ALTOG 22564A 00 59 02 233 716 051 127AA2FA Z256C0 00 59 02 2651244 286 12732807 224A30 00 59 01 832 691 051 12734407 224A4C 00 59 01 839 015 757 aoa co co co oO oO co oo oo Figure 2 Playback allows the user to look at images while pre processing the data By selecting the Maximum Playback RUN and choosing one of the two options Playback will create the HVPSview files as quickly as possible This option is to the right of the Help tab The option to not display timestamps is fastest The other option is quite slow To the right of the Folder icon there are four green buttons The farthest right is the Play button This button will play the file back at a speed set by clicking the icon to the right of the folder icon You can use the Pause button to stop playback You can step through the data particle by particle using the Step button On the Main Display there is a Stereo Pair counter This area will display horizontal and vertical particles that arrived at nearly the same time Clicking the Settings button allows the user to set the detection settings 1 1 Masked Diode Filling During data collection a diode may become noisy If the condition is serious enough the real time software will detect this and ignore mask the diode This results in a white line running the length of any particle image that includes the mask diode Examples are shown in Fig 1 and Fig 2 Playback has an
7. Bette bette bette tebe bette bette tettetet ttt Are you sure to do above files 7 yes no Enter y or rv ANSWER _ as fee Mee tae SB Db fp Ready Figure 26 The sorted files will be processed one at a time in order Some initial plots and warning messages will pop up be saved and deleted The original GUI screen will remain open where the progress is displayed in the status bar The IDL standard output screen will also print out some of the steps the archiving the process has reached WE Aubochearnd 115000 of 826565 End of third loop tre ave_wi_t calculation H wfed Jun 10 12 52 04 2009 Weed Jun 10 12 52 04 2009nnn 30 Done H channel Cleaning A Weed Jun 10 12 5204 2009nnn 291888 Begin wS_t_temp calculation ued Jun 10 12 52 end w_t_bemp calculation Wed Jun 10 12 52 17 oho ss e apb Ready Figure 27 The status bar in figure 27 gives the following information 1 The V channel automatic noise removal is being used 2 Of the 828565 images it has examined 115000 particles 3 The end of the third loop noise removal process for the H channel was done at time Wed Jun 10 12 52 04 2009 As each SORTED HVPS file is processed many files are generated and saved in a sub directory labeled datal data2 data3 etc in the directory where the original SORTED HVPS files are stored These files are initial plots of the file warning messages encountered when loading the data
8. SV default Area x pxs x strbsz 1000000 a 0 115 B 1 218 o 0 917 and where K and K are unit conversion factors 3 Kx Adj x 2 x L x strbsz a LWC g m 6 SV Where K is unit conversion default factor Counts PSD For each size bin the formula for bin height is simply the number of particles in that size bin Use size bins with width equal to strbsz for now Bin L x strbsz into the size bins The first size bin has events for which L 1 and the bin edges are 0 5 strbsz and 1 5 strbsz the n size bin has events for which L n and n 0 5 strbsz and n 0 5 strbsz as bin edges Concentration PSD L um For each size bin the formula for bin height is K x Adj l ae al dth pam x SV pran where the sum is over all events in that size bin and Where K unit conversion factor Bin as above for counts PSD Area PSD mm7 L um For each size bin the formula for bin height is K x Adj x A l OIN x SV sna where the sum 1s over all events in that size bin and A Area x pxs x strbsz 1000000 and Where K unit conversion factor so the units come out Bin as above for counts PSD Ice Mass PSD g m um K gt Adj x minfa x A p a x L x strbsz oe bin _ width pu SV sein Mee A Area x pxs x strbsz 1000000 a 0 115 B 1 218 po 0 917 and where K and K gt are unit conversion factors Liquid Mass PSD g m um K x gt Adj x x L x strbsz i bin _
9. Curent Time 11 03 39 531 Masked pels will show a yellow color in the table Double click again to remove Butes Received H Particles W Particles Mask the masking Piel 0 is the left mast roo 27230 eye aE write Data To Disk column of the sensor 127 the right most real 32009 Displayed setting are sticky and will be retained if the program is restarted 11 03 39 531 Reset Bin Counts Figure 3 Example of the Pixel histogram analysis window showing a couple dead diodes V 12 and V 117 and a detectable noisy diode V 124 2 0 HVPSview 2 1 HVPSview Installation HVPSview runs on an IDL platform of 6 0 or higher The projects programs should be stored in a folder called HVPSview version Please note that HVPSview has been derived from the 2DSview software The 2D S has two input channels Horizontal and Vertical while the HVPS has only one channel For various reasons stemming from both the hardware electronic configuration and the software the single HVPS channel has been called Vertical and V although the probe usually is mounted horizontally Please note that some example images below may include a Horizontal channel as the test data imaged better illustrated a concept such as masking HVPS data will only show Vertical data To install the HVPSview software follow these steps 1 Open IDL Click File gt Preferences gt Path and set the path to the HVPSview version folder first and IDL default second Th
10. Therefore playback has an option to add a diode to the mask during pre processing The dead diode must be identified first This is accomplished via the Pixel histogram analysis window Fig 3 The number of counts each diode has had is displayed Clicking in a numbered box turns it yellow and indicates that that diode will be considered masked for the next time playback is run That is adding diodes to the mask is a two step process First for the period being processed you must run playback monitor the histogram of counts display and click on any diodes that you want added to the mask Next exit and restart playback the clicked boxes remain yellow and this time they will be added to the mask for processing If a noisy diode is not detected and masked during data collection it may be detected during these same steps and also added to the mask This can shorted the time needed for HVPSview to process as it must also remove noisy diode effects im SPEC 2D5 data collection and processing program Yer 1 0 16 File Tools Window Help Maximum Playback RUN lo r Jo RES Bila cs pe A Bll 205m Main Display T H gt 94 763 wH eee ES E p i Stereo Pairs S Pixel Histogram Analysis Hor Ver Awg bin 119528 7 117660 0 Mia 17756 20381 Wax Bin 5 124 Horizontal diode 124 l of E r 7 s Double click a cell in the table to cause that pixel to be masked during playback Show Current Count
11. file the auto clean results and the horiz bin and vert bin files The most important two files are H_archive_HVPS _xxxx_Ns txt and V_archive_HVPS _xxxx_Ns txt They are located in the subdirectory png under the datal data2 data3 and etc These two files contain the 1Hz concentration extinction ice water content times series and particle size distributions The header lines contain the detail data format of the archived file 3 2 2 Archiving Images After H archive HVPS_ Ns txt and V_archive HVPS_ Ns txt 1Hz archive files are generated for each sorted HVPS file the HVPSview software can be used on the data To archive the images 1 Select Archive Multi Ice png from the png drop down list see Figure 20 2 A window will ask how many images per second to archive This is dependent upon cloud type and computing power Type in the desired number of images per second and press Enter The images are saved in the subdirectory png under the corresponding datal data2 data3 etc subdirectory 3 2 3 Combining the archived Ice ascii files To combine all the smaller archive files into a single data file for the whole flight a separate IDL program is used To combine all ascii files follow these steps 1 Copy all H_archive_HVPS _Ns txt into one folder or directory 2 Start IDL 3 ClickFile gt Open gt FindHVPSview_ver gt combine_archive_TC4 _Steve_gains_of_61bins_HorV pro 4 ClickRun gt Compilecombine_archive_
12. in Figure 25 See Appendix A for a description of the options Ice_archive_options ol x Arhive W ME M4 M2 M Accept Figure 25 5 Click Accept 6 Return to the IDL front screen All of the files for overnight processing will be displayed 7 If all the files are present type y gt Enter in the IDL command line This will make the computer process all the sorted data files in the selected directory mw Here are all Sorted 205 data files which are Found in current directory for overnight processing bitt ttti ttt SS SS Claran June _1_2009 50RTED081031123842_123042_123923_0000 205 Canlar June _1_200950RATED081031123842_123921_12394_0001 205 Canlar June _1_2009 50RTED081031123842_123939_124004_0002 205 CAlanslan_June_1_2003 S0RTEDOSIOS1 123842 124002 124022 0003205 CMlarslar_June_1_2009 SORTEDOSIOS1 123842 124020 124049 0004 205 CAlanslar_June_1_2003 S0RTEDOSOS 1 12s842 124047 124110_0005 205 CMlarklar_June_1_2003 SORTEDOSIOS1 123842 124108 124132_0006 205 CAlanslan_June_1_2003SS30RTEDOSIOS1 123842 124130_124156_0007 205 CMlarnklar_June_ 1 2009 SORTEDOSIOS1 123842 124154 124225 0008 205 CAlanslarn_June_1_2005 S0RTEDOSIOS 12s842 124221_ 124246 0009 205 CAlanslan_June_1_2003 S0RTEDOSWS1 123042 124244 12450 O00 205 CMlar lar_une_1_2005 SORTEDOSTOS1 123842 124305 125943 00117 2D5 CAlanslan_June_1_2003 SO0RTEDOS10S1 123842 125941_ 130003 0012 205 CMlar lar_June_1_2009 SORTEDOSIOS1 123842 130001_0013 2D5
13. the extreme shaded end pixels from all the slices whereas for Lq it is the end pixels in a given slice See Figure 1 A3 n 4 L F1 0 if neither edge was shaded 1 if left edge only was shaded 2 if right edge only was shaded 3 if both edges were shaded for any slice during the event Adj adjustment factor 1 defined below Adj adjustment factor n defined further below PC position of particle center on array determined as the center between the extreme edge points of the slice that determined L2 PC3 the position of the center of the circumscribing circle PC position of particle center on array determined as the center between the extreme edge points that determined Ls For logistic reasons the derived files will also have TAS included Le the number of white pixels in the strip s that defined L4 So Lg is the strip for which the end shaded pixels are farthest apart which can be a surrogate for the outer diameter of an out of focus ring L then is a surrogate for the inner diameter of the ring as it is the number of white pixels for the strip that defined L4 If there are two or more strips for which L4is maximized then let Le be the maximum value for those strips L is the Alexei scheme corrected size as follows First of all use the maximum value of the two inside perimeter area estimates Next we simply use the table at the end of Alexei s report For each image we have a
14. total area Az max of 2 inside perimeter area estimates from Alexei s Matlab algorithm and an image area shaded pixels also in Alexei s code but we already have this A We take the square root of minus the ratio of the later to the former We look up the closest value in the appropriate column of the table that 1s the Dspot Dmax column we are using table 40 shadow depth We find the corresponding value in the 2nd column that is the Dmax Do column We take the inverse of that and multiply by Dmax to obtain Do which is our corrected size estimate for the bead If the corrected size Do is larger than the original size Dmax do not use the correction use Dmax instead For round images Dmax might be better estimated from total area but for the bead data we will have to use the along the array size L4 as the bead s speeds through the probe varied unless TAS is well adjusted and speeds don t vary too much i L Lg is just as L7 above except instead of Dmax L4 use D 2 A in both m 1 L and Lg above we are using L instead of the possibly more logical Ls because of the tilted arrays giving skewed images Lg tries to account for elongated images with the ratio L4 L For both L and Lg above L4 should be used for in focus images at the 90 level I E after calculating L7 and Lg replace their values with L 4 s value for the subset of data that 1s in focus at the 90 or better level Adj is calculat
15. um 16300 A2 alexei 163 W8_t tick W8_t_clean 262 Roundness 0 or 1 1 0 Warn Figure 14 If the filter box is not checked images will be displayed including accepted particles noise splashes and shatters The rejected images are red and are highlighted in yellow as shown in figure 18 S Zoomed Showing Horizontal channel from 15 43 20 003 126 470 to 15 43 20 999 425 032 m x A uv z o 2 a c o DO z had o i 7 i z oc _ Io t g 7 20 e o V o 42 l P 15 oe 20 336 546 666 Speed in Secs ool zt adaj a Generate_images_png_pages Cleaning by start End Clicks Label_one_image_as_bad_with_one_click Off On Figure 15 This window has the same functionality as described in section 2 3 Clicking Generate images prg pages i Will e e OF s to be displayed on the computer screen saved to the data png directory Buttons cleaning by Start end clicks and label one image as bad with one click are for manually removing bad images or noise if the auto cleaning method doesn t do a perfect job The Timeshift fishing and De spur are special buttons used inside SPEC INC We use them to look at the laser beam alignment We also use them to look at the cloud particle special cluster distribution and to study the noise image distribution in the initial stage of the software They are for development and automatic removal of noise splashes and shatt
16. 1244307_0010 205 CMlanslan dune _1_2009 50RTED081031123842_124305_125943_0011 205 CAlanslan_June_1_2003 S0RTEDOSIOS1 123842 125941_ 130003 0012 205 CMlarn lar_June_1_2009 SORTEDOSIOS1123842 130001_0013 205 HMA RAE A A A A Are you sure to do above files 7 yes no Enter y or rv Figure 21 The sorted files will be processed one at a time in order Some initial plots and warning messages will pop up be saved and deleted The original GUI screen will remain open and the progress is displayed in the status bar The IDL standard output screen will also display some steps that the archiving process has reached wW tAutocleanl 115000 of 220565 End of third loop true ayve_wi_t calculation H fed Jun 10 12 52 04 2009 Wed Jun 10 12 52 04 2009nnn 30 Done H channel Cleaning All Wed Jun 10 12 52 04 2009nnn 891888 Begin w8_t_temp calculation Www ed Jun 10 12 52 14 2009 end wi_t_temp calculation Wed Jun 10 12 52 14 2009 ANSWER KI F a e Locas SF IDL gt L Ready Figure 22 The status bar in Figure 22 displays the following information 1 The V channel automatic noise removal is being used 2 Of the 828565 images it has examined 115000 particles 3 The end of the third loop noise removal process for the H channel was done at time Wed Jun 10 12 52 04 2009 and etc As each SORTED HVPS file 1s processed many files are generated and saved in sub directories labeled datal data2 data3 e
17. 7 z 3 0 Pannel 2 default V diode 2 Tracet Vdiode0 bas v gt 2 5 Trace2 Vdiode64 i LH prd m Trace3 Vdiode1 27 3 g 1 5 15 43 08 15 43 10 15 43 13 15 43 15 15 43 18 15 43 20 15 43 23 15 43 25 15 43 28 time 1 blue Ydiode 0 fA 2 red Vdiodeb4 V 3 Black Vdiodel 27 V oa F PEOR Pannel 2 default Tas Press Co x Tracel tas ai 1500 Q Trace2 VertLaserD B 1000 z Trace3 HorizLaserD x s00 5 15 43 08 15 43 10 15 43 13 15 43 15 15 43 18 15 43 20 15 43 23 15 43 25 15 43 28 time 1 blue tas m s 2 red YertLoserD 3 Black HorizLaserD Figure 11 Note Click update after changing any options in the zoomed data window 5 5 2 Comparing time series Use the CompTMS button to compare time series parameters In this window different sizing methods arrays and parameters can be plotted and compared Sizing methods are described in Appendix A ASCII and image files can be saved from this window SJ cmp_tms data from 15 43 08 000 000 000 to 15 43 28 000 000 000 10000 H 20000 Data Points Help Notes save_contig load_contig save_png Print ave_ascil secon 20 Nbine 300 Reset Update horiz070806121344_154254_154348_0000 bin dT 20 0s 15 43 08 000 000 000 15 43 28 000 TC F 1 2x10 Compare Time Series df conc Trace 1 1 0x105 MIMI 7 CIH 7 Picone Y _ B 0x104 Trace 2 7 6 0x1G lt MIMI civ z plconc z 4 4 0x10 i l 4 f Ymin 0 00000
18. Navigate to the directory containing the SORTED HVPS requiring processing Click on one of the sorted files and then click open on the lower right corner That please select a file window will disappear SI SPEC 2DS iew ersion 0 59 E Oj x File ar H y Q CleanAl CleanNoise 0 9 F UselnFocusony Archive Multi_Liq asci x Archive_this_png y Make N_asci_help 7 Please Select a File Look in lan_June_1_2009 J eE Pe SORTEDO81031123842_123921_123941_0001 2D5 a 3 SORTEDOS1031 123842_123939_124004_0002 2D5 My Recent 2 SORTEDO81031123842_124002_124022_0003 2D5 Documents SORTEDO81031123842_124020_124049_0004 2D5 png Drop Down List 3 SORTEDO81031 123842_124047_124110_0005 2D5 8 SORTEDOS1031123842_124108_124132_0006 2D5 Desktop 2 SORTEDOS1031123842_124130_124156_0007 2D5 13 SORTEDO81031 123842_124154_124223_0008 2D5 S 4 SORFEBRETI 1 123842_124221_124246_0009 2D5 3 SORTEDO81031 123842_124244_124307_0010 2D5 My Documents 8 SORTEDO81031123842_124305_125943_0011 2D5 Select any a SORTED081031123842_125941_130003_0012 2D5 wy la SORTED081031123842_130001_0013 2D5 SORTED 2DS file My Computer kic 4 gt a O Filename S0RTED081031123842_123921_123941_000 7 Places Files of type Soe 205s 0 Cancel ASCII Drop down list with Archive option Figure 24 4 A pop up window with the ice archiving options is shown
19. Si PC3 xy 48 618 6 735 PC4 xy 0 500 6 500 Figure 16 Another way to filter the data in a very specific way is to use what has been named the in focus only option To access this go back to the first main HVPSview window and click the use in focus only check box You must also select a value from the drop down list Value options are 0 9 0 8 and 0 7 This filters the data to use only those images that are at least 90 80 or 70 shaded This could be accomplished via the Boolean option as well The difference is that when using the check box the sample volume is adjusted using depth of Field values for in focus particles 3 HVPS data Batch processing Archiving As described in playback processing section 1 0 after many split sorted files are generated the data can be processed in batch mode allowing the computer to run over night When processing batches all the SORTED files are converted to BINS and an All Clean filter is applied to them This allows large quantities of data to be processed and filtered for use at a later date The batch processing can be done in Liquid mode or Ice mode When the user knows if the 2 DS probe is flying through liquid cloud by looking at the temperature or by observing the playback image types the batch processing should be done in Liquid mode When the probe is flying in the ice cloud the user should use Ice mode 3 1 Liquid mode Batch processing 3 1 1 Archiving ASCII Files
20. TC4_Steve_gains_of_61bins _HorV pro Ctrl F5 5 Click Run gt Run combine_archive_TC4_Steve_gains_of_61bins_HorV pro F5 Personal use license for Paul Lawson only File Edit Search Run Project Macros Window Help z Compile Combine _Lig_Ns_archiwe_Files_of_61bins_HorY pro aL ee ee Compile Combine Liq _Ns_archive_files_of_61bins Hor pro From Memory g Compile All Run Combine Lig Ms_ archive Files _of 61bins_Hory K a PE Resolve Dependencies CEF Frofile Test Gil trl 4 T Break Ctrl Break Stop Ctrl F Reset step Imta Fo Step Ower Flo Step Gut Gtrl Fa Thrace Rum bo Gursor ime Figure 17 6 A prompt you to select H or V this determines whether it is the horizontal or vertical files being combined 7 Go to the directory containing the files select a file 8 In the IDL command line type y gt Enter From then on all the H_archive HVPS_ Ns txt files will be read and combined in a single archive file saved in the same directory as the smaller archive files 3 2 4 Additional Notes e To batch archiving with the clean option of removing noise only you should click the archive multi Ice ascii WS button instead of the archive multi Ice ascii button and follow the same instructions as above To do batch archiving without removing noise you should click the archive multi Ice ascii NC button instead of the archive multi Ice ascii button and follow the same instructio
21. _124305_125943_0011 2D5 Select any 3 SORTEDO81031123842_125941_130003_0012 2D5 BE la SORTED081031123842_130001_0013 2D5 SORTED 2DS file My Computer i 4 gt NEE File name SORTEDOS1031123842_123921_123941_000 7 oe Files of type sos 0 Cancel eg ASCII Drop down list with Figure 20 Archive option 4 Return to the IDL front screen All of the files for overnight processing will be displayed 5 If all the files are present type y gt Enter in the IDL command line This will make the computer process all the sorted data files in the selected directory mw Here are all Sored 205 data files which are Found in current directory for overnight processing SPE SSE EDAD SSE STATA SAE SBD ASAE SASS SAS SA SO CAlarn lar_June_1_2003 SORTEDOSIOS1 123842 123642 123923_0000 2D5 CAlanklan_June_1_2003 S0RTEDOSOS1 123842 123921_125941_0007 205 CAlarn lar_June_ 1 2009 SORTEDOSIOS1 123842 1239939124004 0002 205 CAlanslan_June_1_2003 S0RTEDOSIOS 1 12s842 124002 124022 0005205 CAlan lar_June_ 1 2009 SORTEDOSIOS1 123842 124020 124049 0004 205 CAlanslan_June_1_2005SS0RTEDOSOS 123842 124047 124110_0005 205 CNlarnslar_June_1_2009 SORTEDOSIOS1 123842 124108 124132_0006 205 CAlanslan_June_1_2005 S0RTEDOSIOS1 123842 124130124156 0007 205 CMlarnklar_June_1_2009 SORTEDOSIOS1 123842 124154 124225 0008 205 CAlanslan_une_1_2005SS0RTEDOSIOS 12s842 124221_ 1244246 0009 205 CMlanlan June _1_200950RATED081031123042_124244_
22. as well To view them click on the Particle Size Dist Plots tab These are also crude quick look products using 1 Hz data files The zoomed in data window allows for a much more in depth data analysis In order to access these options select Hi Res and click UPDATE The start and end times can be changed by typing new time values in the Stime and Etime boxes and clicking the UPDATE button To save any time series or PSD plot as an image click the Save png button in the droplist To save the plot as an ASCII file click the Save Ascii button To print a plot click the Print button The ST button on the left is for displaying stereo images which 1s currently not used The HK is for displaying the Housekeeping data Figure 14 is an example of the plots obtained by selecting HK This data is used primarily by SPEC Inc It can also be used for troubleshooting Sj HKzoomed 15 43 08 000 000 000 to 15 43 28 000 000 000 E 0 x Help Notes save_config load_contig save_png Print save_ascil Reset z070806121344_154254_154348_0000 bin HK dT 52 0s 15 43 08 000 000 000 15 43 28 000 w Al Zoom Pannel 1 default H diode 2 6 on Tracel Hdiode0 v 2 4 2 2 trace2 Hdiode64 gt 2 0 k Trace3 Hdiodel27 Y 1 8 1 6 2 OO eee yn 15 43 08 15 43 10 15 43 13 15 43 15 15 43 18 15 43 20 15 43 23 15 43 25 15 43 28 N time 1 blue HdiodeO 2 red Hdiode64 V 3 Black Hdiode1 2
23. criteria k is the fraction of events eliminated by the criteria given a true waiting time distribution with mean of TaW 1 e is the probability of or fraction of events with inter arrival times less than C 1 e is the fraction of events eliminated due to the first 2C Taw criteria leaving not eliminated by the first criteria Of those a fraction l1 e _ are eliminated by the second criteria
24. e window should look similar to the one below Make sure HVPSview is at the top of the search path Preferences ew C MisherkSPEC 20S SH PS view HPS view lt IDL_DEFAULT gt LC fishersw JC hishenSPEC ZDS 2d 2eedsview verna Figure 3 3 Click OK 4 In the command line at the bottom of the window type compileHVPS gt Enter NOTE IDL is not case sensitive so it is OK to type compilehvps and hvpsview 5 Type in the command line HVPSview gt Enter The following GUI window will open up SI SPEC H PSView Version 1 0 ajal x rea v e ose r Archive this liqasci Archive this_png Make_N_asci_help Figure 4 2 2 More pre processing Before performing quantitative analysis some additional pre processing steps must be done First BIN files are needed 1 Click the left folder icon P This Brings up a list of all the sorted files created during the playback 2 Select the SORTED file with the desired time interval 3 Once the file is loaded a window displaying warning messages will pop up There may be messages regarding time inconsistencies found while processing the sorted file Usually there are no messages as the case shown in Fig 6 Sj Messages generated while loading the sorted data file g ol x File Array dimensions must be greater than Unable to display E 00_data_for testing ver sorted0 70606121344 154052 154056 0000 warr 7 Figure 6 Example of a warming m
25. ea PSD mm7 L uum For each size bin the formula for bin height is K x Adj x A bin iah x SV sep where the sum is over all events in that size bin and A Area x pxs x strbsz 1000000 and where K unit conversion factor Bin as above for counts PSD Ice Mass PSD g m um Adj x minfa x A p iz x L x pxs Kisa E E We A Area x pxs x strbsz 1000000 a 0 115 B 1 218 o 0 917 and Ky and K are unit conversion factors Liquid Mass PSD g m um K x X Adj x7 x Ly x pxs bin _ width pan x SV penn where K unit conversion factor Method 3 M3 All in simple non adjusted spheres For each time bin of size At concentration EX Lid where SV default s SV Jefau SA ge fau Adj ae ay A where SA 127 L x min pxs strbsz x min d dy 3 and Adj O if either edge pixel is shaded F 0 Where dy mm 5 134 x Ly x min pxs strbsz 4m 1000 and where K unit conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Counts time series Simply plot the number of accepted particle events per time bin a i Z Kx Adj x L x pxs x strbsz Extinction Km 2 7 where K unit default conversion factor Rejects events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 T 3 0 917 x K x Adj x x L x 4 pxs x strbsz IWC g m 6 SV where K unit d efault conversion factor Reject eve
26. ed as follows S V am S A default Adj SV a SA na S Vojan TAS x At x S A default SV a TAS i At A SA a SV stands for sampled volume Liters SA for sample area mm2 TAS m s the speed of air through the probe sample area and At seconds the time period of the measurement SA team 128 x pxs x d Where pxs 0 01 mm 10 um is the optically realized size of the array elements and dww 63 mm 6 3 cm is the window to window distance Strbsz is the pixel size in the TAS direction strobe size and should also 10 um but could differ if the aircraft flies too quickly or an incorrect TAS is sent to the probe during data acquisition SA 127 L xstrbsz pxs x pxs x min d d where dor stands for depth of field and d mm F x L x strbsz 4m 1000 where Fagor 5 13 um PC can be found as the halfway point between the edge points of the image on the array for the slice that defined L PC can be found as the center of the circle that defined L3 Methods Equations Method 1 M1 TAS direction all particles For each time bin of size At Concentration L K x ay Where K unit conversion factor default Counts time series the number of particle events per time bin Extinction Km _ 2xKx gt Adj x Area x pxs x m Whee K limit default conversion factor K gt Adj x minfa x A po Zy L x strbsz J IWC g m K x pie where A
27. equires the Spec2D Executable written in C and Visual Basic and referred to as Playback After some initial processing with Playback the main processing and display of HVPS data is accomplished via HVPSview an IDL based program After a data file has been recorded the first step to analyzing the data 1s to open the Executable program which is the same program used to record the data file Make sure that HVPS Mode is enabled and that 150 microns per slice is set in the Display window Click on the folder icon It will prompt you to open a baseY YMMDDHHMMSS HVPS which is a raw data file It is binary and the images are compressed Once a file is chosen a time period selector box will be displayed Select Playback Period Data range of base060201161320 2D5 File Splitting Start Date Time 02 01 2006 f 16 13 26 625 Enable File v Splitting End Date Time 02 01 2006 19 28 57 234 m Select Playback Range ul Offset Particles 7 File Start Time 02 01 2006 16 13 26 625 i T _ 600 000 End Time 02 01 2006 19 28 57 234 pi 7 _ i j Start Playback Cancel Playback Figure 1 To select a time period move the two sliders on the Start Time and End Time For an exact time period use the U and D buttons to increase and decrease time by one second intervals To the right is the File Splitting option HVPSview attempts
28. ering You do not need to use them 2 5 5 Boolean Filtering The Boolean feature allows for further sub setting of data Itis possible to filter out particles based on size location and various other parameters To use Boolean filtering 1 From the zoomed data window select Boolean On as shown 2 Specify the filtering parameter from the drop down list 3 Use Boolean logic from the drop down list to add filter conditions 4 If using numerical values insert a space between the logic and numerical value 5 Hit UPDATE The IDL syntax for the Boolean symbols is defined below And And Or Or Not Not Eq Equal to Ne Not equal to Ge Greater than or equal to Gt Greater than Le Less than or equal to Lt Less than Multiply Addition Subtraction Used for order of operations For example if user types L1n gt 10 are entered the filter box is checked and the Update button is clicked only those accepted and rejected particles with size Lln greater than 10 pixels will be selected and displayed in the H strip and V strip fields These will also be used to make the time series and PSD plots Figure 19 shows an example of when Lin gt 10 was input for updated processing SJ Zoomed Showing Horizontal channel from 15 43 21 668 681 960 to 15 43 25 592 876 993 15 43 23 634 200 849 PCI 50 Mi Bas aoe ve A Cleaning by start End Clicks Label_one_image_as_bad_with_one_click Off C On
29. essages window Once a file is selected a status bar in the lower left of the GUI window shows the number of particles counted The program will automatically generate the files required All the BIN files are placed into a subdirectory labeled data This process is only required one time for each sorted file The program automatically looks in the data subdirectory and uses the appropriate BIN file if it already exists from a previous running Note Due to file size the time to create BINS can be quite extensive but must be done for each SORTED file To continue open a BIN file 1 Click the right Folder icon 2 Open the Data folder 3 Select a BIN for the desired time Again additional files are created in the data subfolder if they do not already exist These are 1 Hz data files There will be several pop up windows depicting the raw data and the images are automatically saved in the data folder These auto saved files are labeled horiz png Figure 7 is an example of the type of window that will appear The images can be viewed at any point from this point forward These are created primarily for SPEC programmers to detect and debug idiosyncratic corrupted data problems F Unsorted Particle_nanas 109 Oo 1210 4 5705x10 4 5700 107 4 569510 4 5690 107 4 568510 4 5680 107 if ae ir 4x10 Axio axic 1x10 UnsortedY Particle_ticks 107 200 150 160 Ticks i 2x10 4x10 Axio axic 1x10
30. f L 20 5 x Ls E an image is determined to be round if L 2 0 5 x Ls and Ls 20 5 x L or Ls gt 50 2 The larger a valid image is the greater the percentage of shaded pixels Splash effects often create large images that have lower percentages of shaded pixels than valid images The follow criteria are used to eliminate such spurious images If any of these five criteria are met the image Is rejected e L5 gt 10orL1 gt 10 and At gt 3 0 X As e L5 gt 150rL1 gt 15 and At gt 2 5 X As e L5 gt 200rL1 gt 20 and At gt 2 0 X As e L5 gt 35orL1 gt 35 and At gt 1 5 X As e L5 gt 2 2 XL7 and L1 gt 10 0 3 Figure B 2 shows some examples of noise generated images appearing in line plus dot patterns 2 m om s 8 e Figure B 2 Examples of line plus dot patterns caused by noisy diodes These are eliminated using the following criteria i L1 As and L2 1 and L1 gt 4 ii As lt 1 35 X L1 and L4 L5 and L1 gt 4 and L2 2 iii L1 gt 10 and L1 gt 0 75 X As and L1 lt 1 5 X As iv L4 L5 and At gt 0 9 X L1X L5 and L2 2 and L2 L4 v L4 L5 and At gt 3 0 X As and L2 2 vi L4 L5 and At gt 4 0 X As If any of these six criteria are met the image is rejected 4 The final step of this first cleaning loop is another noisy diode removal method based on the statistics of image center locations calculated over 4000 images approximately centered on t
31. he image being evaluated When a diode is noisy there are more image centers located on that diode than are located on quiet diodes A diode is labeled bad when it has more particle center locations than the threshold value TH max M 5 x JM 1 5 Where M is the mean number of image centers per diode across the array for diodes having more than a minimum number of image center counts images centered on this diode This minimum number is Mt 3xVMt where Mt is the mean number of image centers per diode across the array for all diodes If less than 33 diodes satisfy the requirement of having more than Mt counts then M is calculated from all diodes Mt and M are recalculated after bad diodes are identified ignoring those diodes and the process repeated until no diodes exceed the threshold An image whose center falls on a bad diode is rejected unless it meets one of the following criteria i L1 215 and L5 2 15 and As gt 0 7 X At li L124 and L5 2 4 and As 2 0 5 X At and L2 gt 0 25 X L5 and L1 lt 50 iii L2 L4 and L2 L5 and L2 2 2 and As 2 0 5 X At Figure B 3 shows an example of noisy diode data and the distribution of image centers across the array 120 100 106 80 62 102 121 6 ooog 70000000000000 00000000000000000000000000000000000O0000000000000000000000000000000000000 O00000000000000000000000000000000000000000000000000000000 00000000000000 0000000000 0 OOOO N AIN dal
32. he previous second are used instead of 5000 particles if there are less than 5000 particles in the previous second similarly for the following 5000 particles or second If aveW8 gt 40000 then C 2000 and C 8000 If aveW8 lt 40000 then C 0 05 X aveW8 and Co 0 2 X aveW8 A particle is rejected if the precipitation status is yes and either a the particle s inter particle time or the following particle s inter particle time is less than C or b both the particle s inter particle time and the following The precipitation status and current mean waiting time are updated every particle instead of every 100 if the valid particle rate is slower than 1000 per second averaged over the surrounding 0 2 seconds particle s inter particle time are less than Co Finally a variable adjo is calculated for each image It is O for rejected images 1 for accepted images if precipitation status is no When precipitation status is yes then for accepted images we calculate a new true average inter particle time TaW the average inter particle time of particles accepted by all the steps minus C Then calculate 2C TaWw 2C TaWw C Taw kel er e TE og a k is limited to a maximum of 0 9 and then adj adjo is used elsewhere in the algorithms for calculating concentrations etc to increase the weight of each accepted particle to account for those good particles rejected by the inter particle time
33. ice water content will be plotted when applicable S cmp_psd data from 15 43 08 000 000 000 to 15 43 28 000 000 000 horiz070806121344_ 154254 154346 0000 bin m iol xj Her Sev tia Frees LoadCtg SevPna Pint Savt Reset xlog 7 xnfi0 xm 2000 Vlog 7 Update more_annatation 1 MjM1 7 CIH 7 Plbinc Mmfm1 7 Cl Plbinc i 2MiM1 7 CIH 7 Picone Mt clv P conc 15 43 08 000 006 000 15 43 28 600 000 0G0 horizQ 78061 21344_154254_154348_G00G bin dT 20 08 OGO 19 bing bin 7 8dBz Ra I ro Ow 107 E a g 107 N 10 be E E 3 a a 10 A lt 1 a E 107 107 107 10 100 100C 10 100 1000 Particle Sizefum Particle Size urn criteria None 3 MiM1 7 CIH 7 Plaea 7 M M1 7 C V 7 Plaea emm cf PiLwe 7 MjJM1 7 CIV PilLwe Figure 13 2 5 4 Viewing the HVPS images from a selected time interval 1 In the zoomed data window Figure 13 check the box to the left of Filter if accepted images only are desired for both display and processing Unclick the filter box to process and display using all the images accepted and rejected Rejected images will be highlighted with a yellow background Ki gt gt gt Speed in Secs 3 F 1 tan 0 10 Cleaning by start End Clicks L4 14 L5 14 Label_one_image_as_bad_with_one_click Off C On L7 14 00 L8 13 99 PC3 xy 0 000 0 000 PC4 xy 0 000 6 500 NoiseType 00000000 Area sq
34. ives 1 100 0 01 and which meanes if Lin GE 0 01 L5 and L5 GE 0 01 L1n or L5 gt 50 then roundness 1 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX L1_L5_ spect_Ratio Accept Figure 8 If a cleaning method is not selected all the images will be used The resultant data is automatically saved in the directory If a cleaning function is selected the program searches for previous cleanings made for each BIN file If one is found you will be prompted with the window shown in Figure 12 If the aspect ratio is the same select yes In doing so the computer will not have to re clean the data question a Zj 2 000 is the Aspect Ratio for roundness defintion Clean_ All or Keep SPlash_option All Use ik to load the cleaned result A Previous AutoCleani v tst File with Aspect _ratio and Cleani Or cleant header lines are ee Figure 9 Note The program processes each particle individually based on statistics of many surrounding particles processing large files can take days The In Focus Only feature allows for further data filtering This removes the particles with holes in them If the area of the hole is greater than the specified size the particle is eliminated from the data For example if 0 9 is the selected criteria then only those images with holes having less than 10 of the total area will be used in the process To use the In Focus Only Feature 1 Check the box to the left of Use In Focus On
35. locations Second Loop 5 Test for splashing events based on inter event distances if the probe is in precipitation Each step will be described in detail after defining terms Four measures of image length are shown in figure A 1 L is the number of slices pixels in the direction of travel for which the event lasted L is the number of shaded diodes pixels in the direction along the array for the slice for which the same quantity is maximized L is the number of diodes between and including the shaded end diodes for the slice that maximizes the same quantity Ls is the distance between and including the shaded end diodes considering all of the slices together Other size parameters are a corrected size for out of focus images based on Korolev s work L7 the number of shaded pixels for the entire image summed over all slices A and an estimate of total number of pixels shaded or not for the entire image A The algorithm for A is part of the Korolev size correction method II pen L m A on c Figure A l Two example particle images designed to demonstrate the four measures of image size described in the text 1 An image is determined to be round if L 20 5 x Ls and L5 2 0 5 x L4 lf an image is not round it is rejected An exception is made for the very large images that do not fit within the array If Ls gt 50 500 um then an image is determined to be round i
36. ly on the toolbar 2 Select a value from the drop down list 0 9 0 8 0 70 2 5 Time Series and Particle Size Distribution PSD Plots 2 5 1 Zooming into a time interval 1 Click on the magnifying glass icon aj 2 Click one of the three time series near the desired start time for analysis 3 Click the desired end time for analysis This creates a smaller time interval to be examined Choosing small time periods significantly reduces subsequent processing time Similar to the previous window this new window initially displays concentration extinction and ice water content read from the 1 Hz data file described above on single time series plots both H and V are shown simultaneously here Figure 13 shows the zoomed in data window Sj Zoomed data from 15 43 08 000 000 000 to 15 43 28 000 000 000 10000 H 20000 Data Points Time Series Ylots Particle Size Dist Plots UPDATE Concentration Ice Water Contant Extinction Particle Concentration H BLUE Y RED Deadtime H 0 000 Secs V 0 000 Secs 15 43 08 15 43 10 15 43 13 15 43 15 15 43 18 15 43 20 15 43 23 15 43 25 15 43 28 Bin Particle Extinction H BLUE Y RED oO 5 15 43 06 15 43 10 15 43 13 15 43 15 15 43 18 15 43 20 15 43 23 15 43 25 15 43 26 in Particle IWC H BLUE V RED 0 08 0 06 0 04 0 02 0 00 m 15 43 08 15 43 10 15 43 13 15 43 15 15 43 18 15 43 20 ee Figure 10 15 43 23 15 43 25 15 43 28 PSDs are plotted
37. ns above Clicking the Archive this Ice ascii or archive this Ice png in the initial zoomed window will archive the current loaded data file in whatever cleaning state the data is in If it is cleaned already it will archive it using only the accepted images If it is not cleaned it will use all images to do the archive etc The make multi pbyp asciis is a spare button for future use and it currently has no function Clicking any of the Make N ascii Liq Make N ascii ice Make N ascii Liq WS Make N ascii ice WS Make N ascii NC will batch generate many zoomed txt ascii files that contain information of each particle in each line for users to do their own data processing and analysis if they choose to Appendix A Derived Parameters Methods Equations in HVPSView Derived parameters are as follows Timestamp Index of same event s image in the sorted file A number of shaded pixels L number of pixels in TAS direction that the event lasted for L2 number of shaded pixels in array direction for the slice during the event for which the number of shaded pixels is maximized Lz the diameter of the circle that just circumscribes the shaded event pixels L4 similar to L except it is the total number of pixels between the shaded end pixels instead of just the shaded pixels Ls 1s similar to L4 except it is the distance between
38. nts for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 IC 3 Kx y Adj x x L x 4 pxs x strbsz LWC g m aAa 6 k SV Where K unit d efault conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Counts PSD For each size bin the formula for bin height is simply the number of particles in that size bin Use size bins with width equal to pxs x strbsz for now Bin L x pxs x strbsz into the size bins The first size bin has edges of 0 5 J pxs x strbsz and 1 5 y pxs x strbsz the n size bin has n 0 5 pxs x strbsz and n 0 5 pxs x strbszas bin edges Reject events for which L4 gt 2 1 L1 or L4 lt 0 49 L1 and area gt 2 Concentration PSD L um For each size bin the formula for bin height is K xX Adj ee ie dth pam x SV pran where the sum is over all events in that size bin and where K unit conversion factor Bin as above for counts PSD Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Area PSD mm L um For each size bin the formula for bin height is x K x gt Adj x 1 x pxs x strbsz bin _ width um x SV gpa where the sum is over all events in that size bin and where K unit conversion factor Bin as above for counts PSD Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Ice Mass PSD g m um For each size bin the formula for bin height is 0 917 x K x P Adj x x L x pas x strb
39. option to fill these lines within particle images Click tool then mask bit replacement a window entitled Mask Bit Recovery settings appears Enable mask filling here and set the maximum number of consecutive masked diodes that can be filled Filling these artificial white lines is important for the measurement of area which also affects length when sizing corrections are used It is even more important to filled masked diodes when they are the edge diode since that diode is used to determine if an image is complete or not q Image with white stripe p File Tools Window Help Maximum Playback RUN xX Async Sebup Bitmap Display Rate gt Analog Settings Housekeeping Display Processed Data v Use Display Separator c Pixel Histogram Raw Data Processing Mode Stuck Bit Handling Current Do Nothing TAS Source Set Probe IP Address Record Speed Mask Bit Replacement w Mask Bit Recovery Settings Ioj x Masmum width of mask bits that Will be recovered 5 Enable Mask Bit Recovery a Include particle offset in output Ignore Housekeeping PC time differences Save Command Info to Disk Save E mit Figure 1 Example of the display used to enable masked diode filling 1 2 Dead diode replacement Similar to a noisy diode a dead diode leaves a white line running through particle images In this case however the real time software cannot detect and mask the diode
40. ounts PSD For each size bin the formula for bin height is simply the number of particles in that size bin Use size bins with width equal to y pxs x strbsz for now Bin L7 x pxs x strbsz into the size bins The first size bin has edges of 0 5 J pxs x strbsz and 1 5 pxs x strbsz the n size bin has n 0 5 4y pxs x strbsz and n 0 5 J pxs x strbsz as bin edges Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Do not adjust bin heights by multiplying by the ratio of total events to accepted events Concentration PSD L um For each size bin the formula for bin height is K x Adj ae Nile ce ath sam x SV prun where the sum is over all events in that size bin and where K unit conversion factor Bin as above for counts PSD Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Area PSD mm7 L uum For each size bin the formula for bin height 1s 5 x K x gt Adj x L x pxs x strbsz bin _ width pam x SV popan where the sum is over all events in that size bin and where K unit conversion factor Bin as above for counts PSD Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Ice Mass PSD g m um For each size bin the formula for bin height is 0 917 x K x P Adj xZ x L x prs x strbsz bin _ width pam x SV sets where K unit conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 LI and area gt 2 Liq
41. ow for moving the view one strip at a time The icons with two arrows gt gt fast forward and rewind The drop down list to the left of the arrows controls the length of time each strip is in the window while fast forwarding and rewinding Placing the cursor over single particles displays the size and other parameters in the lower left hand corner An enlarged image of the particle is displayed as shown in Figure 10 2 4 Artifact Removal The BIN files contain all the data captured by the HVPS Probe This includes noise splashed and shattered events as well as valid images The program has the ability to remove unwanted data Using the cleaning functions creates concentrated data for the time series and particle size distribution plots Before quantitative analysis there are two options for data filtering Clean all Removes noise shattering and splashing Clean noise Removes noise only To use the cleaning functions 1 Specify the type of cleaning by clicking the corresponding icon on the toolbar 2 A text box will ask for an aspect ratio as shown in figurell If the cloud is known to be liquid enter in 2 0 If the cloud is known to be ice enter a larger value ioi please input your roundness aspect ratio definition For Liquid The aspect ratio should be 2 0 which gives 1 2 0 5 and which meanes if Lin GE 0 5 L5 and L5 GE 0 5 L1n or L5 gt 50 then roundness 1 For Ice The aspect ration should be 100 0 which g
42. s follow these steps 1 Copy all H_archive_HVPS _Ns txt into one folder or directory 2 Start IDL 3 Click File gt Open Find HVPSview_ver gt combine_Lig_Ns_archive_61bins_HorV pro 4 Click Run gt Compile combine_Liq_Ns_archive_61bins_HorV pro Ctrl F5 5 Click Run gt Run combine_Liq_Ns_archive_61bins_HorV pro F5 Personal use license for Paul Lawson only File Edit Search Run Project Macros Window Help E c a Compile Combine_Lig_Ns_archive_files_of_61bins Hor pro Compile Combine_Lig_Ns_archive_Files_of_61bins_Hor pro from Memory ee Compile al Run Combine Lig _Ns_archive_files_of_61bins Hor K go be Resolve Dependencies Profile Test GUI rT Break Ctrl Break Stop Ctrl F Reset Step Into Eg Step Giver Flo Step tut atr FE Trace Rum bo Gursar Fy Figure 23 6 You will be prompted to select H or V This determines whether it is the horizontal or vertical files being combined 7 Go to the directory containing the files gt and select a file 8 In the IDL command line type y gt Enter From then on all the H_archive HVPS_ Ns txt files will be read and combined in a single archive file saved in the same directory as the smaller archive files 3 1 4 Additional Notes To batch archive with the clean option of removing noise only you should click the archive multi Liq ascii WS button instead of the archive multi Liq ascii button and follow the same ins
43. sz bin _ width um x SV jojauts where K unit conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Liquid Mass PSD g m um K x Adj x x L x pxs x ee bin _ width pam x SV pepan where K unit conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 Method 4 M4 All In Ring Spot Adjusted Spheres A For each time bin of size At Concentration Liti where Adj ir ew where default SV SA oa SA 127 L x min pxs strbsz x min d d 4 and Adj 0 if either edge pixel is shaded F 0 Where dy mm 5 13 4an x L x min pxs strbsz 4am 1000 and where K unit conversion factor Also reject events for which L5 gt 2 1 L1 or L5 It 0 49 LI1 and area gt 2 Counts time series Simply plot the number of accepted particle events per time bin IC x K xX Adj x L x pxs x strbsz Extinction Km 2 i SV where K unit default conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 T 3 0 917 x K x Adj x x L x y pxs x strbsz IWC g m t 6 SV where K unit default conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 IC 3 K x Adj x x L x 4 pxs x strbsz LWC g m a SV where K unit default conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 LI1 and area gt 2 C
44. tc in the directory where the original SORTED HVPS files are stored These files are initial plots of the file and warning messages encountered when loading the data file the auto clean results and the horiz bin and vert bin files The most important two files are H_archive HVPS xxxx Ns txt and V_archive HVPS xxxx_ Ns txt They are located in the subdirectory png under the datal data2 data3 and etc These two files contain the 1Hz concentration extinction ice water content time series and particle size distributions The header lines contain the detail data format of the archive file 3 1 2 Archiving Images After H_archive HVPS_ Ns txt and V_archive HVPS_ Ns txt 1Hz archive files are generated for each sorted HVPS file the HVPSview software can be used on the data To archive the images 1 Select Archive Multi Liq png from the png drop down list see Figure 20 2 A window will ask how many images per second to archive This is dependent upon cloud type and computing power Type the desired number of images per second and press Enter Only the first images up to that number will be archived each second The images are saved in the subdirectory png under the corresponding datal data2 data3 etc subdirectory 3 1 3 Combining the archived liquid ascii files To combine all the smaller archive files into a single data file for the whole flight a separate IDL program is used To combine all ascii file
45. ticle Cencentration ieee 3x10 f 2x107 1x10 Concentration H Taigi z Ta ire 125 12041550 250 meter ees 124138 500 12 41 39 625 Laie Pod 124i ai 1 2341 49 00 In Vertical Particle Extinetion 20 1214192 4 000 qapi 12h 122417550 250 12 41 33 375 12341258 500 12141539 625 12rd 42 7 a0 12rd 18 87h 1234149 000 in Vertical Particle IWE lee Water Content Extinction ta EA gA EA ea Bom Of L fa ote P in om be aed HAH Teg 124 000 Tas lee 125 1241 30 250 1241 31 17i 12 41 36 T2 12 41 39 625 12142 7a 12 41 45 875 12341 4902 Move cursor to aranh to select specific data Figure 6 By default the horizontal array time series are displayed The arrays can be switched by clicking on the Y vertical or horizontal icons located on the upper toolbar 2 3 Viewing Images To view the particle images before artifact rejection has been applied 1 Choose the desired array either vertical or horizontal 2 Move the cursor over any one of the three plots near the time you want to view 3 Single click a window similar to Figure 10 will open S Showing Horizontal channel from 16 47 35 095 179 304 to 16 47 35 224 014 608 g kd 10 x oa 2 A p z x Sa a g e se i 16 47 35 185 110 695 Speed ir ecs Lenath L1 um 70 44 4 gt gt gt Height L2 um 50 Figure 7 The arrows are used for changing the time interval The single arrow icons gt all
46. to keep information on each particle in memory When the memory is full the program slows due to the use of virtual memory This can result in unacceptably slow processing This can be remedied by enabling file splitting which cuts the original large data file into processed data files of manageable sizes according to a preset number of particles A typical setting is 300 000 particles per file but the optimum size will depend on the attributes of the computer used Once the time period and file splitting options are chosen click the Start Playback button A window will ask whether to create HVPSview pre processed files which are called sorted files These files are also binary with compressed images but various simplifications to the file structure have been made For example the data is sorted into time sequence A sorted data file is necessary to run HVPSview Playback also produces HK_BINY YMMDDHHMMSS HVPS HK_TXTYYMMDDHHMMSS HVPS and PAIRSYYMMDDHHMMSS HVPS files The first two contain probe housekeeping data in binary and ASCII forms respectively Thus these files allow the user to obtain the housekeeping data independently of HVPSview The Pair file is now obsolete Once you click yes you will return to the Main Display window SPEC 2D 5 Data Collection and Processing Program Yersion 1 0 57 a Pixel Histogram Analysis L27S9E74 224B6E 00 59 02 148 570 757 L27946F1 225688 00 59 02 196 629 286 12798686
47. tructions above To do batch archiving without doing any removal of noise click the archive multi Liq ascii NC button instead of the archive multi Liq ascii button and follow the same instructions above Clicking the Archive this liq ascii or archive this Lig png in the initial zoomed window will archive the current loaded data file in whatever cleaning state the data is in If it is cleaned already it will archive it using only the accepted images If it is not cleaned it will use all of the images to do the archive etc The make multi pbyp asciis is a spare button for future use and it currently has no functionality Clicking any of the Make N ascii Liq Make N ascii ice Make N ascii Liq WS Make N ascii ice WS Make N ascii NC will batch generate many zoomed txt ascii files that contain information about each particle in each line for users to do their own data processing and analysis if they choose to 3 2 Ice mode Batch processing 3 2 1 Archiving Ice ASCII Files 1 Copy all sorted HVPS and the associated HK_bin HVPS HK_TXT HVPS and Log_TXT HVPS files into the same directory 2 Get IDL started and follow the steps described in section 1 1 1 3 The initial GUI screen will appear Select archive multi Ice ascii from the drop down list A dialog window will appear for the user to select an input file of type Sorted HVPS as displayed in Figure 20
48. uid Mass PSD g m um K x Adj x L x a PXS X sirbsz bin _ width um x SV iepa conversion factor Reject events for which L4 gt 2 1 L1 or L4 It 0 49 L1 and area gt 2 where K unit Method 5 M5 All In Ring Spot Adjusted Spheres B The same as Method 4 above except replace all L7 with Lg Method 6 M6 in focus only combined This method is for when there are sufficient particles to use in focus only and when there is need to extend to larger sizes than can be adequately sampled by all in techniques It uses M2 1n focus only after CleanAll for sizes smaller than 265 It uses M1 in focus only without cleaning for sizes greater than 325 For sizes between 265 and 325 it uses the average of the preceding two methods A future improvement should be to replace the M1 part with a reconstruction technique Appendix B Removing Spurious HVPS Events HVPS raw data include spurious effects These are primarily from instrument noise and from splashing of precipitation Algorithms used to clean the data of the majority of these spurious effects while retaining the majority of the valid images are described here There are 5 quasi independent steps to the algorithm implemented via two loops through the data First Loop 1 Test for roundness 2 Test for splashing events based on black and white area considerations 3 Test for noise via line and dot patterns 4 Test for noise via statistics of particle center
49. width um x SV Where K unit conversion default factor Method 2 M2 All in along array For each time bin of size At Concentration L ee where default SV efau SA efault 1 a oy _ os WETE SA na T027 Ly x pxs x minfd ww do 2 and Adjz 0 if either edge pixel is shaded F 0 Where dp mm 5 13 um x L x pxs m 1000 and where K unit conversion factor Counts time series the number of particle events per time bin Extinction Km 7 2 Adj x Area x paxs x strbs vs where K unit default conversion factor K f gt Adj x minla x A p eg ths x pxs IWC g m K x where SV vefault A Area x pxs x strbsz 1000000 a 0 115 B 1 218 0 917 and K3 and K are unit conversion factors Kx Adj Sd x pxs LWC g m 6 SV a Where K whatever you have to make it so the units come out Counts PSD For each size bin the formula for bin height is simply the number of particles in that size bin Use size bins with width equal to pxs for now Bin L x pxs into the size bins The first size bin has events for which L2 1 and the bin edges are 0 5 pxs and 1 5 pxs the n size bin has events for which Ly n and n 0 5 pxs and n 0 5 pxs as bin edges Concentration PSD L um For each size bin the formula for bin height is K x Adj at Oa em where the sum is over all events in that size bin and where K unit conversion factor Ar

Download Pdf Manuals

Related Search

Related Contents

  ご利用規約はこちら（135KB）  TX 2AD User Guide  Livret Accueil CADA 2009  Samsung 172V Käyttöopas  Samsung GALAXY S    

Copyright © All rights reserved. Failed to retrieve file