########################################################################## # # This file is part of FORCE - Framework for Operational Radiometric # Correction for Environmental monitoring. # # Copyright (C) 2013-2018 David Frantz # # FORCE is free software: you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation, either version 3 of the License, or # (at your option) any later version. # # FORCE is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with FORCE. If not, see . # ########################################################################## Changelog ########################################################################## ### FORCE v.2.1 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L2PS --- Bugfix: One line was missing in the parameter file generated by force- parameter-level2. You can now ignore the red comment in the 2.0 User Guide on p. 21. --- Bugfix: There was a bug in a bugfix introduced in v. 2.0.. When creat- ing datacubes (DO_REPROJ = TRUE and DO_TILE = TRUE), data were consis- tently missing in some areas at the South-East border of Sentinel-2 MGRS tiles. The occurence of this effect was dependent on the relative location of the MGRS tiles to the upper-left tile in the custom grid system. Should be all OK now. --- Bugfix: The cloud cover percentage (and similar) in the metadata was not correct for tiles that are located at the intersection of differ- ent MGRS/WRS-2 frames. Only the numbers provided in the QAI product were correct. Is now correct for all products. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ########################################################################## ### FORCE v.2.0 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ General changes --- Follow FORCE on social media. ResearchGate: https://www.researchgate.n et/project/FORCE-Framework-for-Operational-Radiometric-Correction-for- Environmental-monitoring. Twitter: #FORCE_EO --- There are many new features throughout all FORCE modules, and there are entirely new modules. The new features generally come with new parameters in the parameter files, which you need to add to your exis- ting parameter files (from old FORCE versions). If you are seeing an error like this, simply add the new parameters (have a look at the user guide for details and possible values): 'OUTPUT_FORMAT is not specified. Unable to read parameter file!' --- All products can now be output as ENVI Standard or GeoTiff. An ENVI header is written for both options. The first option produces uncom- pressed flat binary files. The GeoTiff come with LZW-compression using horizontal differencing. A new parameter OUTPUT_FORMAT was added to all parameter files. All FORCE modules can digest both formats. --- All FORCE modules are more verbose and will show some progress bar. --- The writing of all output is now accompanied by locking the file (even if it does not exist yet). This prevents possible data races when wri- ting one and the same output file from two threads. Note that there is the possibility that the lockfile cannot be generated, in which case no output can be written. If this happens, please refer to the FAQ section in the user guide to resolve this issue. This generally occurs if you have aborted FORCE before (lockfile was not deleted correctly) or if too much I/O is causing a timeout of the lock attempt. --- Directories are now created with 775 permissions instead of 700. Note that your umask settings might reduce these permissions. --- If there is a reading error for any file, the error message includes the filename. --- There are more metadata written to all outputs, e.g. the parameteriza- tion, information about your system, processing time, FORCE version, band names etc. --- A bibtex file including FORCE citations is now included in the sofware package. --- A bibtex file including studies that used FORCE, is now included in the sofware package. If you publish anything enabled or related to FORCE, please send me your published papers (pdf+DOI). They will be included in the bibtext file of the next version: david.frantz@geo.hu-berlin.de #FORCE_EO --- Recognizing the different user needs, quality screening for all higher level FORCE modules (above Level 2) is now in full user control. The parameter SCREEN_QAI was added to all parameter files. Have a look at the user guide for all possible keywords and quality filters. Please make proper use of the quality information and never analyze any data without. --- The sensor list (SENSORS = ...) in the parameter file of all higher- level FORCE modules (above Level 2) has two additional key words: sen2a and sen2b (all lowercase) refer to the original Sentinel-2A/B 10m bands only. Note that FORCE matches the Landsat NIR band with the original Sentinel-2A/B 20m NIR band; therefore, using "SENSORS = LND08 sen2a" will only return the VIS bands. Have a look at the matching table FORCE uses for combining several sensors, and read the next change entry. SENSOR B G R RE1 RE2 RE3 PNIR NIR SW1 SW2 LND04 1 2 3 - - - - 4 5 6 LND05 1 2 3 - - - - 4 5 6 LND07 1 2 3 - - - - 4 5 6 LND08 1 2 3 - - - - 4 5 6 SEN2A 1 2 3 4 5 6 7 8 9 10 SEN2B 1 2 3 4 5 6 7 8 9 10 sen2a 1 2 3 - - - 7 - - - sen2b 1 2 3 - - - 7 - - - --- The filenames of the higher-level FORCE outputs (above Level 2) have changed slightly. Before, the tag "MULTI" was a placeholder for the sensor. Now, it is one of the following and merely refers to a named set of spectral bands instead of individual sensors: 1) "LNDLG" refers to the 6 Landsat legacy bands (B/G/R/NIR/SW1/SW2). It occurs if you select one or several Landsat sensors (even if you additionally use SEN2A and/or SEN2B). "SEN2L" refers to the 10 Sentinel-2 land surface bands (B/G/R/RE1/RE2/RE3/PNIR/NIR/SW1/SW2). It occurs if you only se- lect SEN2A and/or SEN2B. "SEN2H" refers to the 4 high-resolution Sen- tinel-2 bands (B/G/R/PNIR). It occurs if you only select sen2a and/or sen2b. "R-G-B" refers to visual bands only. It occurs if you select one or several Landsat sensors plus sen2a and/or sen2b. See the match- ing table above for details. --- The spatial data cube definition is appended to each gridded data cube. The file contains (1) projection as WKT string, (2) origin of the tile system as geographic Longitude, (3) origin of the tile system as geographic Latitude, (4) origin of the tile system as projected X- coordinate, (5) origin of the tile system as projected Y-coordinate, and (6) width of the tiles in projection units. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L1AS --- force-level1-sentinel2 shows the filename of the current download, progress bar, downloaded size, download speed and eta per download. Note that the progress bar and eta are produced by wget. Afaik, there is no way to change the format. --- The L1AS programs have a new mandatory argument. The file name of the file queue needs to be given. This enables you to manage several pro- ject in one data pool. If the file queue exists, new files will be appended at the end. If not, the file will be created. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L2PS --- The GeoTiff driver is more picky with the output projection. Make sure to validate your WKT string using "gdalsrsinfo -v 'YOUR_WKT_STRING'". I have improved the definition of the default projection in force-para meter-level2 to work with GeoTiff. --- The number of CPUs can be re-adjusted while(!) force-level2 is run- ning. A file 'cpu-$TIME' is temporarily created in DIR_TEMP. This file can be modified. Note that the effect is not immediate, as the load is only adjusted after one of the running jobs (images) is finished. --- Added a custom delay for starting new jobs in force-level2. This pre- vents a huge I/O-related bottleneck at the beginning of processing. As a result, it may take a while (depending on how many cores you are us- ing) until force-level2 runs at full efficiency. An additional parame- ter needs to be given for force-level2 (delay in seconds). --- force-level2 shows eta (this will only show reasonable numbers after a couple of jobs are finsihed). The estimate is based on the runtime of finished jobs, so the first estimate will only be shown when the first job has finished. Note that the progress bar and eta are produced by GNU parallel. Afaik, there is no way to change the format; sorry for the eta in seconds... --- The processing time is written to the logfile. --- There is a slight problem, which seemingly occurs on CIFS-mounted win- dows shares. Attempting to lock the file queues will fail and will throw a warning. This is not a fatal error. Please see the FAQ in the user guide if this happens on your system. I didn't find a solution yet. --- Hopefully fixed a bug that changed file permissions of the file queue to read-only (this is related to "sed -i" and seems to occur on some Linux systems only). --- Changed behaviour of the subzero QAI flag: Subzero = any band < 0.0. If reflectance < -1.0, the pixel is set to nodata. --- Changed behaviour of the surface reflectance output when saturation is detected. Previously, reflectance was capped at 1.0. Now, values > 1.0 and < 2.0 are allowed, but still flagged as saturated. Pixels with any reflectance > 2.0 are set to nodata. --- Improved the performance of the shadow probability computation in the cloud/shadow detection module. The shadow probability is only computed for pixels that are close enough to a cloud. A maximum cloud height of 12000m + max. sun zenith of scene is assumed; sun azimuth is not taken into account so far. --- Snow mask switched on for Sentinel-2. Increased NDSI threshold to re- duce comission because of the missing thermal band, included a SWIR2 threshold to counterbalance. --- Reduced cloud/cloud shadow/snow/water ambiguity. Now, a pixel cannot be cloud and snow, or water and snow at the same time. Water takes precedence over snow. Snow takes precedence over cloud. Note that snow and cold clouds can be easily confused. No shadow is matched for snow pixels (will increase error of omission to a certain degree). Opaque clouds, and buffered clouds take precedence over cirrus clouds. Clouds take precedence over cloud shadows. --- Decreased the aggresivity of the cirrus mask by incorporating a blue test. Cirri are not buffered anymore. --- Reduced cloud shadow comission errors by incorporating the shadow pro- bability in the computation of the shadow match similarity. This will favor stonger shadows and shadows won't simply snap to all dark fea- tures (like coniferous forest). --- Snow coverage in percent is written to the logfile. --- Implemented more efficient method to interpolate coarse resolution at- mospheric parameters, + elevation-dependent parameters are computed for smaller elevation increments -> resulting atmospheric variables like AOD look much smoother --- Fixed a bug in the geometric module, which appeared when warping and tiling images. The result was a possible subpixel misalignment as the images were simply warped into the target coordinate system without aligning the new extent with the tiling grid. --- The conversion from DN to physical units is now based on the reflec- tance calibration factors for Landsat as suggested by Kurt Thome @ ACIX. As the ESUN spectrum is not used anymore, the remove-then-add calibration employed for Sentinel-2 was removed, too. --- There are three different resolution merge options available to in- crease the spatial resolution of the 20m Sentinel-2 bands to 10m. The default algorithm has changed from REGRESSION to IMPROPHE. Other op- tions are STARFM and NONE. See user guide for more details. --- The use of the DODB was deprecated. The parameters were removed from the parameter files. This feature has never made it into an official FORCE release; before the release of FORCE v. 1.0, measures were im- plemented to make the AOD estimation more reliable; therefore results were fine without using it. Given the extra processing step (non-line- arity in processing) and used disk space, I decided to completely re- move this functionality. --- Fixed a potential data race when writing one and the same output file from two threads. Now, a lockfile is created to prevent this. Parallel writes can happen as FORCE implements a Level 2 compositing to reduce data redundancy, i.e. if two different Level 1 products cover the same Level 2 tile. By doing so, the compositing method was changed too. Be- fore, it was simply overwriting with the latest data. Now, the exis- ting file is read, then the average between the old and new data is computed, and the file is written completely again. This method pro- duces more smooth transitions between Level 1 footprints. Note that the old overwrite with latest data method is still used for the QAI layer. --- It is now possible to process images without a DEM, although this is not recommended. If doing so, the DEM is assumed to be flat @ z=0m. Cloud detection and atmospheric correction both benefit from using a DEM. If topographic correction is to be used, L2PS will exit with an error. To disable, use FILE_DEM = NULL, and DO_TOPO = FALSE. --- Added wavelength domain tags to bandnames for BOA/TOA products. --- Data cover, cloud cover (opaque, buffered and cirrus), cloud shadow cover, snow cover and water cover is written to the metadata. --- force-quicklook-level2 has an additional argument, which defines the stretch (max vis value in linear stretch). Before, it was fixed to 1000 (10% reflectance), which is inappropriate in bright landscapes. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L3PS --- If the target resolution is lower than the Level 2 data, the spatial aggregation can be performed using an approximated Point Spread Func- tion. Note that this option severely affects input time. A new para- meter REDUCE_PSF was added to the parameter file. --- force-level3 is more verbose, shows progress, eta, relative time spent for input/computing/output operations (note: it is not pure I/O time, it is rather a measurement of all input/output related things, e.g. the CPU time spent with REDUCE_PSF would count as input), and composi- ting statistics at the end. The eta is based on the runtime of finis- hed tiles, so the first estimate will only be shown after the first tile has finished. --- BAP composites look mostly crappy over water because it is a very va- riable surface type. Therefore, the compositing method is switched over permanent water, i.e. if > 90% of obs are water-flagged. Over water, minimum SWIR2 compositing is used, while enforcing that NIR re- flectance is greater than SWIR2. Will only be used for sensors with NIR and SWIR2 band. --- force-level3 can use Level 2 ImproPhe data (will use original data, i.e. BOA/TOA, if there is no IMP product for any specific date), gene- rated by the new tool force-level2-improphe (see section FORCE L2IMP below). A new parameter USE_IMPROPHE was added to the parameter file. --- Added wavelength domain tags to bandnames for reflectance products. --- force-quicklook-level3 has an additional argument, which defines the stretch (max vis value in linear stretch). Before, it was fixed to 1000 (10% reflectance), which is inappropriate in bright landscapes. --- force-level3-mosaic was deprectared, and replaced by the new force- mosaic, which can mosaic all FORCE-generated, tiled outputs. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE TSA --- FORCE TSA has received a major update and many new features. There are many new parameters that you need to include in your existing parame- ter files. --- If the target resolution is lower than the Level 2 data, the spatial aggregation can be performed using an approximated Point Spread Func- tion. Note that this option severely affects input time. A new para- meter REDUCE_PSF was added to the parameter file. --- force-tsa is more verbose, shows progress, eta, relative time spent for input/computing/output operations (note: it is not pure I/O time, it is rather a measurement of all input/output related things, e.g. the CPU time spent with REDUCE_PSF would count as input). The eta is based on the runtime of finished tiles, so the first estimate will on- ly be shown after the first tile has finished. --- force-tsa can use Level 2 ImproPhe data (will use original data, i.e. BOA/TOA, if there is no IMP product for any specific date), generated by the new tool force-level2-improphe (see section FORCE L2IMP below). A new parameter USE_IMPROPHE was added to the parameter file. --- Only bands that are needed to compute the requested index are read from disc. This decreases time spent for reading data. As an example, only red and NIR are read for calculating NDVI time series. --- Fixed a bug that caused a core dump when outputting the TSS product with more than 1169 images. --- Instead of a 'simple' spectral index or band, linear spectral unmixing can be used. Use INDEX = SMA. Several parameters were added to the pa- rameter file. The endmember file is specified with FILE_ENDMEM. Only the abundance image related to one endmember is retained (this will likely change in future versions). The SMA can be Sum-to-One constrai- ned using the parameter SMA_SUM_TO_ONE. The SMA can be Non-Nega-tivity constrained using the parameter SMA_NON_NEG. The fractions can be sha- de normalized, using the parameter SMA_SHD_NORM. The shade spectrum (photogrammetric zero or measured shade) need to go into the last column of the endmember file. The RMSE time series of the SMA can be output using the parameter OUTPUT_RMS. --- Red Edge bands (RE1, RE2, RE3) and the broad NIR band (BNIR) of Senti- nel-2 can be used as INDEX. --- Implemented Disturbance Index without rescaling, i.e. no spatially tu- ned z-transformation. INDEX = TC-DI0. --- Time series can be interpolated. Several parameters were added to the parameter file. The interpolation method is set with INTERPOLATE. Cur- rently implemented are NONE (no interpolation), LINEAR (linear inter- polation), MOVING (running mean interpolation / smoothing), and RBF (Radial Basis Function interpolation / smoothing). The maximum tempo- ral distance (i.e. filter width) for the MOVING filter, is set with MOVING_MAX (in days). Several kernels can be used for the RBF filters using RBF_SIGMA, the sigma(s) are given in days. The kernel strengths are adapted by weighting with actual data availability within the time covered by each kernel. The parameter RBF_CUTOFF specifies a cutoff value for temporally truncating the kernels, such that e.g. 95% of the Gaussian bell is retained. The time step for the interpolation is set with INT_DAY (in days). The interpolated time series can be output with OUTPUT_TSI. --- Land Surface Phenology (LSP) can be derived from dense timeseries. The Spline Analysis of Time Series (SPLITS) API was incorporated for this purpose. FORCE can be compiled with and without SPLITS, in the latter case, the new functionality won't be available. See the user guide for instructions on how to install with SPLITS. A number of new parameters were added to the parameter file, and new products can be output. LSP extraction is triggered by using LSP as folding option (FOLD = LSP), or when OUTPUT_LSP = TRUE. The number of segments for fitting the splines needs to be specified using the LSP_N_SEGMENT parameter. To derive LSP metrics for a given year, some data from the previous and next year need to be included (LSP_DOY_PREV_YEAR, LSP_DOY_NEXT_YEAR). The user can select if a Northern hemisphere, Southern hemisphere or mixed phenology is expected (LSP_HEMISPHERE). In total, 26 LSP metrics will be derived for each year. --- The change and trend (TRD/CAT) products can be computed on the annual LSP metrics. This will produce 26 change / trend products. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE CSO --- New module FORCE Clear Sky Observations (FORCE CSO) is intended for data availability mining. For given time steps (e.g. 3 months), per- pixel statistics about data availability are calculated, i.e. number of CSOs, and average (standard deviation, min, max, etc.) days between consecutive CSOs. FORCE CSO includes the programs force-parameter-cso to generate an empty parameter file, and force-cso to do the data mi- ning. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE ImproPhe --- New module FORCE Improving the Spatial Resolution of Land Surface Phe- nology (FORCE ImproPhe) is intended to to increase the spatial resolu- tion of coarse continuous fields. It was originally developed to re- fine Land Surface Phenology metrics derived from MODIS, using sparse Landsat data as spectral and multi-temporal targets for data fusion. Regarding phenology, it can be used to obtain a Landsat-like phenology even in areas / during times when Landsat data alone is insufficient (in terms of temporal density). FORCE permits the use of Landsat and/ or Sentinel-2 data as target datasets for the improPhement. ImproPhe can also be applied to other coarse resolution data. FORCE ImproPhe includes the programs force-parameter-improphe to generate an empty parameter file, and force-improphe to perform the data fusion. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L2IMP --- New module FORCE Level 2 ImproPhe (FORCE L2IMP) is intended to increa- se the spatial resolution of lower resolution Level 2 ARD using higher resolution Level 2 ARD, e.g. to improve the spatial resolution of 30m Landsat imagery to 10m using Sentinel-2 data as targets. This only works for years where both data sources exist. The data fusion is per- formed with the ImproPhe algorithm. Note that this module is heavy on processing time. FORCE L2IMP includes the programs force-parameter- l2imp to generate an empty parameter file, and force-l2imp to perform the data fusion. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE WVDB --- Collection 6 data were removed from LAADS' servers, leaving force-lut- modis nonfunctional. The code was updated to make use of collection 6.1. --- Due to new NASA policies, LAADS' FTP was shut down, leaving force-lut- modis nonfunctional. The code was updated to make use of the HTTPS server. --- The water vapor climatology has a new column: standard deviation of monthly water vapor. This information is just for estimating the va- riability for each coordinate/month, and so for the usability of the climatology. This information is not used by FORCE L2PS, thus the old tables can still be used. --- We compiled a ready-to-use, global water vapor database. The dataset is comprised of daily global water vapor data for February 2000 to Ju- ly 2018 for each land-intersecting Worldwide Reference System 2 (WRS- 2) scene, as well as a monthly climatology that can be used if no dai- ly value is available. The dataset is freely available at https://doi.pangaea.de/10.1594/PANGAEA.893109 under the terms of the CC BY 3.0 license. This dataset may relieve you of the burden to gene- rate the water vapor database on your own. --- Fixed a bug in force-lut-modis in finding intersecting MODIS granules. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE AUX --- Added new program 'force-mosaic', which generates virtual mosaics from tiled products in Virtual Dataset format. This tool can be used with any tiled FORCE output, i.e. Level 2 / Level 3 / TSA / CSO / ImproPhe. force-mosaic mosaicks everything with the same basename. It Will like- ly fail for products that have different number of bands in different tiles; this especially applies to the TSS product of force-tsa. The previous, per-module mosaic tools are deprecated. --- There is a new mandatory parameter in force-qai-inflate, which speci- fies the output format, i.e. ENVI or GTiff. --- The usage of force-tile-finder has slightly changed. Before, the Level 2 parameter file needed to be given as input (containing the defini- tion of projection and grid). Now, each data cube is accompanied by a spatial data cube definition, Therefore, only the directory of any gridded data cube needs to be given (containing a data cube definition file). An additional parameter ‘resolution’ was added to relate geo- graphic coordinates to pixel positions, denoting that products with multiple resolutions may be present in a data cube. --- The usage of force-tabulate-grid has slightly changed. Before, the Level 2 parameter file needed to be given as input (containing the de- finition of projection and grid). Now, each data cube is accompanied by a spatial data cube definition, Therefore, only the directory of any gridded data cube needs to be given (containing a data cube defi- nition file). ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ########################################################################## ### FORCE v.1.1 beta ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L1AS --- Download and management support for Sentinel-2B, higher level support was already implemented. --- The Sentinel-2 downloader/archiver has 4 new mandatory arguments/fil- ters: sensing-start-time sensing-end-time min-cloud-cover max-cloud- cover. --- Fixed a bug when importing Landsat images with white-space characters in the directory name; now, this thows an error. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L2PS --- Reduced memory requirements for holding TOA reflectance internally by ~0.72GB (Landsat 4-7) ~0.95GB (Landsat 8) ~2.92GB (Sentinel-2) --- Reduced memory requirements when correcting for adjacency effects by ~1.20GB (Landsat 4-8) ~2.24GB (Sentinel-2) --- Fixed a bug in the BRDF correction of the Landsat 8 SWIR-2 band. --- BRDF correction is now available for Sentinel-2. --- Relative Spectral Response Function for Sentinel-2B is now included. --- Relative Spectral Response Function of Landsat 4 TM was updated. --- Improved gasweous vapor transmittance calculations. The water vapor absorption spectrum was updated to the HITRAN 2016 database (obtained from HITRAN on the Web, http://hitran.iao.ru/) The computation of wa- ter vapor absorption and water vapor and ozone transmittance was cor- rected: a numerical error in calculating transmittance resulted in in- creased SWIR-1 reflectance in Landsat-TM data. --- The spectral resolution of all tables was reduced from 0.1 nm to 1 nm. --- Added bandnames to all output products. --- Cloud shadow matching for Sentinel-2 images is now performed on the 20 m resolution, which significantly speeds up the work, especially for poorly illuminated winter images where the base height is iterated in small vertical increments ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE L3PS --- Added new parameter OFF_SEASON, which allows the compositing algorithm to select clear-sky observations that are not within the prime season of interest. This may be advantageous if gap-free images are needed. This parameter also affects the temporal-spectral metrics. If enabled, the statistics will be derived over the complete year, instead of the prime season of interest. --- Added new parameters OUTPUT_Q25, OUTPUT_Q50, OUTPUT_Q75 and OUTPUT_IQR that allow the computation of 25%/50%/75% quantiles and interquantile range, defined as Q75 - Q25. --- Added new parameter LSP_START, which allows you to input LSP data with a custom starting point. Previously, the LSP data had to be provided as YEAR*365+DOY. The new parameter specifies an offset as Y*365+DOY. Thus, if the LSP are provided as before, LSP_START = 1. If the LSP da- ta would be provided relative to January 1 2000, LSP_START = 730001, i.e. 2000*365+1. --- Fixed a bug, which resulted in no data when using the phenology-adap- tive compositing with only one year of LSP data. --- Removed unused parameter LSP_NUM_YEAR. --- Added new program 'force-level3-mosaic', which generates virtual mosa- ics from tiled Level 3 products in Virtual Dataset format. Due to per- formance considerations, it is adviced to enable tiled output in force -level3, and if necessary, use this new tool to create mosaics. force- -quicklook-level3 was updated to generate quicklooks from the vrt mos- aics. --- Added bandnames to all output products. --- Negative tile numbers are now allowed. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE TSA --- Added product id as metadata for time series stack output. --- Parallel processing can be used. New parameter was added to parameter file. Computation of TC-DI is currently not parallelized. --- Added bandnames to all output products. --- Negative tile numbers are now allowed. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE WVDB --- Collection 5 geometa tables were removed from LAADS' FTP server, lea- ving force-lut-modis nonfunctional. The code was updated to make use of collection 6. --- Discovered a randomly occuring bug, seemingly related to a data race in localtime(): the function timer was set to tomorrow, resulting in a never-ending while loop. Measures were implemented to avoid this data race. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ FORCE AUX --- Added a new program, 'force', which prints the version number and a short disclaimer. --- Fixed a bug of missing resolution in force-qai-inflate outputs. --- force-tabulate-grid was completely overworked. Instead of a table with coordinates, a shapefile is directly generated, which let's you easily relate to the processing grid. --- Added bandnames to all output products (force-qai-inflate). ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ########################################################################## ### FORCE v.1.0 beta ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++ First public release ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++