Overview of ACOLITE settings QV 2014
Transcription
Overview of ACOLITE settings QV 2014
Overview of ACOLITE settings QV 2014-2015 1. Main menu (1) (2) (3) (4) (5) (6) (8) (9) (7) (10) (11) Figure 1. Main menu of ACOLITE (version 20150408.0) (1) Input file Select the input Landsat-8 tile directory or MTL file. (2) Output directory (optional) Select the directory to store ACOLITE output files. Default is the image input directory. (3) Region crop (optional) Crop to provided region boundaries, in decimal degrees. Coordinates west of the Greenwich meridian and south of the equator should be prefixed with a minus sign (-). A test is performed whether provided coordinates make sense (i.e. S<N and W<E), and the Landsat tile is tested if it covers at least part of the region of interest. It is recommended to process a cropped region on computers with low memory or running Windows XP. (4) RGB processing Check these boxes to output a PNG RGB colour composite of the Landsat-8 tile. RGB composites are made using bands 4, 3, and 2 (655, 561 and 483 nm), at top of atmosphere and/or after Rayleigh correction. (5) L2 processing These options allows you to choose the output of the L2 processing, with the atmospheric correction described in (Vanhellemont and Ruddick, 2015, 2014a). Output options are: output parameters to NetCDF or GeoTIFF files and PNG maps. Datasets can be written to a single NetCDF file, a NetCDF file per parameter or per parameter group. Writing separate files is recommended for full scene processing with a large number of output parameters. (6) L2 products These are the basic output products of the atmospheric correction algorithm: marine reflectances in the first 5 bands (443, 483, 561, 655 and 865 nm) and the multiple-scattering aerosol reflectance in the NIR band (865 nm). Other L2 output parameters can be specified in the Advanced settings menu. (7) Save and restore settings These buttons allow you to save and restore processing settings. When launched, the application automatically tries to restore settings from a file named acolite_settings in the same directory. (8) Advanced settings Brings up the advanced settings menu, see below. (9) About... Shows brief info about the ACOLITE processor. (10) Run Run the ACOLITE processing with the selected options. (11) Exit Exit the ACOLITE application. 2. Advanced settings (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Figure 2. Advanced options toolbox of ACOLITE (version 20180408.0) (1) Options for Rayleigh correction Allows you to change the atmospheric pressure if it is significantly different from the default (=1013.25 hPa), recommended for elevated inland waters, and when the pressure is known. The Rayleigh reflectance is scaled according to the given atmospheric pressure in hPa, scale factor, or elevation in m (using a standard atmosphere). (2) Use DEM for pressure correction Similar to the above, but a 3 arcsecond SRTM digital elevation model (DEM) is used to determine elevation and pressure for a standard atmosphere. SRTM data is available globally between 56°S and 60°N. No DEM 1° tiles are distributed with ACOLITE and they need to be acquired in .hgt format by the user. Some example websites that provide the data: http://e4ftl01.cr.usgs.gov/SRTM/SRTMGL3.003/2000.02.11/ http://dx.doi.org/10.5067/MEaSUREs/SRTM/SRTMGL3.003 https://lpdaac.usgs.gov/products/measures_products_table/srtmgl3 ACOLITE will determine and list which tiles are needed. Tiles need to be downloaded, unzipped (or .gz compressed – not .zip) and placed in the ACOLITE/Input/SRTMGL3.003/ directory. Note that tiles with only sea - thus without land - do not exist in the SRTM tileset. ACOLITE will list those non-existing tiles in its estimate. They are not present on the data server and are not needed for correct processing: Missing tiles are set to 0m – sea level. (3) Reflectance scaling for RGB These options allow you to change the stretching of the RGB composites. Reflectances in the three bands (R=655nm, G=561nm, B=483nm) are scaled linearly between 0 and 0.15 (or entered value) to the three 8-bit (0-255) channels used in the RGB map. (4) Geolocation on maps This option adds longitude and latitude labels to the output PNG maps, derived from the UTM projection in the GeoTIFF files. (5) Pan-sharpen the RGB maps This option uses band 8 to pan-sharpen the RGB images to 15m resolution. Details on the simple method that is used can be found in (Vanhellemont and Ruddick, 2014b). Not recommended for full-tile processing as this will create a very large (~250MB) PNG file! (6) Parameter colour scale These boxes allow the user to specify output ranges for the PNG maps. Note that all parameters will be linearly scaled to this range. Various options (7) Use SWIR or NIR atmospheric correction You can choose between using the default SWIR atmospheric correction that uses bands 6 and 7, 1609 and 2201 nm, for the aerosol correction (Vanhellemont and Ruddick, 2015) or the NIR atmospheric correction that uses bands 4 and 5, 655 and 865 nm (Vanhellemont and Ruddick, 2014a) The default is recommended in moderately to extremely turbid waters (turbidities > ~30 FNU). (8) Aerosol epsilon options ACOLITE determines a per-tile epsilon or aerosol type as the ratio between the Rayleigh corrected reflectances in the aerosol correction bands, for pixels where the marine reflectance can be assumed negligible; i.e. in all pixels for the SWIR atmospheric correction, and in clear waters for the NIR correction. These buttons allow you to choose aerosol correction options: - Full tile fixed epsilon - Per pixel variable epsilon (SWIR correction only) - Sub-scene fixed epsilon (recommended if processing sub-scene) This option computes the aerosol epsilon for the selected crop. There can be significant variability of aerosol types over the full Landsat-8 tile. E.g. for inland waters, the per-tile epsilon can be significantly different from the sub-scene epsilon if a large number of water/sea pixels are present elsewhere in the image. - User defined epsilon See (Vanhellemont and Ruddick, 2015) and their Supplementary Data #1 for details. (9) USGS TOA reflectances USGS provides L1 GeoTIFF data with the options to convert the digital numbers to radiances or reflectances. The default in ACOLITE is to use the top of atmosphere radiances which are then converted to top of atmosphere reflectances within ACOLITE using sun-earth distance, extraterrestrial irradiance from (Thuillier et al., 2003) and the cosine of the sun zenith angle. This option allows you to directly use the USGS TOA reflectances. Various options (10) Cloud masking options These options allow you to specify a different band and threshold for cloud masking. By default, cloud masking is performed using the Rayleigh corrected reflectances at 1609 nm (B6) with a threshold of 2.15%. See (Vanhellemont and Ruddick, 2015) for details. The Quality Assessment Band (BQA) is given as option, but this is currently not recommended. When using the BQA for masking, pixels are masked when cloud confidence is medium or high (bit 15 set to 1) and when water confidence low or not defined (bit 5 set to 0), more info on the BQA see http://landsat.usgs.gov/L8QualityAssessmentBand.php/ This option is currently not recommended over the simple band threshold method. o High water confidence levels are not given. o Green fields often have a medium water confidence level, and similar to the threshold method, cloud shadows over land have medium water confidence levels. o The algorithm returns “not defined” over turbid waters (bit 4 and 5 set to 0) that often have a high snow/ice confidence (bit 10 and 11 set). (11) Gains This option allows you to set the gains for bands 1-5, 6 and 7. Results from vicarious calibration efforts by (Pahlevan et al., 2014) and (Franz et al., 2014) are provided as presets. The Pahlevan L and Franz Ave gains should be used when the USGS radiances are used (default). The Pahlevan R gains are specifically for the USGS reflectances (see 2.6 above). Note that these vicarious system calibration gains were computed using other processing chains and might not be appropriate for ACOLITE processing. (12) These three tabs allow you to select extra L2 output parameters: 1) rtoa: Top-of-Atmosphere reflectances in bands 1-7 (443, 483, 561, 655, 865, 1609, 2201 nm). 2) rrc: Rayleigh-corrected reflectances in bands 1-7 (443, 483, 561, 655, 865, 1609, 2201 nm). 3) Other parameters: Allows you to specify a comma separated list of ouput parameters that are currently hard-coded in the ACOLITE processor. A partial list is given below in alphabetical order: BT_B10, BT_B11: At-sensor brightness temperature in the TIRS bands 10 and 11. Products are not atmospherically corrected, and there are calibration issues with band 11. Use with caution. CHL_OC2, CHL_OC3: blue-green ratio chlorophyll algorithm using bands 483-561 (OC2) and bands 443-483-561 (OC3). Algorithm coefficients provided by the OBPG (http://oceancolor.gsfc.nasa.gov/). This type of algorithm should be used with caution in turbid waters, and the performance of the OLI chlorophyll product has not been evaluated. Use with caution. DEM: SRTM digital elevation model for the given tile/region FAI: Floating algae index from (Hu, 2009) using Rayleigh corrected reflectances at 655, 865 and 1609 nm. The index is the difference between the NIR reflectance and a baseline between the red and SWIR. LTOA_XXX: top of atmosphere radiance at different OLI wavelengths (443, 483, 561, 655, 865, 1609, 2201 nm) RTOA_XXX: top of atmosphere reflectance at different OLI wavelengths (443, 483, 561, 655, 865, 1609, 2201 nm) NDVI, NDVI_TOA: Normalised Difference Vegetation Index: using reflectances at TOA (NIR-RED)/(NIR+RED). NDVI and NDVI_TOA are the same product. NDVI_RCO: Normalised Difference Vegetation Index: using Rayleigh corrected reflectances (NIR-RED)/(NIR+RED) RHOAM_XXX: multiple-scattering aerosol reflectance extrapolated to different OLI wavelengths (443, 483, 561, 655, 865, 1609, 2201 nm) RHOW_XXX: marine reflectance at different OLI wavelengths (443, 483, 561, 655, 865 nm) LRC_XXX: Rayleigh corrected radiance at different OLI wavelengths (443, 483, 561, 655, 865, 1609, 2201 nm) RRC_XXX: Rayleigh corrected reflectance at different OLI wavelengths (443, 483, 561, 655, 865, 1609, 2201 nm) SPM_NECHAD_561: suspended sediment concentration from (Nechad et al., 2010) using the 560 nm setting from their paper with the OLI 561 nm band. SPM_NECHAD_655: suspended sediment concentration from (Nechad et al., 2010) using the 655 nm setting from their paper with the OLI 655 nm band. T_DOGLIOTTI: blended turbidity from (Dogliotti et al., 2015) using the 645/859 nm setting from their paper with the OLI 655/865 nm band. “RED” turbidity is used for ρw 655 < 0.05, “NIR” turbidity for ρw 655 > 0.07, with linear blending for ρw 655 between 0.05 and 0.07. T_DOGLIOTTI_RED: red-band turbidity from (Dogliotti et al., 2015) using the 645 nm setting from their paper with the OLI 655 nm band. T_DOGLIOTTI_NIR: NIR-band turbidity from (Dogliotti et al., 2015) using the 859 nm setting from their paper with the OLI 865 nm band. T_GARABA_645_LIN: turbidity from (Garaba et al., 2014) using the linear 645 nm model from their paper with the OLI 655 nm band. T_NECHAD_645: turbidity from (Nechad et al., 2009) using the 645 nm setting from their paper with the OLI 655 nm band. 3. Common error messages 3.1 IDL error messages ERROR -152 Unable to allocate memory: to make array. ACOLITE uses a lot of memory to process Landsat-8 scenes. The bands for the aerosol correction are already downsampled when computing a per-scene aerosol type. However in some cases, memory management in Windows XP/IDL 6.4 is insufficient, even on machines with a lot of RAM. Try processing for a smaller sub region e.g. ~0.5 by 0.5°. NCDF_CONTROL: Attempt to take the file out of define mode (ENDEF) failed. (NC_ERROR=-62) This error occurs when the size of the output NetCDF file becomes too large (>~2GB). This happens when processing full tiles with a large number of output products. An option is to export each dataset to a separate NetCDF file (option added in ACOLITE 20150511.0) 3.2 ACOLITE error messages Given region limits fail sanity check. The provided region limits are wrong: the bounding coordinates for South > North or West > East. Region not in image. The provided region limits are fully outside the given Landsat tile. Not a valid Landsat-8 tile directory or MTL file. The provided input path is not a Landsat tile directory or MTL file. Please point to a Landsat scene. Required DEM tiles not found. The listed DEM tiles are not found in the Input/DEM/SRTMGL3.003 directory. Please download and extract the tiles in that directory. Only needed if using the DEM for scaling the Rayleigh reflectance. (See 2.2 above.) Others? Please send your error message and processing settings to: quinten.vanhellemont@naturalsciences.be 4. ACOLITE version history (since v20140930.0) v20150508.0 - added option to output different individual parameters (e.g. rhow_655, rhow_865) or parameter groups (e.g. rhow, rtoa, l2par) to separate NetCDF files - added option to output parameters to GeoTIFF files rather than NetCDF files (both full and cropped scenes) - added option for user to specify a fixed aerosol epsilon value (between B4 and B5 for NIR, B6 and B7 for SWIR correction) - added per pixel variable epsilon option for the SWIR correction - added option to choose cloud masking channel (865, 1609, 2201 nm) and threshold, defaults are 1609 nm and 0.0215 - added option to use the Landsat Quality Assessment Band (BQA) masking information With this option, pixels are masked when cloud confidence is medium or high (bit 15 set to 1) and when water confidence low or not defined (bit 5 set to 0), more info on the BQA see http://landsat.usgs.gov/L8QualityAssessmentBand.php/ This option is currently not recommended over the simple band threshold method. o High water confidence levels are not given. o Green fields often have a medium water confidence level, and similar to the threshold method, cloud shadows over land have medium water confidence levels. o The algorithm returns “not defined” over turbid waters (bit 4 and 5 set to 0) that often have a high snow/ice confidence (bit 10 and 11 set). - fixed bug that still occurred in some cases when given limits are outside the image - fixed bug for bulk processing where one image was not over the given limits - fixed bug when loading old settings file: new settings not yet in that settings file are now set to defaults - for bulk processing the image list can now be specified in a text file, for example: idl -rt=acolite.sav -args image=imagelist.txt settings=acolite_settings_file - trailing slash is now added to the output path if needed - save and restore buttons now remember last used directory - fixed full tile pan-sharpening error - added support for geolocation on full tile maps - added Dogliotti et al. (2015) blended turbidity algorithm v20150306.0 - changed cloud mask limit to Wang and Shi (2006): 0.0215 for the 1609 nm channel - added Rayleigh corrected radiances as output - added BQA “quality assessment” band as output parameter - added range for parameter colour scale of maps - SPM products are scaled linearly depending on P90 (0-5 / 0-10 / 0-15 / 0-50 / 0-100 gm-3) - Added support for running a batch of scenes through the command line: An ACOLITE settings file needs to be generated in the GUI, and the images need to be given as a comma separated string. The examples below use the included IDL runtime, but it is also possible to use your own IDL installation. Linux: open a terminal cd /path/to/acolite/ idl82/bin/idl -rt=acolite.sav -args image=/path/to/image1,/path/to/image2 settings=acolite_settings_file Windows Select Start > Run… enter cmd and press return cd /path/to/acolite/ bin/bin.x86/idlrt.exe acolite.sav –args image=/path/to/image1,/path/to/image2 settings=acolite_settings_file Note that the Windows version does not output any messages to the command line. v20141210.0 - added PAN sharpened RGB option in advanced settings - fixed wrong 1 pixel offset for certain crops and fixed size of PAN band when cropped - added text box allowing export of extra L2 products - added support for SRTM 3 arc second DEM for Rayleigh correction scaling - added check for crop fully outside scene - turbidity products are scaled linearly depending on P90: from 0 to 15 or from 0 to 40 v20141119.2 - fixed bug where cropped dimensions differ from lat/lon arrays by 1 pixel v20141119.1 - fixed "crop_pos" bug when one end of the crop in latitude is outside the scene extent v20141119.0 - added elevation as option to calculate air pressure / Rayleigh scaling for inland waters (standard homogeneous atmosphere with T=288.15K) - added NetCDF output of TOA reflectances for all bands and Rayleigh corrected reflectances for the SWIR bands - option to use USGS/LGPS provided reflectances, rather than calculate reflectances from USGS/LGPS radiances (the latter is still the default) - support for setting gains on B1-7 (with presets for vicarious calibration results from Nima and Bryan) - fixed an important bug in finding epsilon when resampling SWIR bands (for memory efficiency) - added the option to get a sub-scene epsilon - which can be quite different for inland waters when there are sea pixels in the scene v20141107.0 (version distributed at Ocean Optics 2014, Portland ME) - more efficient (accurate?) calculation of lon/lat from GeoTIFF tags and better crop positioning - crop when rather than after reading GeoTIFF files - mapping now uses UTM coordinates and no additional reprojection to equirectangular projection (should fix REBIN issue) v20140930.0 (internal version) v20140722.0 (version distributed at IOCCG Summer School 2014, Villefranche) References Dogliotti, A., Ruddick, K., Nechad, B., Doxaran, D., Knaeps, E., 2015. A single algorithm to retrieve turbidity from remotely-sensed data in all coastal and estuarine waters. Remote Sens. Environ. 156, 157–168. Franz, B.A., Bailey, S.W., Kuring, N., Werdell, P.J., 2014. Ocean Color Measurements from Landsat-8 OLI using SeaDAS, in: Proc. Ocean Optics 2014, Portland Maine (USA), 26-31 October 2014. Garaba, S., Badewien, T., Braun, A., Schulz, A.-C., Zielinski, O., 2014. Using ocean colour remote sensing products to estimate turbidity at the Wadden Sea time series station Spiekeroog. J. Eur. Opt. Soc.-Rapid Publ. 9. Hu, C., 2009. A novel ocean color index to detect floating algae in the global oceans. Remote Sens. Environ. 113, 2118–2129. Nechad, B., Ruddick, K., Neukermans, G., 2009. Calibration and validation of a generic multisensor algorithm for mapping of turbidity in coastal waters, in: SPIE Europe Remote Sensing. p. 74730H–74730H. Nechad, B., Ruddick, K., Park, Y., 2010. Calibration and validation of a generic multisensor algorithm for mapping of total suspended matter in turbid waters. Remote Sens. Environ. 114, 854–866. Pahlevan, N., Lee, Z., Wei, J., Schaaf, C.B., Schott, J.R., Berk, A., 2014. On-orbit radiometric characterization of OLI (Landsat-8) for applications in aquatic remote sensing. Remote Sens. Environ. 154, 272–284. Thuillier, G., Hersé, M., Foujols, T., Peetermans, W., Gillotay, D., Simon, P., Mandel, H., others, 2003. The solar spectral irradiance from 200 to 2400 nm as measured by the SOLSPEC spectrometer from the ATLAS and EURECA missions. Sol. Phys. 214, 1–22. Vanhellemont, Q., Ruddick, K., 2015. Advantages of high quality SWIR bands for ocean colour processing: examples from Landsat-8. Remote Sens. Environ. 161, 89–106. doi:10.1016/j.rse.2015.02.007 Vanhellemont, Q., Ruddick, K., 2014a. Turbid wakes associated with offshore wind turbines observed with Landsat 8. Remote Sens. Environ. 145, 105–115. doi:10.1016/j.rse.2014.01.009 Vanhellemont, Q., Ruddick, K., 2014b. Landsat-8 as a Precursor to Sentinel-2: Observations of Human Impacts in Coastal Waters., in: ESA Special Publication SP-726. Presented at the 2014 European Space Agency Sentinel-2 for Science Workshop, Frascati.