QUICK GUIDE TO THE USE OF AHS AND CASI IMAGES I-General considerations Note: most of this information is found in the AHS and CASI product specification documents that are located in each sensor parent directory. AHS images are found in the [REFLEX/AHS_12072*/...] folders CASI images are found in the [REFLEX/CAS_12072*/...] folders At-sensor radiance images are in the [calibrated_data] folders. For each at-sensor image, a small subset of orthorectified bands (3 bands for AHS or 4 bands for CASI) are found in the [geolocated_QL] folders. The [geolocation_info] folders contain the IGM (AHS) or GLU (CASI) files that are needed to orthorectify the complete set of bands from L1b or L2b products. For this purpose you should use ENVI>geolocate from IGM or an equivalent tool. The [Level2] folders contain the surface magnitudes, i.e. reflectance (or HDRF), temperature and emissivity. The [metadata] folders contain metadata, either in text files or in image quality layers. Image and auxiliary filenames follow a specific pattern that reports: SENSOR_DATE_TIME_LINE_LEVEL_otherinfo. Time and FlightLineId fields (for example 0902Z_P01ID) are used to identify a specific image. All image products are distributed as ENVI-like files. Look for the .bsq extension to open them. II-Methods Calibrated data has been produced with nominal calibration factors. Reflectance, Temperature and emissivity have been computed with ATCOR4 using the following configuration water vapour: estimated from the image (more details in ATCOR4 manual) aerosol type: maritime visibility: 40 km estimated from the analysis of two field spectra (bare soil and black fabric) Warning! These values are selected as those that produce the best fit with the field spectra, but shall not be considered as estimation of the actual atmospheric conditions!. Geolocation has been performed with the nominal boresight correction. Image inspection shows that the obtained accuracy is slightly above 1 pixel. III-Specific issues AHS File surfing Images from the night flight are included in the folder AHS_120725N The VNIR-to-SWIR bands and the MIR-to-TIR bands are found in separate files, namely: VNIR-to-SWIR: *PT12 (not delivered for the night imagery) MIR-to-TIR: PT34 for at-sensor radiance, PT4T for temperature, PT4E for emissivity. Radiometric considerations Bands 22, 23 and 64 are at the edge of atmospheric windows - they have very low SNR and are seldom useful. Bands 44 and 46 have a very low SNR due to detector malfunctions. They should be used with reservations. Bands 60 to 63 are not calibrated because they are outside the Labsphere useful range. File AHS P11BD on July 26th shows a recording anomaly, namely a set of corrupted image rows (rows 4587 to 4600) followed by 20 rows with duplicated info and finally a small data gap. Temperature is computed from AHS band 75 assuming emissivity=0.98 Emissivity is computed inverting for each band the Planck Law from the temperature estimated above. Geometric considerations Pixel size for geolocated quicklook is set to 3 meters for low ("B") images and 5 meters for high ("I") images. Actual ground IFOV and PSF are complex. Specifically, AHS GIFOV increases significantly to the image edges. Please note that the user might select a different grid size when generating the georeferenced cube. CASI File surfing CASI images acquired binning spectral bands by 2 are labelled E2. CASI images acquired with full spectral resolution are labelled SF. Spectral considerations. All CASI images suffer a slight spectral shift under flight conditions. A resampling of the input radiance from the estimated band centers to the reported band centers is perfoemd before calibration. This procedure has a limited accuracy and might affect the radiometric accuracy in absorptions regions. Actual FWHM has been analysed at INTA. A value around 3.1 microns is our best estimation. Radiometric considerations Column #1441 in the E2 images is not valid. SNR for each band is reported in the stats.csv file. Please check this file before data analysis. As a general rule, bands <425 nm and >1000 nm have a low SNR. Some samples from very low reflectance targets display a radiance < 0 in the water absorption region around 935 nm and beyond 1020 nm. This is due to the combination of low signal, noise and an excess straylight correction, which unfortunately couldnīt be optimized. Geometric considerations. Resampling rectangular pixels to a square grid requires trade-offs. To get a compromise between lost pixels (pixels which are acquired but are not used in the new grid) and gaps (positions in the output grid that do not have an input pixel) we have selected 1.25 m pixels for E2 images and 1.5 for SF images. Please note that the user might select a different grid size when generating the georeferenced cube.