Helicopter Applicators Catskills Small Pass Data Processing January 4, 2005 The original reflectance files from 4 passes are of Phoenicia, NY, Catskills, flown on September 3, 2004 and was obtained from Helicopter Applicators, Inc. Radiance data was converted to reflectance by Helicopter Applicators, Inc.; methods on this conversion are unknown. Wavelength values for each band are also unknown. The files were modified in the following ways: 1. Individual Runs (Passes) were mosaicked together in ENVI using original rectification obtained from CaliGeo software by Helicopter Applicators, Inc. This original rectification was between 10 and 30 m off of the 1m resolution DOQ. 2. The mosaiced cube was further warped against the 1m resolution DOQ of Phoenicia, NY using Imagine 8.7 on /net/nfs/spectrum. Warping details: a. Polynomial degree 2 warp in Imagine b. Including 41 gcp points primarily concentrated across the northern border of the image and the south eastern corner. Points within the image are sparse. GCP points can be found at /net/nfs/gromit/d1/catskills_hai/Small/SP_refl_mosaic_gcp_in.gcc /net/nfs/gromit/d1/catskills_hai/Small/SP_refl_mosaic_gcp_ref.gcc c. These points resulted in a RMSE of 5.14. 3. This georeferenced reflectance file was then exported as .bil for easier use in all versions of ENVI. Final File Details: Location: /net/nfs/gromit/d1/catskills_hai/Small/SP_refl_mosaic_geo_bil Dims: 3047 x 2254 x 130 [BIL] Size: [Unsigned Int] 1,785,663,880 bytes. File Type : ENVI Standard Sensor Type: Unknown Byte Order : Host (Intel) Projection : UTM, Zone 18 North Pixel : 2 Meters Datum : North America 1983 Wavelength : None Upper Left Corner: 1,1 4. A checksum was run on the .bil file. File fingerprint = 466e083bcbd8c3dace9061b2d4bd913d Saved as: net/nfs/gromit:/d1/catskills_hai /Small/SP_refl_mosaic_geo_bil_cksm 5. The file was then compressed using gzip for faster file transfer and can be found at: net/nfs/gromit:/d1/catskills_hai /Small/SP_refl_mosaic_geo_bil.gz Note: We will use these same gcp points to warp any prediction images produced by Jim Sokolowski for accuracy assessment. Determination of pixels for data Extraction 1. Trimble measured plot center UTM coordinates (zone 18N, WGS84) were used to calculate tree level coordinates based on distance and magnetic bearing to each tree from plot center using scripts written by Mary Martin. Compared to the 1m resolution DOQ these appear to be accurate within 2m of visible hemlock canopies. 2. Original ground measured tree coordinates were overlain on the georeferenced .img file and viewed using multiple band combinations (58,37,15 and 90,60,20) and geographically linked to the 1m resolution DOQ. 3. New AOI’s were created (1 pixel for each tree) based on the apparent offset to the DOQ, pixel appearance in the .img file, geometry of all trees on the plot, and known species and health characteristics for those trees. This was manual and hence, highly subjective. Note, however, that if pixels were selected off of the original coordinates (say for example the correct pixel appears to be 2m northeast of the original point location) pixels would be selected in the same way for ALL trees on that plot (The assumption is that if the registration/warping is off by 2m to the northeast, it should be off in the same way for ALL trees in that location). New coordinates for each tree are thus created from the AOI’s. 4. Spectra was extracted from these new AOI’s for input to calibration equations and will be used to extract predictions for accuracy assessment.