SPDSS_Task65_EstimateSouthPlattePhreatophyteGroundwaterEvapotranspiration

Home Browse Search

Table 3. Dark object subtraction values for correcting Bands 3, 4 and 5 for each image. Scene Band 3 Band 4 Band 5 32/32 0.020 0.018 0.000 33/32 0.026 0.019 0.004 33/33 0.018 0.016 0.000 4.3 Calculation of NDVI* from TM Data NDVI was calculated from the DOS-corrected Band 3 and 4 data according to Equation 1 using the data in Table 3. Analysis of the cumulative distribution functions using linear regression for both NDVIo and NDVIS resulted in the values in Table 4. Equation 2 used these values for calculation of NDVI*. Table 4. Parameters that stretch the NDVI distribution from 0-1 (NDVIo and NDVIS) developed for the three TM Images. Scene NDVIo NDVIS 32/32 0.249 0.954 33/32 0.263 0.947 33/33 0.223 0.930 Overlap between images provided the opportunity to cross check the calculated values of NDVI* to assure continuity across the entire study area, despite using multiple images. Ideally, pixel values should be identical for the same point within the two datasets, but slight differences in geocorrection, DOS correction values, and also the nine-day temporal difference between scene 32/32 and scene 33/32, introduced the potential for data discrepancy between adjoining datasets. Overlapping pixels were examined for verification that the values of NDVI* were comparable. Adj acent images were displayed in ArcMap, and using the Identify tool, pixel values were compared between the two datasets where overlap was present. Pixel values for some irrigated cultivated fields outside of the riparian polygon were significantly lower for the later dataset due to harvest (probably alfalfa, but not confirmed), however, phreatophyte vegetation showed little change between adjacent datasets (Figure 8). Thus, the processing values (DOS, NDVIo and NDVIS) were accepted as correct because they yielded comparable values of NDVI* for the riparian vegetation of interest across the three images. 13