Summary of First Spatial Analysis at Fort Benning. Optimizing Fixed Sample Designand Data Compression from Remote Imagery

摘要

The digital SPOT image data from the study area chosen by TEC within Fort Benningcomprise three wavebands, near infrared, red and green. The standard vegetation index, NDVI, was computed from the first two channels. The pixel data for the four variables were mapped, and the summary statistics were computed. Wavebands 2 and 3 were strongly skewed and so the values were transformed to logarithms. Variograms were computed in two directions, along the rows and down the columns, for the three wavebands using both raw and transformed data, and for NDVI. Several mathematical models were fitted to the variograms. The double exponential model provided the best fit in all cases. The parameters of the models were different for the different directions and for the different variables. Two distinct scales of spatial variation are present in each case, a short range one Of about 9 pixels and a long range one of over a hundred pixels. The scale of variation is greater in the N-S direction. Based on these results we have recommended a reconnaissance sampling plan comprising a series of transects over the study area for obtaining field data to correlate with the image. (AN).