Laserfiche WebLink
<br />. <br /> <br />o <br />..:J <br />(~..:..) <br />w <br />00 <br /><Xl <br /> <br />. <br /> <br />status of lands within their division. Each office reviews the maps for its <br />area and assigns a unique ID to each field, based upon the township, range, <br />and section number. For many of the fields, crop type information is also <br />specified. A minimum of 15% of the fields on each map must be validated. <br />Many of the maps are 100% field-checked. <br /> <br />When review by the State Engineer's Office has been completed, the <br />.maps are returned to D-3744. Based upon this new information, the <br />ARC/INFO database is corrected and any additional data added. Final plots <br />are prepared and sent to the State Engineer's Office. This completes the <br />ARC/INFO vector database development. <br /> <br />IMAGE PROCESSING STEPS <br /> <br />Although significantly more costly than MSS (multi-spectral scanner) <br />data, TM (thematic mapper) was chosen as the most appropriate data <br />available for this project. Numerous studies have demonstrated that <br />substantiallY more land cover information can be derived from TM data <br />than from MSS data (Hopkins et.aI., 1990; Prout and Sutton, 1986; Anuta <br />et.aI., 1984). This is primarily attributable to the increased number of <br />spectral bands and increased radiometric sensitivity of TM. As a result of <br />the greater dimensionality of the TM data, the number of separable classes <br />produced is twice that of MSS data (Anuta et. aI., 1984). Image processing <br />steps for the Upper Colorado Irrigated Lands project are as follows: <br /> <br />1) download TM data from 9-track tape <br />2) extract AOI (area of interest) <br />3) edge enhance <br />4) apply Tasseled Cap transform <br />5) classify image data <br />6) evaluate signatures <br /> <br />Edge Enhancement <br /> <br />Once the image data have been downloaded and the particular AOI <br />extracted, the data are edge enhanced. Lee and Price (1993) have <br />successfully demonstrated that the integration of edge information can <br />increase land cover classification accuracies by approximately 10%. Edge <br />enhancers are kernels (Le., windows) that effectively increase the spatial <br />frequency of the data. In doing so, the edges between homogenous groups <br />of pixels are amplified. A low pass 3x3 box filter (Figure 1) was selected for <br />use in this study. This filter produces a "sharper" image with minimal data <br />. modification. Unlike edge detectors (zero-sum kernels) however, the edge <br /> <br />5 <br />