Utility of remotely sensed data for identification of soil conservation practices

摘要

Discussed are a variety of remotely sensed data sources that may have utility in the identification of conservation practices and related linear features. Test sites were evaluated in Alabama, Kansas, Mississippi, and Oklahoma using one or more of a variety of remotely sensed data sources, including color infrared photography (CIR), LANDSAT Thematic Mapper (TM) data, and aircraft-acquired Thermal Infrared Multispectral Scanner (TIMS) data. Both visual examination and computer-implemented enhancement procedures were used to identify conservation practices and other linear features. For the Kansas, Mississippi, and Oklahoma test sites, photo interpretations of CIR identified up to 24 of the 109 conservation practices from a matrix derived from the SCS National Handbook of Conservation Practices. The conservation practice matrix was modified to predict the possibility of identifying the 109 practices at various photographic scales based on the observed results as well as photo interpreter experience. Some practices were successfully identified in TM data through visual identification, but a number of existing practices were of such size and shape that the resolution of the TM could not detect them accurately. A series of computer-automated decorrelation and filtering procedures served to enhance the conservation practices in TM data with only fair success. However, features such as field boundaries, roads, water bodies, and the Urban/Ag interface were easily differentiated. Similar enhancement techniques applied to 5 and 10 meter TIMS data proved much more useful in delineating terraces, grass waterways, and drainage ditches as well as the features mentioned above, due partly to improved resolution and partly to thermally influenced moisture conditions. Spatially oriented data such as those derived from remotely sensed data offer some promise in the inventory and monitoring of conservation practices as well as in supplying parameter data for a variety of computer-implemented agricultural models.