Detecting, evaluating, and monitoring land-use change on the southern High Plains of Texas
The study used remote sensing data and a geographic information system (GIS) to investigate relationships between changes in groundwater levels and changes in irrigated and non-irrigated land use in Hockley County, Texas from 1974 through 1982. The goal was to produce information for use in regional planning activities and to develop forecasting models. Objectives were detection of irrigated land use locations, identification of patterns of change in irrigated and non-irrigated land use, and forecasting locations and time frames of future changes. Data were organized as cells representing areas of 67m x 67m within GIS layers that contained Landsat data values, classified land uses, soil mapping units, surface elevation, depth to water and depth to base of the aquifer. Eight main classes of land-use change patterns across the study period were identified and compared with underlying saturated thicknesses using mean separation tests. These classes were tested for sensitivity to surface conditions, energy costs associated with pumping lift and artifacts produced by interpolation algorithims. Additional classes of change out of irigated land use at intervals of 2, 4, 6 and 8 years were related to saturated thicknesses; regression models were produced for each class. Conclusions were that irrigated land use is most closely related to saturated thicknesses of the underlying aquifer. Regression models for 1974 through 1980 indicated that the percentage of land irrigated over a given saturated thickness could be predictive of land use over the same thickness in a future year. Land use in 1982 could not be predicted. Inspection of Landsat and classified data suggested that the introduction of center pivot irrigation technology accounted for twenty-five percent of land that was not irrigated previously becoming irrigated. This distinctly affected the relationship that existed between saturated thickness and land use under row irrigation technology. Reliable forecasting models could not be developed without a longer time series that would permit evaluation of effects of this innovation on the aquifer-land use relationship.