Comprehensive evaluation of the improved SPUR model (SPUR-91)

SPUR (Simulation of Production and Utilization of Rangelands) is a physically-based model designed to simulate the complexity of rangeland ecosystems (Wight and Skiles, 1987). Results of extensive validation testing in Texas indicated a need to modify the model to improve submodel integration and procedural guidelines. A description of the modified model (SPUR-91) is provided. A sensitivity analysis was conducted, examining the influence of changes in model output corresponding to perturbations made to individual input parameters. The model was then extensively validated using a Texas data set composed of simultaneous measurements of hydrology, plant, livestock, and meteorological parameters. Results of validation testing of SPUR-91 confirmed that the model: (1) predicted initial soil water content within 3% of observed; (2) predicted evapotranspiration, even under very low cover conditions, within 1% of observed; (3) provided a good correlation of temporal fluctuation (R = 0.72) of observed herbaceous production, the observed 4-year mean live standing crop for the major individual species was predicted within 1-13%; (4) could produce a determinant-type growth curve to approximate the long-term response of shrubs and trees; and (5) responded to management (grazing, vegetation manipulation) in a manner consistent with observations. A discussion of why modifications to SPUR improved the performance of SPUR-91 is presented.