Deterministic model of soil moisture to predict forage yield on semiarid rangelands

Abstract

Rangeland plays an integral part in the economy of the United States' Southwest. The past is characterized by a history of range degradation. Consequently, rangeland managers are demanding new techniques to evaluate alternative grazing practices. The development of the model described in this thesis is in response to those demands. The developed model is a continuous, deterministic computer simulation model that predicts the soil moisture regime of a soil profile. This in turn is used to calculate a stress index value of the soil. Annual forage yield is then estimated as a function of these stress index values. The model is designed to accept temperature and precipitation data, either actual or simulated, and to calculate forage yield. The model consists of five parts: evapotranspiration, interception, soil moisture movement, surface runoff and forage yield. The model was tested on data from the Santa Rita Experimental Range near Tucson, Arizona. Ten years of forage data were available from the experimental range. The model was run with input data of temperature and precipitation for the same ten years. After calculating stress index values for that period, multiregression analysis was used to relate stress index to actual forage production; an r² of .73 was obtained. The same ten year period was then simulated to predict forage yield. The average annual difference between the actual and simulated yield data was .1 kg/hectar.