Optimal irrigation decisions with limited water

MORGAN, COLORADO. THE OBJECTIVES OF THE STUDY WERE TO MAXIMIZE THE EXPECTED CROP RETURNS AND MINIMIZE VARIANCES DUE TO RANDOM PRECIPITATION. RECENT EXPERIMENTAL DATA FROM FIELD TRIALS ON CORN AT COLORADO STATE UNIVERSITY WERE USED. AN EVAPOTRANSPIRATION PREDICTION MODEL WAS DEVELOPED TO RECONSTRUCT SOIL MOISTURE OVER THE SEASON FOR VARYING IRRIGATION TREATMENTS. REGRESSION ANALYSIS WAS USED TO DETERMINE PARAMETERS IN PRODUCTION FUNCTIONS RELATING EVAPOTRANSPIRATION IN VARIOUS TIME PERIODS TO YIELDS. A DYNAMIC PROGRAMMING MODEL, ALSO DEVELOPED, DETERMINED OPTIMAL AMOUNTS OF IRRIGATION AT PRESCHEDULED TIMES FOR A SINGLE CROP. THE EVAPOTRANSPIRATION PREDICTION MODEL AND FIRST THE MULTIPLICATIVE AND THEN THE ADDITIVE PRODUCTION FUNCTIONS WERE INCORPORATED INTO THIS PROGRAM. THE MODEL WAS EMPLOYED TO SOLVE SEVERAL TYPES OF CASE PROBLEMS, INCLUDING ONE INVOLVING UNLIMITED WATER AVAILABLE AT A PREDETERMINED PRICE, AND ONE INVOLVING THE WATER"S BEING LIMITED DURING VARIOUS TIME PERIODS. THE RESULTS OF THIS LATTER PROGRAM SIMULATE THE SITUATION INVOLVING A CROP COMPETING WITH OTHER CROPS FOR THE LIMITED WATER. DETERMINISTIC AND NONDETERMINISTIC CASES WERE STUDIED. IN THE FIRST ALL VARIABLES WERE ASSUMED TO TAKE ON THEIR EXPECTED VALUES, THUS REDUCING THE PROBLEM TO A DETERMINISTIC LINEAR PROGRAM OF MAXIMIZING RETURN. IN THE SECOND, PROBABILISTIC RAINFALL WAS USED AS A SUPPLEMENT TO KNOWN SURFACE AND GROUNDWATER SUPPLIES. A SIMULATION PROGRAM WAS FORMULATED AND RUN TO EVALUATE THE EXPECTED RETURN AND VARIANCE FOR VARIOUS PRECIPITATION PLANNING POLICIES.