A FRAMEWORK TO DEAL WITH UNCERTAINTY IN SOIL AND MANAGEMENT PARAMETERS IN CROP YIELD SIMULATION - A CASE-STUDY FOR RICE

A framework is developed for crop yield simulation that accounts for uncertainty in soil and management input parameters. The framework consists of three steps: (1) the calculation of 'standard yield' using estimated soil and management parameters that are supposed to be representative for the land unit under consideration; (2) sensitivity analysis on the soil and management parameters; and (3) the calculation of yield probability from probability distributions for the soil and management parameters through Monte Carlo simulation. The results of sensitivity analysis are useful in setting up cost- and time-efficient measurement strategies. When a large number of actual soil and management measurements have been made, Monte Carlo simulation results in actual yield variability in the land unit. The developed framework is illustrated for rainfed upland rice on a loamy soil in The Philippines, using the MACROS crop growth model. The uncertainty in the simulated yield was large. there was 90% probability that simulated yield was between O.6 and 1.65 times the simulated standard yield in average years. In general, accurate yield prediction will be conditional on the availability of accurate soil and management parameters.