Estimating solar radiation for plant simulation models

Abstract

There is considerable interest in using plant growth simulation models for large area yield forecasting in the United States and foreign areas . These models generally require daily input values of solar radiation, temperature, and rainfall. For calibration and evaluation of these models. historical data are needed. However, historical solar data are rarely available because solar radiation has been measured at only a few locations, frequently for only limited time periods, It would be necessary to develop surrogates for the historical ⢠data values. Five algorithms producing daily solar radiation surrogates using daily temperatures and rainfall were evaluated using measured solar radiation data for seven U,S. locations. The algorithms were compared both in terms of accuracy and of daily solar radiation estimates and in terms of response when used in a plant growth simulation model (CERES-wheat). Requirements for accuracy of solar radiation for plant growth simulation models were discussed. One algorithm was recommended as being best suited for use in these models when neither measured nor satellite estimated solar radiation values were available.