MODELING RICE FIELD MOISTURE-CONTENT DURING THE HARVEST SEASON .2. MODEL IMPLEMENTATION AND VALIDATION

An algorithm was developed to implement the model of Lu and Siebenmorg en (1993) for predicting rice moisture content (m. c.) throughout a ha rvest season. Equations are presented for generating diurnal patterns of air temperature, relative humidity, and solar radiation from daily maximum and minimum temperatures and relative humidities, and the dail y total solar radiation. Rice m.c. was simulated using both hourly and daily field meteorological data for the harvest seasons of 1988 throu gh 1990 at two locations in Arkansas. The m.c.s predicted from both ho urly and daily input weather data were closely matched throughout the entire harvest seasons, with the differences being usually less than o ne percentage point (the average absolute relative differences for the three harvest seasons ranging from 3 to 5%). The model, using both ho urly and daily field meteorological data, predicted the rice m.c.s wit h reasonable accuracy; the average absolute prediction error between m easured and predicted m.c. varying between 1.2 and 2.2 percentage poin ts for the three harvest seasons. Large prediction errors (up to six p ercentage points) were obtained for some of the early harvest dates or when rice was harvested shortly after rain. Dew was predicted to appe ar on the kernel surface when the air relative humidity was about 94% or higher and could cause up to a four percentage point increase in ri ce m.c. overnight.