DEVELOPING GUIDELINES FOR REPLANTING GRAIN-SORGHUM .2. IMPROVED METHODS OF SIMULATING CARYOPSIS WEIGHT AND TILLER NUMBER

Enhancing the yield prediction accuracy of SORKAM, a growth model for sorghum [Sorghum bicolor (L.) Moench], requires improving the simulati on of grain growth and the relationship between tiller number and mini mum temperature. A new grain growth equation, developed from 1990 data collected at Manhattan, KS, relates grain-filling rate to the rate of change of plant dry matter per caryopsis during the effective filling period. Additional functions are specified for the lag and leveling-o ff phases of caryopsis growth, resulting in a new submodel, which was tested against data from Kansas and two Australian sites. The tiller n umber calculation was revised using three years of Manhattan data. Aft er incorporating the new submodels, SORKAM outputs were compared with observed yields and yield components in three Kansas regions. The R(2) values for yield increased from 0.27, 0.46, and 0.79 to 0.76, 0.69, a nd 0.90 for eastern, central, and western Kansas, respectively. Caryop sis weight comparisons showed that the revised model accounted for 48 to 72% of observed variability, up from 15 to 57% for the original mod el. Tiller number predictions also improved, although modestly.