SIMULATING FORAGE SORGHUM YIELDS WITH SORKAM

A validated crop model can be a valuable tool for evaluating new produ ction and management practices, particularly for crops routinely grown in environments that produce large variation in field performance tes ts. Previous work has been done to validate SORKAM a crop growth model for grain sorghum [Sorghum bicolor (L.) Moench]. The purpose of this study was to determine the ability of SORKAM to simulate forage sorghu m yields for a wide range of environmental conditions. Data for the si mulations used in the validation process were taken from performance t ests for four cultivars ('Early Sumac', 'Atlas', 'Sugar Drip', and 'Ca nex') conducted at two sites in Kansas over 25 yr. Simulated bloom dat e, grain, stover, and total dry matter yields were regressed on measur ed data. The SORKAM model simulated bloom dates without bias; however, the root mean square error (RMSE) was 8 d, and the model tended to ov erestimate bloom date under conditions of water deficit. The model gen erally overpredicted yields of the four cultivars for grain (bias = 3. 2 Mg ha(-1)), stover (bias = 2.0 Mg ha(-1)), and total dry matter (bia s = 5.2 Mg ha(-1)). In addition to overestimation of total dry matter production, weaknesses in the SORKAM model in simulating forage sorghu m growth included inability to simulate phenological development accur ately under moisture deficit stress and to partition dry matter into g rain and stover. These problems must be resolved before SORKAM can be used to address management decisions.