Improvement and validation of ORYZA model through introducing "transplanting time effect" and light sensitivity of rice

Objective Rice growth model is an important tool for realizing automatic management of rice production. The existing rice model does not fully consider the "transplanting time effect" in rice growth, and not easy to use as too many parameters required. Therefore, the model was improved for its more accurate and practical use.

Methods On the base of ORYZA and RCSODS models, combining with algorithms for the rice growth period and the importance of the periods in dry matter accumulation, the time effect of transplanting and the equation of dynamic extinction coefficient of rice were introduced into the model. The simulated growth period and yield were validated using the meteorological data in Suihua City, Heilongjiang Province from 2003 to 2012.

Results The mean square root of residual difference (RMSE) of simulation were respectively 1.17 days, 1.41 days, 1.00 day, 2.23 days and 2.12 days in the periods of greening, tillering, booting, heading and maturity. The RMSE and RE (relative error) of simulation of the entire development period were averaged 1.6 days and 1.3%. The observed annual yields of rice ranged from 6495 kg/hm² to 8715 kg/hm², and the simulated yields ranged from 7230 kg/hm² to 8207 kg/hm², with the minimum error in yield of 210 kg/hm², and the average RMSE and RE of 714 kg/hm² and 9.2%, respectively. The relative error rate of simulation by the model was kept within 10%.The acquired effective cumulative temperature before transplanting was significantly and positively related to that required for rice to complete the basic vegetative period (R² = 0.7837, P = 0.003), the growth rate parameter of the first growing stage in the model was adjusted according to this character, which decreased the simulation error in the date of growing period by three times, and in the yield by 34%.

Conclusions The introduction of the “Transplanting time effect” and the “Dynamic extinction coefficient” improved the simulation accuracy of the model in rice growth period and yield significantly. The response equation of the acquired effective accumulation temperature before transplanting and that required for whole vegetative period is introduced into the model, which greatly decreases the simulation error in the date arriving a growth period. The introduction of equation of dynamic rice extinction coefficient decreases the simulation errors in yield. All the modifications have improved the accuracy of the model.