Abstract: We examined whether the tomato growth simulation model DSSAT-CROPGRO-Tomato can accurately simulate the growth, development, and yield formation of tomatoes in northern China solar greenhouse under the condition of returning straw to the field. The model was verified based on the greenhouse tomato plot test data in both 2016 and 2018. There were four treatments of straw returning, including 0 (S₀), 1.5×10⁴ (S₁), 3×10⁴ (S₂) and 4.5×10⁴ kg·hm^-2 (S₃). The corresponding crop genetic parameters of different schemes were obtained using GLUE parameter estimation module. Moreover, the simulation accuracy of the model was verified by comparing the measured and simulated values of tomato phenology, fresh fruit yield, maximum leaf area index, soil moisture, soil inorganic nitrogen, and aboveground dry matter quality. The results showed that the coefficients of variation of the parameters PODUR (the duration of light 〖JP2〗and heat required for the final fruit load under optimal conditions) and SLAVR (specific leaf area)were 28.53% and 14.13%, respectively. The model had the highest simulation accuracy in 2016 when all treatments were under parameter estimation schemes, and its ARE (absolute relative error) and nRMSE were 10.33% and 7.12%, respectively. The model had high accuracy in simulating soil moisture, tomato growth, and yield. The results of leave-one-out cross-validation method showed that the overall error of the model to the yield of greenhouse tomato was 18.68%-21.95%. Our results indicated that the CROPGRO-Tomato model can accurately simulate tomato growth and yield formation in the solar greenhouse in Shenyang with straw returning to the field.

Key words: solar greenhouse, tomato, DSSAT-CROPGRO-Tomato, straw return, leave-one-out cross-validation.