【摘要】 为研究不同模型对花生生长发育的模拟效果，确定花生发育期的最佳模拟方法，基于正阳、驻马店、方城和黄泛区4个农业气象观测站26年花生发育期数据，选择分段线性模型(LM)、Wang-Engel模型(WE)、有效积温模型(EM)、活动积温模型(AM)和钟模型(CM)5种作物发育期模型，随机选取其中18年观测数据进行模型参数估算，根据均方根误差(RMSE)、标准化均方根误差(NRMSE)和平均绝对误差(MAE),比较了上述5种模型对河南省花生出苗、开花、下针和成熟4个主要发育期的拟合效果，并利用其余8年数据对模型效果进行验证。结果表明：(1)不同模型的积温指标存在明显差异，全生育期积温表现为AM的>WE的>EM的>LM的趋势。(2)WE模型对花生发育期的拟合效果相对较好，4个发育期RMSE、NRMSE和MAE的平均值分别为5.25 d、2.41%和3.87 d,均小于其他4种模型的。同一模型对不同发育期的拟合效果表现为出苗期的好于开花期的，开花期的好于下针期和成熟期的。(3)模型验证结果与拟合效果类似，也表现为WE模型中RMSE、NRMSE和MAE的平均值均小于其他模型的，分别为4.91 d、2.32%和3.97 d。对不同发育期的验证也表现为出苗期效果相对较好，其次是开花期的，下针期和成熟期效果相对较差。综上，WE模型可用来模拟河南省花生生长和发育过程。更多还原

【Abstract】 In order to study the simulation effectiveness of various models on peanut growth and development, and identify the best simulation method, the five models of crop development stages, piecewise linear model(LM), Wang-Engel model(WE), effective accumulated temperature model(EM), active accumulated temperature model(AM), and clock model(CM) were chosen, based on 26-year peanut development stage data on four agrometeorological observation stations in Zhengyang, Zhumadian, Fangcheng, and Yellow River Floodplain. The simulation performance of four main stages of peanut development, emergence, blooming, needling, and maturity fitted by the five models were compared with 18-year observation data which were randomly selected, using the metrics of root mean square error(RMSE), normalized root mean square error(NRMSE), and mean absolute error(MAE), and the performance was evaluated using the data from the remaining eight years. The results were as follows.(1) The accumulated temperature indexes of different models varied significantly, and the accumulated temperature of the whole period showed the AM>WE>EM>LM pattern.(2) The WE model fitted the developmental stages relatively well. The mean values of RMSE, NRMSE, and MAE were 5.25 d, 2.41%, and 3.87 d, respectively, at four stages of development, all lower than other four models. The fitting of the same model on different development stage was better for emergence stage than for blossoming stage, and better for blossoming stage than for needling stage and maturity stage.(3)The results of model validation were similar to the fitting results, also showing that the WE model had the lowest average RMSE(4.91 d), NRMSE(2.32%), and MAE(3.97 d). The verification of various development stages revealed that the effect was better in the emergence stage, followed by blossoming stages, while relatively worse in the needling and maturity stages. In conclusion, the WE model is suitable for simulating the growth and development process of peanuts in Henan Province.更多还原