基于气象-生理的夏玉米作物系数及蒸散估算

准确估算作物系数对预测作物实际蒸散量和制定精准的灌溉计划至关重要。为反映作物逐日作物系数变化,综合考虑气象和生物因子对作物生长的共同影响,采用五道沟水文实验站大型蒸渗仪夏玉米实测蒸散及气象数据,基于地温及叶面积指数建立了气象-生理双函数乘法模型,并结合梯度下降法对模型进行了精度优化。结果表明,在整个玉米生长期中,作物系数实测值和计算值平均绝对误差为0.12,均方根误差为0.15,相关性为0.91,蒸散量实测值与计算值平均绝对误差为1.0 mm/d,均方根误差为4.5 mm/d,相关性为0.75。该模型计算的全生育期蒸散量准确率(误差在2～3 mm/d以内)相比使用联合国粮农组织(FAO)推荐的作物系数计算所得准确率提高了3倍以上,可更精确用于作物系数及蒸散量计算。

Estimation of summer maize crop coefficient and evapotranspiration based on meteorology-physiology

Abstract: Accurate estimation of crop coefficient is critical to predicting actual crop evapotranspiration and developing accurate irrigation schedules. In order to reflect the daily crop coefficient changes of the crops, this study comprehensively considered the common influence of meteorological and biological factors on crop growth, a meteorological-physiological double function multiplication model was established based on geothermal temperature and leaf area index by using the evapotranspiration data and meteorological data of summer maize of lysimeter of the Wudaogou Hydrological Experimental Station. Combining with the gradient descent method, a meteorological-physiological double function multiplication model was constructed to estimate crop coefficients and optimize the accuracy of the model. The results showed that the model calculated the crop coefficient of summer maize. The model could also be used for the estimation of the evapotranspiration of summer maize with higher precision. In different samples, including training samples and test samples, and the entire growth period of summer maize, the precision calculated by the model was high. In the entire growth period of summer maize, between the measured value and the calculated value, the average absolute error was only 0.12. In addition, the root meant square error was 0.15. And the correlation was 0.91 between the measured value and the calculated value on the crop coefficient of summer maize. Regarding the estimation of evapotranspiration, between the measured value and the calculated value, the average absolute error was only 1.0 mm/d, and the root meant square error was 4.5 mm/d. In addition, the correlation was 0.75 between the measured value and the calculated value on the evapotranspiration of summer maize. And by using the crop coefficient of the Food and Agriculture Organization of the United Nations (FAO) which was recommended, the evapotranspiration of summer maize was calculated. And the model constructed in this study was compared with the calculated results, it was found that the evapotranspiration accuracy calculated by meteorological-physiological double function multiplication model was increased by more than 3 times when the error between measured value. The calculated value was within 2 mm/d and 3 mm/d. And the error was almost all within 4 mm/d. In this study, the meteorological-physiological double function multiplication model based on ground temperature and leaf area index could be more accurately used for crop coefficient and actual evapotranspiration calculation of summer maize. And the research incorporated a machine learning algorithm (a multi-disciplinary interdisciplinary) to further improve the accuracy of crop coefficient calculation. The construction of an accurate crop coefficient model had great significance for the accurate prediction of evapotranspiration and further for the development of accurate irrigation plans. The model simplified the calculation of summer corn crop coefficients, clarified the comprehensive influence degree of ground temperature and leaf area index on crop coefficients, improved the calculation accuracy, and could be used for the dynamic calculation of corn crop coefficients. The evapotranspiration data measured by a large-scale weighing lysimeter at the Wudaogou Hydrological Experimental Station was used to calculate the actual crop coefficient in combination with a series of meteorological data. The crop coefficient and leaf area index were fitted using the Michaelis-Menten equation to obtain the influence of biological factors. The multi-factor regression was used to select the meteorological factor-ground temperature (0 cm), which had the closest effect on the crop coefficient, and performed exponential fitting. Finally, the gradient descent method was used to optimize the model parameters, and the measured results were used to evaluate the calculation results. It had great significance to grasp the dynamic change characteristics of crop coefficients during the summer maize growth period and accurately estimate the actual evapotranspiration of crops.