Chinese Journal of Agrometeorology ›› 2024, Vol. 45 ›› Issue (10): 1109-1122.doi: 10.3969/j.issn.1000-6362.2024.10.002

Previous Articles     Next Articles

Simulation of Evapotranspiration in a Well-facilitated Paddy Field Based on Machine Learning Algorithms and Energy Balance Closure

TAI Jiu, WANG Wei, XU Min, HU Ning, CHEN Shang, XU Jing-zheng, HU Xiao-xu, LV Heng, ZHU Zi-han, LAI Yu-jing   

  1. 1. Key Laboratory of Meteorological Disaster, Ministry of Education/ Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters/ Key Laboratory of Ecosystem Carbon Source and Sink, China Meteorological Administration/ Jiangsu Key Laboratory of Agricultural Meteorology, Nanjing University of Information Science & Technology, Nanjing 210044, China; 2. Climate Center of Jiangsu Province, Nanjing 210008; 3. Aerospace New Weather Technology Co., Ltd, Wuxi 214028
  • Received:2023-11-08 Online:2024-10-20 Published:2024-10-16

Abstract:

To find the optimal machine learning model for simulating the actual paddy evapotranspiration (ETa) in each growth stages, and to quantify the impacts of forcing energy balance closure on simulation results, authors firstly analyzed the temporal variations in ETa and its influencing factors (air temperature Ta, relative humidity RH, wind speed U, vapor pressure deficit VPD, soil moisture at 5cm depth SWC and incident shortwave radiation K) with in-situ observations in a well-facilitated paddy filed in Nantong city, Jiangsu in 2020. Then the ETa of each growth stages were simulated by two machine learning algorithms: back propagation (BP) neural network and random forest. Finally, the impact of forcing energy balance closure on ETa simulation by BP model was quantified. The results showed that the relative importance of influencing factors to paddy ETa differed among growth stages. K was the most important influencing factor for ETa, while SWC had negligible effect on ETa. Therefore, including Ksignificantly improved the ETa simulation by BP neural network algorithm with correlation coefficient (R) increased by 14.9% and root mean square error (RMSE) reduced by 51.1%. BP1 model containing the five meteorological factors (Ta, RH, U, VPD and K) ranked the best for simulating ETa before tillering stage, while the BP3 model including Ta, RH, and K was more suitable after tillering stage. Forcing energy balance closure had improved the simulation performance of the BP neural network algorithm especially before tillering stage. After forcing energy balance closure, the simulation of BP2 model (Ta, RH, U and K) had been improved most obviously among five variable combinations. The R between BP2 simulations and field observations increased by 3.5% and the RMSE reduced by 25.7%. 

Key words:

Evapotranspiration, Machine learning algorithm, Energy balance closure, Paddy field, Growth stage