Using machine learning to improve the estimation of actual evapotranspiration in China

Time:2022-5-5

In this issue, I share a paper on using machine learning method to improve China‘s actual evapotranspiration estimation received on the top journal of Hydrology in the field of hydrological resources. The title of the paper is
Improving local evaluationtransmission estimation across China during 2000-2018 with machine learning methodshttps://www.sciencedirect.com/science/article/pii/S0022169421005850Download the original text. This article discloses the final 1km actual evapotranspiration data set of China’s 10 day average annual 36 periods from 2000 to 2018, which can be found inhttps://doi.org/10.6084/m9.figshare.12278684.v5. Download
The main technical route of this work is as follows:
Firstly, five process models are output through meteorological data and remote sensing data, and then the relationship between simulated evapotranspiration and actual evapotranspiration of the five models is trained on the site. Six machine learning methods are used in the training, from which the optimal machine learning method can be obtained. In this paper, Gaussian process regression GPR is obtained, and then GPR is applied to integrate the actual evapotranspiration of China of the five process models, A new set of actual evapotranspiration data of China (chinaet10day / 1km) is obtained, and the product is compared with 8 existing high-resolution et products covering China. The results show that the product has higher accuracy

Using machine learning to improve the estimation of actual evapotranspiration in China

flow chart. jpg