The potential impacts of environmental factors change on evapotranspiration in the Gaoping River Basin, Taiwan: A sensitivity analysis using Artificial Neural Networks (ANN)

Evapotranspiration (ET) returns more than half of the precipitation into the atmosphere, playing an important role in the water cycle. Past research indicates that global ET shows an increasing trend under climate change. However, the dominant drivers of ET differ between climate zones, plants also react differently under different environmental conditions. Research on the sensitivity of ET to combinations of multiple environmental factors remains limited. We focus on the Gaoping River Basin located in southern Taiwan, using the 2-kilometer gridded ET and meteorological data, including Temperature (T), Solar radiation (SR) and Vapor Pressure Deficit (VPD) from 1980 to 2021 produced by Taiwan Climate Change Projection Information and Adaption knowledge Platform (TCCIP), combined with an Artificial Neural Network (ANN) to develop the model. We divide the data into bins based on the percentiles of environmental factors to represent various environmental conditions, the sensitivity of ET to changes in environmental factors was calculated in each bin. The objectives of this study include (1) Identifying the dominant drivers of ET within Gaoping River Basin; (2) Examining whether the sensitivity of ET to environmental factors exhibits seasonal or interannual variations; and (3) Assessing the potential impacts of changes in environmental factors on ET under climate change. Preliminary results show that ET is most sensitive to T within Gaoping River, and the sensitivity of ET to SR varies between seasons.