Methods for predicting water temperature in data-scarce areas under different climate regions of China

Abstract

Water temperature is an important index that affects physical, chemical and biological reactions in water environments, and accurate water temperature prediction is important. Water temperature prediction in a data-deficient area along the Yangtze River trunk stream was selected as the research object, the factors influencing water temperature changes, such as air temperature, latitude and elevation, were analyzed, and the main factors were determined. A linear regression equation of water temperature and air temperature under different climate types was constructed. The Air2stream model was used for water temperature prediction, and the model prediction accuracies were compared. (1) Water temperature changes are mainly controlled by air temperature, and (2) the averaged root mean square error (RMSE) of water temperatures predicted by the linear regression equation and Air2stream model were 1.79 °C and 1.40 °C, respectively. The averaged determination coefficients (R²) for the Air2stream model under the plateau alpine and subtropical monsoon climate types were 0.97 and 0.95, respectively. (3) The prediction accuracy of the Air2stream model exceeded that of the linear regression equation. Although the phenomenon of water temperature lagging behind air temperature is becoming increasingly obvious in high-flow areas, the water temperature prediction method of the water temperature-air temperature linear regression equation coupled with the Air2stream model can provide more reliable prediction results, thereby providinge a reference for water temperature prediction in data-deficient areas.