Abstract: When the daily average water temperature exceeds 32℃, the hatching rate and feeding rate of some warm water fish decrease with the increase of temperature, grow slowly or stop growing, and are easy to induce diseases. In summer, the water temperature in the fish pond is not always in the high temperature period. In this paper, based on the water and air temperature data of Fushun Aquaculture Meteorological Service Test Base from 2017 to 2019, the variation characteristics of water temperature in each layer of fish pond in summer were analyzed, and the prediction models of average and maximum water temperature on the sunny days were established by stepwise regression analysis. Results showed that the daily max. water temperature in each layer in summer occurs from 18：00 to 19：00, which lagged behind the temperature by 1−3 hours. The period when the water temperature ≥32℃ was mainly concentrated from 16：00 to 22：00, and the hourly average water temperature of each layer in this period was 32.9−33.1℃, 32.6−32.7℃ and 32.1−32.3℃, respectively. The water temperature of each layer fluctuated and rose with the temperature in summer, especially from mid-June to early-July, while the maximum values of temperature and water temperature appeared in late-July. The average number of days with water temperature ≥32℃ in summer was 23 days, accounting for a quarter of the whole summer, mainly from late July to mid-August. During this period, the daily average water temperature of each layer was 28.5−34.4℃, 28.2−34.1℃ and 27.9−33.9℃, respectively. The back substitution test in 2017−2018 and the forecast test in 2019 showed that the average absolute error of the model was 0.3-0.8℃, and the average relative error was within 2.6%, indicating that the model has good prediction effect and can be better applied to the prediction of fish pond water temperature in summer.

Key words: Sichuan basin, Fish pond water temperature, Variation characteristics, Prediction