To assess the decadal water temperature (WT) changes observed at the center of Lake Kasumigaura, a shallow turbid lake in Japan, we constructed a prediction model for WT. The thermal interactions among air, water, and sediment were simulated by a one-dimensional differential equation using hourly observations of meteorological and limnological parameters. The validated model showed good performance (i.e., <1 °C RMSE for the daily averaged WT) for 2012, 2013, and 2015, respectively, in both upper and lower water layers, which were attained by incorporating a physical sub-model regarding the surface water temperature (surface WT). The good description of seasonal, daily, and hourly changes in surface WT obtained with the sub-model was confirmed by the observed and simulated time series. We then quantified and evaluated the trends of the meteorological and limnological parameters influencing WT changes from 1979 to 2015 based on their effects on WT by using the prediction model with the increases in air temperature (AT), solar radiation (SR), wind velocity (WV), and turbidity during that period; then, WT in the former period (1979) was predicted according to the trends and compared with the measured WT. The observed WT changes during the period were quantitatively explained by the compound effects of the parameters' changes, i.e., the AT and SR raising the WT, and the WV and turbidity lowering the WT.
Read full abstract