Abstract: Stratified water temperature data and concurrent meteorological data were collected by an automatic meteorological station specific for a lake crab farming microclimate from 2022 to 2024, water temperature variation characteristics at different depths under diverse weather conditions were analyzed, along with their correlations with meteorological factors．Additionally, a stepwise regression approach was adopted to develop predictive models for stratified water temperature under various weather conditions．Results revealed that significant variation characteristics in water temperature at each depth were demonstrated, aligning with trend in air temperature．Water temperature lagged behind air temperature, with smaller fluctuations than air temperature．A significant positive correlation between water temperature and air temperature was observed．Water temperature predictive models for sunny to cloudy, cloudy to overcast, and overcast and rainy weather conditions were constructed using average temperature, maximum temperature, and minimum temperature for current day and previous 1 to 3 d．Models were tested using retrospective validation and predictive assessment, with absolute errors ranging from 0.8 to 1.4 °C, showed robust fitting and could be used to predict water temperature in crab ponds．