The optimal irrigation management could be implemented by determining the crop water status corresponding to the negative stem water potential (SWP). However, there are several limitations in measuring SWP directly, especially the intrusive nature of the measurement. This paper aims to forecast a time when SWP starts to be negative by analyzing the linear relationship between vapor pressure deficit (VPD) and the canopy-air temperature differential. To predict the time the SWP drops negatively, an ARIMA model is used to forecast VPD values 30 minutes ahead. The proposed method was evaluated by conducting experiments in an actual environment. The experimental results showed that, from 24 days, the proposed method can forecast the water stress of durian trees correctly within 1-hour and 2-hour intervals for 14 and 6 days, respectively, which is promising in practice.