Irrigation is very important as it affects many active processes during the growth and development of plants.However, inappropriate watering may cause many issues, which may lead plant to dead or reduction output of plant.One of the common problems is water stress, which is indicated by the closure of the stomata, restricts transpiration and finally affects the fruit yield and internal quality.As a result, detecting the water stress level and avoiding the excessive watering are essential to secure fruit quality and yield.Water is one of the most essential ingredients for the growth of grape vines as it effects many active processes such as photosynthesis and the development of grape berries.In this study, hyperspectral imaging technology was used to measure the reflectance and predict water potential based on those features for more effective irrigation controls.Grape vine specimens were potted in glasshouse, and three levels of water treatment with two replicates were planned to create controlled water stresses.To our knowledge, water stress is often indicated by changes in leaf color and temperature, furthermore, there are also some other stress symptoms including morphological changes such as leaf curling or wilting.Measurements using leaf reflectance may provide a better method to standardize water status measurements for specific grapevine varieties.Grape vine leaves absorb most radiance in the visible band in the plant pigments such as chlorophyll and xanthophylls, but reflects most radiance in the near-infrared (NIR) band.Both plant nutrition and water stress would change the reflectance pattern for an efficiency drop of photosynthetic absorbance which would in turn cause a reflectance increase in the visible band but decrease in the NIR band.Therefore, the combination of data from spectral bands into indexes, such as the (NIR-Red)/(NIR+Red) or the NIR/Red, could enlarge spectral differences and provide additional details useful for water stress detection.In our study, vine leaf reflectance captured by hyperspectral imaging technology were extracted and then transformed to normalized difference vegetation index (NDVI) and Ratio.The water potential of leaves (WPL) which related to the captured hyperspectral images was detected by water potential instrument.Then the reflectance and WPL were used to develop partial least squares (PLS) models to predict water potential.PLS model was built to predict water potential through using reflectance features (NDVI and Ratio).The results showed PLS model using the reflectance features had the rp and RPD to 0.908 and 2.220, and RMSEP and bias to 0.775 and 1.717e-06 for estimating water potential of grape leaves.The experimental results indicated that using hyperspectral imaging technology enable to use reflectance to predict water potential for grape vine and be beneficial for grape field management.In our future work, the sample size would be increased and grape variety to extend the application scope of the prediction model and optimize the model to be more accurate and stable.