Understanding and researching water level changes are fundamental to many aspects, including climate change, environmental balance and the necessary conservation measures that have been better taken. Deep learning (DL) methods have become increasingly crucial for predicting drought and water levels using remote sensing satellite imagery. In this work, the Landsat-Normalized Difference Water Index (NDWI) is used to predict the drought and water level based on the Deep Convolutional Neural Network (DCNN) in the Chennai region, using time series data sets. DCNN has separately trained these NDWI values in both areas of future dynamics prediction. To compute Root Mean Square Error (RMSE) the Deep Convolutional Neural Network's is used. Also, unexpected changes in the NDWI sequence trend are well adapted through the network. According to the prediction of the future NDWI value and reasonable accuracy, the RMSE is kept less than 0.03% without providing supplementary data. By adopting the method specified in work, the water level change can be accurately carried out in advance. Active measures are taken to ensure and improve the water content in any area. © 2021, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.