Abstract:
Soil moisture content is the most significant element in fit farming output and circulation of water, and its accurate
forecast is critical regarding water resource management. Mostly soil moisture is complicated through structural
features in addition to climatic complications. It’s tough to come up with an optimal mathematical model for soil
moisture since there are so many variables for calculation. Presented forecasting models have issues with
prediction accuracy, generality, plus other factors like Prediction performance, as well as multi-feature processing
capabilities etc. considering all these factors taking Gallipoli, Turkey as a reference site for developing a deep neural
network model to forecast moisture with good accuracy and minimum error. The dataset contains entities since
2008 to 2021. Doing quite a bit of mathematical analysis and establishing the correlation between selected features
with the spearman coefficient, the appropriate weather data is able to give proper weight to forecast soil moisture.
The output of the proposed method proves that the deep learning approach is realistic as well as efficient for the
prediction of moisture. Also, deep learning technique is able to make model generalizations with excellent accuracy
and minimum errors which is used to save irrigation water with controlling drought.
