Can AI Predict Karst Spring Flow? Using Support Vector Machines скачать в хорошем качестве

Can AI Predict Karst Spring Flow? Using Support Vector Machines

Prediction of flow rate of karstic springs using support vector machines Authors: Manish Kumar Goyal, Ashutosh Sharma & Konstantinos L. Katsifarakis Predicting spring discharge in karstic aquifers is challenging due to their complex void structures and highly nonlinear flow pathways. This video presents a machine learning–based approach to forecast discharge from two adjacent karstic springs in Greece—Mai Vryssi and Pera Vryssi—using Support Vector Regression (SVR). Four SVR models with different kernels (linear, polynomial, Gaussian RBF, and exponential RBF) are tested using precipitation and spring flow data at daily and monthly scales. Model hyperparameters are optimized through grid search, and performance is evaluated using RMSE and correlation coefficients. Results show that kernel choice strongly influences performance, with the polynomial kernel performing best for Mai Vryssi and ERBF for Pera Vryssi. Overall, the SVR models outperform traditional approaches such as GRNN, RBF neural networks, and ARIMA, demonstrating the strength of AI-driven methods for understanding and predicting karst hydrological behavior. DOI: https://doi.org/10.1080/02626667.2017.1371847 This video is created by Saral AI https://saral.democratiseresearch.in/