Hybrid wavelet–ANN modelling for LAI forecasting under climatic variability: comparative case studies from the mediterranean basin

Seasonal variability in the Mediterranean precipitation regime significantly affects vegetation cover, particularly due to frequent and severe drought conditions. In this study, the Leaf Area Index (LAI) was adopted as a key ecological indicator for assessing vegetation status. Monthly total precipitation and near-surface air temperature were used as predictor variables, while MODIS-based LAI data from 2007 to 2023 served as the response variable. A hybrid Wavelet–Artificial Neural Network (W-ANN) approach, in which Daubechies wavelet coefficients of meteorological variables were provided as inputs to a Levenberg–Marquardt backpropagation ANN, was compared to a conventional ANN model applied directly to the raw data. Four urban locations with contrasting Mediterranean climates—Antalya and Istanbul (Kandilli) in Türkiye, and Enna and Trieste in Italy—were selected for model evaluation. Using both approaches, LAI was forecasted for the period 2024–2030, and predictive performance was comparatively assessed. Results indicated that the W-ANN model outperformed the conventional ANN, yielding 15–85% higher accuracy, with mean squared error (MSE) values ranging from 0.01 to 0.04 on the test datasets. Scenario simulations revealed a declining trend in LAI for Antalya and Enna, and an increasing trend in Istanbul and Trieste. The proposed framework offers a transferable tool for vegetation monitoring and climate adaptation in semi-arid regions.