Modeling and predicting reference Evapotranspiration (ET₀) is crucial for effective water resource management and the protection of natural ecosystems. While the Penman–Monteith (PM) equation is widely recognized for its high accuracy in estimating ET₀, it requires extensive meteorological data, which is often unavailable at many synoptic stations. In contrast, the Thornthwaite (TH) model is simpler and more widely used but suffers from limited accuracy. Previous studies have demonstrated that calibrating the TH model with additional variables can improve its accuracy; however, these efforts have been largely preliminary. This study addresses the accuracy gap by developing EvapoDeep, an advanced model for ET₀ modeling and prediction that leverages deep learning techniques. Initially, the difference in ET₀ (DET₀) between the TH and PM models is calculated. This DET₀ is then modeled using an advanced Convolutional Neural Network (CNN)-based deep learning method. The modeling component of EvapoDeep uses Global Navigation Satellite Systems (GNSS)-based Precipitable Water Vapor (PWV) data from California, utilizing one year of GNSS measurements. In the final phase, the prediction component of EvapoDeep employs the generated ET₀ time series to predict ET₀ for the subsequent day using Long Short-Term Memory (LSTM) networks, demonstrating the model's ability to handle temporal dependencies through its LSTM component. The EvapoDeep model significantly improved accuracy, achieving a Root Mean Square Error (RMSE) of 0.32 mm, compared to the TH model's RMSE of 1.09 mm, representing a 70% improvement. In addition, the next-day ET₀ predictions yielded an RMSE of 0.29 mm, underscoring its high predictive performance.

Urban hotspots are the points with elevated surface temperatures compared to their surroundings and are a critical component of the Urban Heat Island (UHI) effect and localized climate change. Over the past 25 years, Wrocław has experienced rapid urban development, which is one of the specific causes of hotspot formation and it is not still well explored. This study investigates the spatiotemporal dynamics of hotspots in different functional urban zones. It considers the land cover, urban form, and human activities. Landsat images from 1999 to 2023 were used to estimate Land Surface Temperature (LST) and Land Use/Land Cover (LULC) maps were created using an Artificial Neural Network (ANN). Hotspots were identified and linked to zones including urban expansion areas, commercial districts, residential zones, and transport infrastructure. Findings show that approximately 60% of hotspots are linked to the urban expansion and 20% to the commercial development. Loss of vegetation and increased built-up density were found to be the primary reasons of elevated surface temperatures. This research shows the importance of urban planning, including strategies such as expanding green spaces, implementing green roofs, and using reflective and permeable materials.

Identifying the key drivers of Surface Urban Heat Islands (SUHIs) is crucial for understanding urban thermal dynamics and informing mitigation strategies. This study models SUHI, examines the contribution of often-neglected atmospheric parameters, identifies key drivers, and pinpoints the spatial regions that most strongly influence SUHI intensity. An explainable deep learning framework integrating multi-source surface and atmospheric data is applied. The framework combines satellite-derived indicators of gray, green, and blue infrastructure, population density, topography from the Digital Elevation Model (DEM), and atmospheric variables from Weather Research and Forecasting (WRF) outputs to predict SUHI intensity. A residual attention-based Convolutional Neural Network (CNN), trained on a 13-year dataset, achieves robust predictive performance, with a Root Mean Square Error (RMSE) of 0.33 °C under normal conditions and 0.39 °C during heatwaves. Results show that atmospheric parameters, especially wind, are strongly associated with variations in SUHI intensity and contribute to enhanced rural cooling, which is linked to stronger urban–rural temperature contrasts. Precipitation and water vapor influence SUHI by altering evaporative and radiative cooling, while mean sea level pressure has a secondary role. Explainable Artificial Intelligence (XAI) techniques, including SHapley Additive exPlanations (SHAP) and Gradient-weighted Class Activation Mapping (Grad-CAM), indicate that urbanization intensity, vegetation, and population density are the primary contributors to SUHI variability, while wind and moisture are associated with changes in thermal contrasts. Spatial analysis further shows that densely built-up urban cores are consistently emphasized in the model’s predictions of SUHI. Grad-CAM highlights spatial interactions between urban and atmospheric parameters. By quantifying surface and atmospheric factors and identifying influential locations, this framework provides accurate SUHI estimates and actionable insights for urban planners, supporting urban ecosystem management and targeted heat adaptation strategies.