The surface urban heat island (SUHI) effect poses a significant threat to the urban environment and public health. This paper utilized the Local Climate Zone (LCZ) classification and land surface temperature (LST) data to analyze the seasonal dynamics of SUHI in Wuhan based on the Google Earth Engine platform. In addition, the SUHI intensity derived from the traditional urban–rural dichotomy was also calculated for comparison. Seasonal SUHI analysis showed that (1) both LCZ classification and the urban–rural dichotomy confirmed that Wuhan’s SHUI effect was the strongest in summer, followed by spring, autumn and winter; (2) the maximum SUHI intensity derived from LCZ classification reached 6.53 °C, which indicated that the SUHI effect was very significant in Wuhan; (3) LCZ 8 (i.e., large low-rise) had the maximum LST value and LCZ G (i.e., water) had the minimum LST value in all seasons; (4) the LST values of compact high-rise/midrise/low-rise (i.e., LCZ 1–3) were higher than those of open high-rise/midrise/low-rise (i.e., LCZ 4–6) in all seasons, which indicated that building density had a positive correlation with LST; (5) the LST values of dense trees (i.e., LCZ A) were less than those of scattered trees (i.e., LCZ B) in all seasons, which indicated that vegetation density had a negative correlation with LST. This paper provides some useful information for urban planning and contributes to the healthy and sustainable development of Wuhan.

Ecosystem service values (ESVs) are crucial to ecological conservation and restoration, urban and rural planning, and sustainable development of land. Therefore, it is important to study ESVs and their driving factors in the Dongting Lake Eco-Economic Zone (Dongting Lake). This paper quantifies the changes in ESVs in the Dongting Lake using land use data from 2000, 2005, 2010 and 2018. The eXtreme Gradient Boosting (XGBoost) model is used to study the effects of individual driving factors and the synergistic effects of these driving factors on ESVs. Our analysis suggests that: (1) From 2000 to 2018, the largest dynamic degree values in the Dongting Lake are in unused land types, followed by construction lands and wetlands. The ESVs of the Dongting Lake show an increasing trend, with those of forestlands being the highest, accounting for approximately 44.65% of the total value. Among the ESVs functions, water containment, waste treatment, soil formation and protection, biodiversity conservation and climate regulation contribute the most to ESVs, with a combined contribution of 76.64% to 76.99%; (2) The integrated intensity of anthropogenic disturbance shows a U-shaped spatial distribution, decreasing from U1 to U3. The driving factors in descending order of importance are the human impact index, total primary productivity (GPP), slope, elevation, population, temperature, gross domestic product, precipitation and PM2.5; (3) When the GPP is low (GPP < 900), the SHAP (SHapley Additive exPlanation) value of the high human impact index is greater than zero, indicating that an increase in GPP increases the ESVs in the Dongting Lake. This study can provide technical support and a theoretical basis for ecological environmental protection and ecosystem management in the Dongting Lake.

Short-term load forecasting (STLF) is crucial for intelligent energy and power scheduling. The time series of power load exhibits high volatility and complexity in its components (typically seasonality, trend, and residuals), which makes forecasting a challenge. To reduce the volatility of the power load sequence and fully explore the important information within it, a three-stage short-term power load forecasting model based on CEEMDAN-TGA is proposed in this paper. Firstly, the power load dataset is divided into the following three stages: historical data, prediction data, and the target stage. The CEEMDAN (complete ensemble empirical mode decomposition with adaptive noise) decomposition is applied to the first- and second-stage load sequences, and the reconstructed intrinsic mode functions (IMFs) are classified based on their permutation entropies to obtain the error for the second stage. After that, the TCN (temporal convolutional network), GRU (gated recurrent unit), and attention mechanism are combined in the TGA model to predict the errors for the third stage. The third-stage power load sequence is predicted by employing the TGA model in conjunction with the extracted trend features from the first and second stages, as well as the seasonal impact features. Finally, it is merged with the error term. The experimental results show that the forecast performance of the three-stage forecasting model based on CEEMDAN-TGA is superior to those of the TCN-GRU and TCN-GRU-Attention models, with a reduction of 42.77% in MAE, 46.37% in RMSE, and 45.0% in MAPE. In addition, the R² could be increased to 0.98. It is evident that utilizing CEEMDAN for load sequence decomposition reduces volatility, and the combination of the TCN and the attention mechanism enhances the ability of GRU to capture important information features and assign them higher weights. The three-stage approach not only predicts the errors in the target load sequence, but also extracts trend features from historical load sequences, resulting in a better overall performance compared to the TCN-GRU and TCN-GRU-Attention models.