The accurate estimation of solar radiation is crucial for optimizing solar energy deployment and advancing the global energy transition. This study pioneers the development of a hybrid model combining Recurrent Neural Networks (RNNs) and the Weather Research and Forecasting (WRF) model to estimate solar radiation in tropical regions characterized by scarce and low-quality data. Using datasets from Sumedang and Jakarta across five locations in West Java, Indonesia, the RNN model achieved moderate accuracy, with R2 values of 0.68 and 0.53 and RMSE values of 159.87 W/m2 and 125.53 W/m2, respectively. Additional metrics, such as Mean Bias Error (MBE) and relative MBE (rMBE), highlight limitations due to input data constraints. Incorporating spatially resolved GHI data from the WRF model into the RNN framework significantly enhanced accuracy under both clear and cloudy conditions, accounting for the region’s complex topography. While the results are not yet comparable to best practices in data-rich regions, they demonstrate promising potential for advancing solar radiation modeling in tropical climates. This study establishes a critical foundation for future research on hybrid solar radiation estimation techniques in challenging environments, supporting the growth of renewable energy applications in the tropics.
Accurate estimation of global horizontal irradiance (GHI) is essential for optimizing photovoltaic (PV) systems, particularly in regions with distinct climatic characteristics. Geostationary satellites, such as GK2A and COMS, provide consistent and spatially extensive data, offering a practical alternative to ground-based measurements. However, the performance of semi-empirical GHI models has been sparsely evaluated across diverse geographic zones. This study aimed to conduct a comparative analysis of four semi-empirical models—Beyer, Rigollier, Hammer, and Perez—applied to two contrasting locations: Seoul, South Korea (temperate) and Jakarta, Indonesia (tropical). Using satellite-derived cloud indices and ground-based pyranometer data, model performance was evaluated via RMSE, MBE, and their relative metrics. Results indicate that the Hammer model achieves the best performance in Seoul (RMSE: 103.92 W/m2; MBE: 0.09 W/m2), while the Perez model outperforms others in Jakarta with the lowest relative RMSE of 58.69%. The analysis outlines the limitations of transferring models calibrated in temperate climates to tropical settings without regional adaptation. This study provides critical insights for improving satellite-based GHI estimation and supports the development of region-specific forecasting tools essential for expanding solar infrastructure in Southeast Asia.
Accurate estimation of global horizontal irradiance (GHI) is essential for optimizing photovoltaic (PV) systems, particularly in regions with distinct climatic characteristics. Geostationary satellites, such as GK2A and COMS, provide consistent and spatially extensive data, offering a practical alternative to ground-based measurements. However, the performance of semi-empirical GHI models has been sparsely evaluated across diverse geographic zones. This study conducts a comparative analysis of four semi-empirical models—Beyer, Rigollier, Hammer, and Perez—applied to two contrasting locations: Seoul, South Korea (temperate) and Jakarta, Indonesia (tropical). Using satellite-derived cloud indices and ground-based pyranometer data, model performance was evaluated via RMSE, MBE, and their relative metrics. Results indicate that the Hammer model achieves the best performance in Seoul (RMSE: 103.92 W/m²; MBE: 0.09 W/m²), while the Perez model outperforms others in Jakarta with the lowest relative RMSE of 58.69%. The analysis highlights the limitations of transferring models calibrated in temperate climates to tropical settings without regional adaptation. This study provides critical insights for improving satellite-based GHI estimation and supports the development of region-specific forecasting tools essential for expanding solar infrastructure in Southeast Asia.
Tracking capacity fade and predicting the lifespan of Lithium Iron Phosphate (LiFePO4) batteries under calendar aging are crucial for the reliability of Battery Energy Storage Systems (BESSs) in tropical regions. Conventional empirical models often rely on static environmental averages and neglect coupled thermal–hygroscopic dynamics. To address these limitations, this paper introduces a multi-physics coupled state-space-based continuous integral model for battery degradation under dynamic tropical boundary conditions. The primary novelty of this research lies in the development of a continuous-time multi-physics state-space degradation model that explicitly captures the interconnected interactions between temperature, humidity, and State of Charge (SoC) under dynamically varying tropical microclimates. Calendar aging tests were conducted for 180 days inside an environmental test chamber under tropical microclimate conditions (average of 29.91 °C, RH of 77.26%), with reference performance tests executed at a low C-rate of C/20 to extract static electrochemical capacity. Parameter identification using an Ordinary Least Squares (OLS) solver demonstrates high model fitting, with R-squared values ranging from 0.8229 to 0.9429. Extrapolation results provide realistic end-of-life projections between 8.9 and 59.8 years and successfully identify the critical physical transition points P1 and P2 at the Solid Electrolyte Interphase (SEI) layer. Overall, this research provides a prognostic instrument for optimizing the operational management of utility-scale BESS in tropical climates.
Accurate solar irradiance forecasting in data-sparse tropical regions remains challenging due to complex terrain, rapid convective cloud formation, and limited ground-based observations. This study introduces a novel hybrid forecasting framework that integrates the Weather Research and Forecasting (WRF) model (10 km) with station-based Long Short-Term Memory (LSTM) bias correction, complemented by three innovative spatial site-adaptation strategies to produce spatially coherent short-term irradiance fields. The hybrid system leverages hourly Global Horizontal Irradiance (GHI) data from eight BMKG stations (2023) alongside GK2A satellite cloud information to dynamically correct WRF forecast biases, capturing nonlinear cloud–irradiance interactions that standard Numerical Weather Prediction (NWP) models fail to resolve. Results indicate that the hybrid WRF–LSTM system reduces 1–3-day root mean square error (RMSE) by 120–127 W/m2 and relative RMSE (rRMSE) by 26%, while lowering relative mean bias error (rMBE) from 31–36% (raw WRF) to 2.5–4.1%, with the largest improvements observed in regions exhibiting initially high WRF errors. Among the spatial adaptation methods, the average-based scheme minimizes RMSE but exhibits weak spatial coherence; the distance-weighted scheme achieves the strongest spatial consistency with regional reanalysis (R2 = 0.37) with minimal bias; and the elevation-based scheme ensures full-domain coverage with moderate skill. This study demonstrates that the integration of dynamical NWP modeling with LSTM-based bias correction and tailored spatial transfer strategies provides a robust, scalable approach for short-term solar irradiance forecasting and resource mapping in tropical environments. The proposed framework offers practical implications for PV power forecasting, grid management, and renewable energy planning in regions where observational data are sparse and the terrain is highly heterogeneous.