Accurate solar irradiance estimation is critical for solar energy development in tropical regions, yet it remains challenging due to complex atmospheric dynamics in maritime climates. This study systematically examined the optimal combination of input parameters for global horizontal irradiance (GHI) estimation in Indonesia using multispectral observations from the GEO-KOMPSAT-2A (GK2A) geostationary satellite. A comprehensive ablation analysis was conducted across six model configurations to quantify the incremental contributions of solar geometry, clear-sky radiative potential, atmospheric attenuation, cloud optical properties from visible and near-infrared channels, thermal infrared cloud characteristics, and moisture information. Multiple machine-learning algorithms were evaluated. Categorical boosting (CatBoost) consistently achieved the best performance, characterized by the lowest relative root mean square error (rRMSE) and minimal bias. Shapley additive explanation (SHAP) analysis indicated that solar zenith and azimuth angles are the most influential predictors, followed by visible (VI006) and near-infrared (NR016) channels, highlighting the dominant role of cloud optical properties in tropical GHI estimation. The full-feature model (M6) achieved rRMSE values of 33.65% under partly cloudy conditions and 25.07% under clear skies, substantially outperforming ERA5 reanalysis and semi-empirical models. Validation using 36 ground-based pyranometer stations across Indonesia confirmed the robustness and generalizability of the proposed framework for operational tropical solar resource assessment. © The Author(s) 2026. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits the use, sharing, adaptation, distribution and reproduction in any medium or format, as long as appropriate credit to the original author(s) and the source is given by providing a link to the Creative Commons license and changes need to be indicated if there are any. The images or other third-party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.