Prospecting Potential Sites for Concentrating Solar Power (CSP) Plants through Embedding Similarity of the AlphaEarth Foundations Model

• Eduardo Chiarani UTFPR / UniSENAI
• Marlon Marcon UTFPR

DOI: https://doi.org/10.5753/eniac.2026.31426

Resumo

Prospecting geographic sites for Concentrating Solar Power (CSP) plants is carried out by weighting and processing several geospatial layers. This work proposes an alternative, in which sites are represented by the embeddings of the AlphaEarth Foundations model. To that end, 26 reference plants are compared among themselves by cosine similarity and the Wasserstein distance. Over those measures, a feature selection based only on positive examples, Sequential Forward Floating Selection (SFFS) with Leave-One-Plant-Out (LOPO) validation, reduces the 64 dimensions to a four-band signature. With the signature fixed, a site is scored by the same measures against the 26, under a top-k criterion. The efficacy of the method was verified with 22 operational and spatially independent plants held out of the reference set, of which 21 were recovered (95.5%). Another seventeen candidate areas were used as an application test, of which four came out adherent.

Referências

Alam, M. M. T. and Simic Milas, A. (2025). Dimensionality optimized machine learning retrieval of canopy chlorophyll, nitrogen, and phosphorus from Google satellite embeddings.

Brown, C. F. et al. (2025). AlphaEarth Foundations: an embedding field model for accurate and efficient global mapping from sparse label data. arXiv. Disponível em: [link].

Chiarani, E., Antunes, A. F. B., Drago, D., Oening, A. P., and Paschoalotto, L. A. C. (2023). Optimal site selection using GIS-based multi-criteria decision analysis (MCDA): a case study to concentrated solar power plants (CSP) in Brazil. Anuário do Instituto de Geociências, 46.

COP30 (2025). COP30 approves Belém Package 1. Disponível em: [link].

Demir, A., Dinçer, A. E., and Yılmaz, K. (2023). A novel method for the site selection of large-scale PV farms by using AHP and GIS: a case study in izmir, Türkiye. Solar Energy, 259:235–245.

Fernández-García, A., Zarza, E., Valenzuela, L., and Pérez, M. (2010). Parabolic-trough solar collectors and their applications. Renewable and Sustainable Energy Reviews, 14(7):1695–1721.

Gao, H. et al. (2026). TSP-Former: a phenology-guided transformer for tobacco mapping using satellite image time series. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 19:2423–2438.

IRENA (2025). Tracking COP28 outcomes: tripling renewable power capacity by 2030. Technical report, International Renewable Energy Agency, Abu Dhabi.

Panaretos, V. M. and Zemel, Y. (2019). Statistical aspects of Wasserstein distances. Annual Review of Statistics and Its Application, 6:405–431.

Pudil, P., Novovičová, J., and Kittler, J. (1994). Floating search methods in feature selection. Pattern Recognition Letters, 15(11):1119–1125.

Sun, Y., Zhu, D., Li, Y., Wang, R., and Ma, R. (2023). Spatial modelling the location choice of large-scale solar photovoltaic power plants: application of interpretable machine learning techniques and the national inventory. Energy Conversion and Management, 289:117198.

Turney, P. D. and Pantel, P. (2010). From frequency to meaning: vector space models of semantics. Journal of Artificial Intelligence Research, 37:141–188.