The research we build on.
These are the papers behind SOLSAT. They cover satellite imaging, sunlight and solar panels. Each entry says how we use it.
The research library.
Filter by topic. Each entry links to the original source and says exactly how it informs SOLSAT.
2021
A global inventory of photovoltaic solar energy generating units
Kruitwagen, L., Story, K. T., Friedrich, J., Byers, L., Skillman, S., Hepburn, C. · Nature 598, 604 to 610
Benchmark for utility-scale detection from Sentinel-2 and SPOT imagery; informs our solar-farm screening.
Read source2018
DeepSolar: A machine learning framework to efficiently construct a solar deployment database in the United States
Yu, J., Wang, Z., Majumdar, A., Rajagopal, R. · Joule 2(12), 2605 to 2617
Proof that rooftop PV can be mapped nationally from imagery; the template for our as-built detection.
Read source2016
Automatic detection of solar photovoltaic arrays in high resolution aerial imagery
Malof, J. M., Bradbury, K., Collins, L. M., Newell, R. G. · Applied Energy 183, 229 to 240
Early high-resolution PV detection baseline we compare panel-counting against.
Read source2016
Distributed solar photovoltaic array location and extent dataset for remote sensing object identification
Bradbury, K. et al. · Scientific Data 3, 160106
Public labelled dataset used for pre-training before Thai fine-tuning.
Read source2015
U-Net: Convolutional networks for biomedical image segmentation
Ronneberger, O., Fischer, P., Brox, T. · MICCAI 2015 · arXiv:1505.04597
The family of models behind how we outline roof planes.
Read source2023
Segment Anything
Kirillov, A. et al. · ICCV 2023 · arXiv:2304.02643
Foundation-model prompting for fast obstacle and roof-edge labelling.
Read source2024
Mapping solar potential: comprehensive PV system detection with aerial and satellite images using SolarSAM
Matar, J., Albrecht, M., Kandelaki, A., Kotthoff, F., Duchon, M. · IEEE SmartGridComm 2024, Oslo
Shows how the Segment Anything model (SAM) can outline solar panels in aerial and satellite images.
Read source2022
Photovoltaics detection in satellite imagery using deep learning and remote sensing
MIT thesis · DSpace@MIT, handle 1721.1/147238
Capacity estimation from detected arrays, a step toward our portfolio scan.
Read source2016
An introduction to Himawari-8/9: Japan’s new-generation geostationary meteorological satellites
Bessho, K. et al. · J. Meteorol. Soc. Japan 94(2), 151 to 183
Sensor specification for the Himawari-8/9 imagery we ingest.
Read source2024
Developing a Thailand solar irradiance map using Himawari-8 satellite imageries and deep learning models
Suwanwimolkul, S. et al. (Chulalongkorn University) · arXiv:2409.16320
A Thai benchmark for mapping sunlight from satellites. It is our closest local reference.
Read source2018
Evaluation of Himawari-8 surface downwelling solar radiation by ground-based measurements
Damiani, A. et al. · Atmos. Meas. Tech. 11, 2501 to 2521
Shows that clouds bias satellite sunlight readings (global horizontal irradiance, GHI). This is why we calibrate on local data.
Read source2023
Thailand regional solar PV forecasting (with EGAT)
Hawai‘i Natural Energy Institute · Project report
One-hour-ahead forecasts from Himawari-8 over Thai test sites. Shows the approach at grid scale.
Read source2002
A new airmass independent formulation for the Linke turbidity coefficient
Ineichen, P., Perez, R. · Solar Energy 73(3), 151 to 157
A clear-sky model we tune for Thai haze and dust (PM2.5) seasons.
Read source2014
Learning phrase representations using RNN encoder-decoder for statistical machine translation
Cho, K. et al. · EMNLP 2014 · arXiv:1406.1078
Origin of the memory unit (GRU) in our sunlight forecasting model.
Read source2018
The National Solar Radiation Data Base (NSRDB)
Sengupta, M. et al. · Renew. Sustain. Energy Rev. 89, 51 to 60
Reference design for a satellite-derived solar resource database.
Read source2016
Floating photovoltaic power plant: A review
Sahu, A., Yadav, N., Sudhakar, K. · Renew. Sustain. Energy Rev. 66, 815 to 824
Engineering background for floating-array yield and cooling effects.
Read source2023
Mapping global water-surface photovoltaics with satellite images
Nanjing University-led team · Renew. Sustain. Energy Rev. (2023)
Global floating-PV inventory; method for water-body screening.
Read source2026
Water-surface PV inventory for China, 2015 to 2024 (SAR + optical fusion)
ESSD authors (see DOI) · Earth Syst. Sci. Data 18, 5951 to 5968
Shows that radar sees through cloud and glare on water. This is our route for floating solar in the monsoon.
Read source2025
Enhancing irrigation reservoirs through floating photovoltaic: a satellite mapping-based multifunctional water-energy strategy
Benito, A. O., Lannutti, E., Verdezoto, P. C. · ISES Solar World Congress 2025
Uses Sentinel-2 satellite images to screen reservoirs for floating solar sites.
Read source2011
Combining solar photovoltaic panels and food crops for optimising land use: towards new agrivoltaic schemes
Dupraz, C. et al. · Renewable Energy 36(10), 2725 to 2732
Land-equivalent-ratio framing for our agrivoltaic research track.
Read source2018
pvlib python: a python package for modeling solar energy systems
Holmgren, W. F., Hansen, C. W., Mikofski, M. A. · J. Open Source Softw. 3(29), 884
Physics baseline for clear-sky and yield modelling.
Read source2013
Photovoltaic degradation rates: an analytical review
Jordan, D. C., Kurtz, S. R. · Prog. Photovolt. 21(1), 12 to 29
Source range for the degradation assumption in the financial twin.
Read source2016
Rooftop solar photovoltaic technical potential in the United States: a detailed assessment
Gagnon, P., Margolis, R., Melius, J., Phillips, C., Elmore, R. · NREL Technical Report NREL/TP-6A20-65298
Methodology reference for building-level technical potential.
Read source2021
Hydro-Floating Solar Hybrid at Sirindhorn Dam begins commercial operation
EGAT · Press release
Thailand’s flagship floating asset. A target site type for our monitoring service.
Read sourceCitations are to third-party work; listing does not imply endorsement of SOLSAT by the authors. Verify each reference against its DOI before quoting.
What we are working on next, and looking for partners on.
Cloud-robust monitoring
Combining radar (SAR) with normal satellite images, so floating and rooftop arrays stay visible through the rainy season.
Dust (PM2.5) and soiling
Linking how much dust is in the air to lost output and cleaning plans.
Crop and yield together
Measuring crops under raised panels to estimate how well land is shared between food and energy, from space.
Solar credit risk
Turning a yield range into loan-risk signals that lenders can check.