Retrieval of Solar Shortwave Irradiance from All-Sky Camera Images

A convolutional neural network (CNN) based model, named CNN-CMF, is proposed to estimate solar shortwave global horizontal irradiance (GHI) from daytime all-sky camera images by retrieving the cloud modification factor (CMF). This work explores the use of all-sky cameras as an additional observational resource for solar radiation studies. The model has been trained and tested using a total of 237,669 sky images paired with pyranometer GHI measurements from Valladolid and Izaña (Spain) and Lindenberg (Germany). A comparison between model results and pyranometer data shows a high determination coefficient (R²) of 0.99 for the test dataset. Statistical metrics show a mean bias error (MBE) of −2% and a standard deviation (SD) of 9%, indicating a slight underestimation of the model. The generalization capability of the model was examined using independent measurements from the Antarctic station of Marambio, which was not included in the training dataset. The retrieved GHI values remained a high correlation, with an R² of 0.95. The statistical metrics show at this location a small overestimation of the model GHI values (MBE ≈ 2%), increasing the uncertainty in the precision of the model (SD ≈ 26%). The model results present an improvement when daily irradiation values (GHI_d) are retrieved. These results show a performance of the model yielding MBE and SD values of approximately 3% and 11%, respectively, and an R² value up to 0.99.

This work was supported by the Ministerio de Ciencia e Innovación (MICINN), with the grant no. PID2024-157697OB-I00 and TED2021-131211B-I00375. Financial support of the Department of Education, Junta de Castilla y León, and FEDER Funds is acknowledged (CLU-2023-1-05). This work was funded by European Comision through the EUBURN-RISK project (INTERREG-SUDOE; S2/2.4/F0327). The authors acknowledge the support of COST Action CA21119 HARMONIA and the Spanish Ministry for Science and Innovation to ACTRIS ERIC and the Marie Sklodowska-Curie Staff Exchange Actions with the project GRASP-SYNERGY (grant no. 10 101131631).

González-Fernández, D., Román, R., Mateos, D., Herrero del Barrio, C., Cachorro, V.E., Copes, G., Sánchez, R., García, R.D., Doppler, L., Herrero-Anta, S., et al. (2024). Remote Sensing, 16, 3821.