Cloud-height mapping from all-sky camera network

Clouds play a key role in the Earth’s radiative balance and atmospheric dynamics, yet large uncertainties persist in their representation in weather and climate models. These uncertainties are partly related to the limited availability of continuous, high-resolution observations of cloud geometry. In this context, ground-based imaging networks provide a valuable opportunity to observe cloud fields with high temporal and spatial detail. In this work, we present a general framework for cloud detection and cloud-height retrieval using a network of 25 all-sky cameras distributed across the city of Valladolid (Spain) and its surrounding areas.

All instruments are identical OMEA-3C all-sky cameras operated and geometrically calibrated within the GOA-SCAN infrastructure of the Group of Atmospheric Optics. The proposed methodology combines image preprocessing, cloud-pixel segmentation, identification of matching cloud pixels, and stereoscopic reconstruction to derive instantaneous cloud-height fields. Cloud heights are retrieved through stereoscopic triangulation from planar-projected image pairs (Nguyen and Kleissl, 2014; Beekmans et al., 2016; Blum et al., 2021). The system provides continuous observations every five minutes, allowing the monitoring of cloud spatial structure and short-term evolution.

For each acquisition time, every camera is paired with all other cameras in the network, producing multiple independent cloud-height estimates based on row-wise correlation techniques. These estimates are filtered using geometric constraints, correlation quality metrics, and physical plausibility criteria. The use of multiple camera distances enables sensitivity to different cloud layers and ensures a spatially consistent coverage of the urban area and its surroundings.

A key component of this study is the validation of the retrieved cloud heights using independent ground-based observations. Cloud-base heights derived from the all-sky camera network are compared with measurements from a co-located ceilometer, allowing an objective assessment of the retrieval accuracy under different cloud conditions. This comparison provides insight into the performance of the stereoscopic approach and its limitations, particularly for low and multi-layer cloud scenes.

The presented framework establishes a robust basis for future developments, including extended validation with additional remote-sensing instruments and satellite products, as well as improvements in retrieval accuracy and operational applicability.

 

This work was supported by the Ministerio de Ciencia e Innovación (MICINN), with the grant no. PID2024-157697OB-I00. This work is part of the project TED2021-131211B-I00375 funded by MCIN/AEI/10.13039/501100011033 and European Union, “NextGenerationEU”/PRTR and is based on work from COST Action CA21119 HARMONIA. Financial support of the Department of Education, Junta de Castilla y León, and FEDER Funds is gratefully acknowledged (Reference: 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 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).

 

Beekmans, C., Schneider, J., Läbe, T., Lennefer, M., Stachniss, C., and Simmer, C. (2016) Atmospheric Chemistry and Physics, 16, 14231–14248.

Blum, N. B., Nouri, B., Wilbert, S., Schmidt, T., Lünsdorf, O., Stührenberg, J., Heinemann, D., Kazantzidis, A., and Pitz-Paal, R. (2021) Atmospheric Measurement Techniques, 14, 5199–5224.

Nguyen, D. A. and Kleissl, J. (2014) Solar Energy, 107, 495–509.