Single infrared channel can correct storm-cloud parallax across all Meteosat generations

Andrzej Kotarba of the Space Research Centre, Polish Academy of Sciences, has validated a cloud-top height method by Šoljan and colleagues (2024) for correcting parallax in geostationary imagery of deep convective clouds. The method matches 11 µm brightness temperature to a moist adiabat approximated with 30 polynomial coefficients, using ERA5 temperature, dew point and pressure.

Tested on about 1.7 million MODIS collocations with CloudSat/CALIPSO from 2007, it underestimated cloud-top height by 2.7 km (18%). Against the CLAAS-3 SEVIRI product (2004–2024) the bias was 1.1 km. A simple regression adjustment brought errors below 7%.

For European summer 2005, unadjusted parallax shifts averaged 1.3 km shorter than the CLAAS-based mean of 32.5 km. 84% of cloud pixels landed in the same image location (97% allowing for CLAAS uncertainty). Skipping correction altered frequency estimates by 13–20%; the method cut this below 7% monthly and 5% seasonally. The paper was published on 1 October 2026.

Why it matters: A single, sensor-agnostic correction could allow consistent storm climatologies from early Meteosat data onward, and possibly other geostationary fleets. Meteosat also covers Africa, but the study itself tested only Europe, and tropical performance was weaker.