Climate reanalysis performed well in Brazil—with a hot-season bias
WHAT THE EVIDENCE SAYS
Monthly ERA5-Land temperatures for 2010–2024 generally had root-mean-square errors below one degree Celsius relative to fourteen stations, but overestimated several hot September–October periods. Accuracy varied spatially.
WHY IT MATTERS / OUR READ
Reanalysis products fill observation gaps; local calibration becomes especially important when a project depends on seasonal heat exposure.
CAVEATS / UNCERTAINTY
This monthly validation does not establish the quality of hourly extremes, microsites or vegetation-response estimates; station coverage is sparse.
SOURCE
Theoretical and Applied Climatology — e Silva et al. (2026)
View original evidence ↗
Monthly temperatures in Mato Grosso were checked against fourteen stations.
Read the Wild Lands assessment →