Frontiers in Climate (Oct 2022)
Comparing extremes indices in recent observational and reanalysis products
Abstract
Monitoring changes in climate extremes is vitally important in order to provide context for both our current and possible future climates. Datasets based on climate extremes indices from in situ observations and climate reanalyses are often used for this purpose. We assess the spatial and temporal consistency between these two classes of dataset on a global basis to understand where they agree or are complementary. As expected, the temperature time series expressed as anomalies, or self-normalizing indices, agree well. While there is sometimes a large spread in absolute values between products, both long-term trends and inter-annual variability are also in agreement. Spatially the temperature indices show high correlations, but comparisons between the cumulative distributions at each grid box show differences in regions at high altitude or where interpolation has been performed across climatic zones. The agreement is lower between the time series from observation-based and reanalysis datasets for precipitation indices. Trends in these indices show larger spatial heterogeneity, and inter-annual variation in the global averages is often larger than the magnitude of the long-term trend. These indices show larger spatial heterogeneity in the trends, which results in comparatively small long-term trends in the global averages, which are also small compared to the inter-annual variation. Spatially these indices show on average smaller correlations than for the temperature indices, but large regions show strong positive correlations for some precipitation indices. A subset of the reanalyses has higher correlations with the latest in situ-based dataset, HadEX3, and also have smaller differences in the per-grid box cumulative distributions, indicating close agreement to the observation-based dataset. Also, we outline how the comparisons herein suggest that the gridding method used when creating HadEX3 may need to be updated for future versions of this dataset, in order to retain detail arising from topographic features, for example.
Keywords
