Two new publications on the added value of high-resolution climate models
21 May 2024
Photo: zdf.de
In two recently published studies, Dr. Benjamin Poschlod explores the capabilities and added value of high-resolution climate models.
In the study “Convection-Permitting Climate Models Can Support Observations to Generate Rainfall Return Levels” Poschlod and his colleague Dr. Jonathan Koh combine climate model simulations and observations to derive spatially consistent information about extreme rainfall. In order to estimate the occurrence probability of these rare events, statistical models can be built using rainfall observations. However, these measurements are mostly point measurements and hence, one needs to interpolate in space when performing regional analyses. Often, topographical features such as elevation, latitude, and longitude are chosen as auxiliary variables to facilitate this interpolation. Poschlod & Koh propose to add high-resolution climate simulations as covariates. They show that the additional information provided by the climate model can improve the spatial representation of extreme daily rainfall. They also show, that this added value is only present for high-resolution (1.5 km) simulations, but vanishes for coarser-resolution reanalysis data (30 km).
In most parts of the world, the rain gauge density is less than in the study area over southern Germany. Therefore, the authors test the approach with smaller subsets of the observational data. The method proves to be robust under these conditions, highlighting potential for regions where observational coverage is scarcer but high-resolution climate simulations are available.
In the article “Snow depth in high-resolution regional climate model simulations over southern Germany – suitable for extremes and impact-related research?” Poschlod and Dr. Anne Sophie Daloz explore different multi-decadal high-resolution climate simulations and their ability to simulate snow depth and extremes. The authors identify a variety of uncertainty sources (deviation of elevation, climatic biases, albedo overestimation, snow density parametrization), which are interlinked. Still, the high-resolution climate models can outperform the state-of-the-art reanalysis product reproducing mean snow depth, the snow cover duration, and the presence of “White Christmas”. As first study, Poschlod & Daloz show that climate models even reproduce the characteristics, timing, and spatial patterns of extreme snow depths. Hence, they show potential for snow depth projections in the context of climate change research, even though careful evaluation is needed before any impact-related interpretation of the simulations.