Observational uncertainty and regional climate model evaluation: A pan‐European perspective

Abstract

The influence of uncertainties in gridded observational reference data on regional climate model (RCM) evaluation is quantified on a pan‐European scale. Three different reference data sets are considered: the coarse‐resolved E‐OBS data set, a compilation of regional high‐resolution gridded products (HR) and the European‐scale MESAN reanalysis. Five high‐resolution ERA‐Interim‐driven RCM experiments of the EURO‐CORDEX initiative are evaluated against each of these references over eight European sub‐regions and considering a range of performance metrics for mean daily temperature and daily precipitation. The spatial scale of the evaluation is 0.22°, that is, the grid spacing of the coarsest data set in the exercise (E‐OBS). While the three reference grids agree on the overall mean climatology, differences can be pronounced over individual regions. These differences partly translate into RCM evaluation uncertainty. For most cases observational uncertainty is smaller than RCM uncertainty. Nevertheless, for individual sub‐regions and performance metrics observational uncertainty can dominate. This is especially true for precipitation and for metrics targeting the wet‐day frequency, the pattern correlation and the distributional similarity. In some cases the spatially averaged mean bias can also be considerably affected. An illustrative ranking exercise highlights the overall effect of observational uncertainty on RCM ranking. Over individual sub‐domains, the choice of a specific reference can modify RCM ranks by up to four levels (out of five RCMs). For most cases, however, RCM ranks are stable irrespective of the reference. These results provide a twofold picture: model uncertainty dominates for most regions and for most performance metrics considered, and observational uncertainty plays a minor role. For individual cases, however, observational uncertainty can be pronounced and needs to be definitely taken into account. Results can, to some extent, also depend on the treatment of precipitation undercatch in the observational reference.

Comparison of statistical downscaling methods with respect to extreme events over Europe: Validation results from the perfect predictor experiment of the COST Action VALUE

“…Section 2 describes the data and methods used. Details on the data can also be found in Kotlarski et al (), on the downscaling methods, predictor variables, and the validation setup in Gutiérrez et al (). Section 3 presents the results of the validation of extremes.…”

Statistical downscaling skill under present climate conditions: A synthesis of the VALUE perfect predictor experiment

“…Kotlarski et al . () evaluated a suite of RCMs against three different gridded reference data sets. They found that the uncertainty inherent in these datasets was typically smaller than climate model uncertainty.…”

“…In addition to the evaluation of downscaling methods, VALUE also addressed the quality of observational datasets used as reference for the evaluation (Kotlarski et al ., ; Herrera et al ., submitted manuscript, ). Kotlarski et al .…”

“…Similarly, it was not possible to systematically compare a range of different predictor selections for a specific method type. This special issue presents the VALUE results so far, including a discussion of the interface between climate modelling and users (Roessler et al ., ), uncertainties resulting from observational data sets (Kotlarski et al ., ; Herrera et al ., submitted manuscript, ), and the evaluation results of the perfect predictor experiment (Gutiérrez et al ., ; Maraun et al ., ; Hertig et al ., ; Soares et al ., submitted manuscript, 2018; Widmann et al ., submitted manuscript, 2018).…”

seNorge2 daily precipitation, an observational gridded dataset over Norway from 1957 to the present day

“…It is worth mentioning that the seNorge2 daily mean temperature dataset has been described in the paper by Lussana et al (2018). Despite being released only recently, seNorge2 has already been used in a few applications, such as snow and permafrost mapping (Gisnås et al, 2017) and evaluation of climate projections (Kotlarski et al, 2017). Most noticeably, the Norwegian Water Resources and Energy Di-rectorate (NVE) uses seNorge2 as meteorological forcing for the national forecasting system for floods, avalanches and landslides.…”