The VALUE perfect predictor experiment: Evaluation of temporal variability

Abstract

Temporal variability is an important feature of climate, comprising systematic variations such as the annual cycle, as well as residual temporal variations such as short‐term variations, spells and variability from interannual to long‐term trends. The EU‐COST Action VALUE developed a comprehensive framework to evaluate downscaling methods. Here we present the evaluation of the perfect predictor experiment for temporal variability. Overall, the behaviour of the different approaches turned out to be as expected from their structure and implementation. The chosen regional climate model adds value to reanalysis data for most considered aspects, for all seasons and for both temperature and precipitation. Bias correction methods do not directly modify temporal variability apart from the annual cycle. However, wet day corrections substantially improve transition probabilities and spell length distributions, whereas interannual variability is in some cases deteriorated by quantile mapping. The performance of perfect prognosis (PP) statistical downscaling methods varies strongly from aspect to aspect and method to method, and depends strongly on the predictor choice. Unconditional weather generators tend to perform well for the aspects they have been calibrated for, but underrepresent long spells and interannual variability. Long‐term temperature trends of the driving model are essentially unchanged by bias correction methods. If precipitation trends are not well simulated by the driving model, bias correction further deteriorates these trends. The performance of PP methods to simulate trends depends strongly on the chosen predictors.

An intercomparison of a large ensemble of statistical downscaling methods over Europe: Results from the VALUE perfect predictor cross‐validation experiment

“…Thus, some clearly poor performing methods have been also included to illustrate problems. We want to remark that this experiment alone is not sufficient to evaluate the limitations of (MOS) BC techniques (see Maraun, Shepherd, et al () for more details). Moreover, it also does not fully validate PP techniques because further results using GCM predictors are needed to evaluate whether well‐represented predictors have been used and the PP assumption is valid.…”

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

“…But RCMs resolve processes and small‐scale variability below the GCM resolution that are crucial to represent short term persistence and spatial structure. Well performing RCMs may thus add crucial value as has been shown for, for example, short term persistence of daily maximum temperature (visible in Maraun et al ., ) and spatial correlations of temperature (Widmann et al ., submitted manuscript, 2018) (see also Figures and ). Of course, deficiencies of the chosen RCM may also deteriorate the performance of the driving model as has been the case, for example, for inter‐annual variability of spring temperatures (Maraun et al ., ; see also Figures and ).…”

“…The definition of reference scales follows Maraun et al . (). Mean: twice the standard deviation of daily values; variance (daily and inter‐annual), spell length, amplitude of seasonal cycle, spatial correlation length: the value itself; return level: absolute deviation from mean; correlations: 1; phase of the annual cycle: 1 month [Colour figure can be viewed at wileyonlinelibrary.com]…”

“…The definition of reference scales follows Maraun et al . (). For all indices apart from annual cycle phase: the value itself; phase of the annual cycle: 1 month [Colour figure can be viewed at wileyonlinelibrary.com]…”

“…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).…”