Validation of Convective Parameters in MPI-ESM Decadal Hindcasts (1971–2012) against ERA-Interim Reanalyses

Abstract

Decadal forecasts of the Max Planck Institute – Earth System Model (MPI-ESM) are validated against ERA-Interim reanalysis data for the period 1979 to 2012 over Europe with respect to (a) the medians of 500 hPa temperature, 925 hPa temperature and 925 hPa mixing ratio, and (b) the 90th percentiles of parameters that are relevant for convective storms. These parameters are the vertical temperature gradient between 850 and 500 hPa, the low-level mixing ratio (25 hPa above the local model topography), and Convective Available Potential Energy (CAPE). The MPI-ESM evaluation reveals very little bias for the median 925 hPa temperature but a pronounced negative bias for the median 500 hPa temperature. Besides, a positive bias of the median 925 hPa moisture is found. The 90th percentiles of the convective parameters exhibit positive biases of the vertical temperature gradient and the low-level mixing ratio, which results in an overestimation of CAPE. It is hypothesized that MPI-ESM’s convective parameterization scheme is reluctant in dissipating CAPE across Europe, often leaving the atmosphere in a state that appears too favorable for convective storms. Biases for convective parameters are slightly larger for an enhanced resolution of MPI-ESM. A categorization with respect to forecast years indicates that these biases mostly result from a systematically deviant model climatology. However, a considerable part of the 500 hPa temperature bias is also introduced by the initial conditions. ERA-Interim shows an increase of low-level moisture over most parts of Europe in the past three decades, in particular after 1990. Steep vertical temperature gradients became less frequent in northwestern Europe and more frequent in southeastern Europe. The 90th percentile of CAPE exhibits little evolution in northwestern Europe, where these two changes largely compensate each other, whereas it increases in southeastern Europe, where they add up. MPI-ESM seems able to reproduce some of these tendencies, but not regional structures.