The main goal of this work package is to quantify the decadal predictive skill of extremes of key climate variables (temperature, wind, and precipitation) and of exemplary user-relevant quantities for Central Europe (e.g. wind energy and losses associated with windstorms), which will be selected in collaboration with end users. In particular, we will quantify the added value of downscaling on the predictability of these quantities, which are mostly influenced by local conditions and physical processes on the regional scale. The objective is to develop or update methodologies for the computation of these variables on the global (GCM) and regional (RCM) scale. Since the temporal and spatial resolution of observations and model output data is not generally uniform, and a post-processing of simulated standard variables is necessary, the computation of the user-relevant quantities is not straightforward. Different skill metrics are applied to quantify the added value of regionalisation. The developed methods will contribute to the MiKlip II central evaluation system (CES) and to an effective dissemination of the project results to the potential users.
The work package C2-WP1 is conducted at the Karlsruhe Institute of Technology (KIT). It is subdivided in two thematic aspects:
Development/application of methodologies to derive user-relevant variables (in cooperation with IPRODUCTS and SUPPORT from Module D)
Quantification of the predictive skill and the added value of downscaling for these variables (in cooperation with C2-WP2).
T1 Development/application of methods to derive user-relevant quantities
Different and appropriate methodologies are required to derive user-relevant quantities. For this task existing approaches will be further extended or new methodologies will be developed. The methodologies will be applied to standard output of the global and regional decadal prediction system and to observational data sets.
T2 Analysis of decadal predictability and added value of downscaling
The intention of this task is to find and apply suitable skill metrics to assess the forecast accuracy and the reliability of the MiKlip decadal prediction system with respect to climate extremes and user-relevant variables. These metrics are applied to both the global and regional hindcast simulations, to quantify the added value of downscaling.
D1a Methodologies to derive user-relevant variables
D1b Homogeneous time-series of variables on regional and global scale
D2a Suitable skill metrics for accuracy and reliablilty
D2b Summary of predictive skill for extremes and user-relevant variables
Several potentially user-relevant variables have been identified. These variables are related to temperature extremes, wind extremes, heavy precipitation and agronomy (e.g. windstorm losses, crop productivity or heat waves).
Regarding windstorm losses, a collaboration with an insurance company has been established in cooperation with GERICS (Module D). Focus is given to forecasts of wind gusts potentially causing losses (gusts > 17.2m/s) for 1-5 years ahead. Especially the change in the number of days per year where wind gusts exceed this threshold is analysed. A “forecast” for 2012-2016 for the preop ensemble is linked to insurance data (Fig. 1). Since the present calibration is based on absolute wind speeds, we currently develop a bias correction and calibration method, together with Module E.
Regarding temperature extremes, we developed an index based on the exceedance of the 90th percentile of daily maximum temperature (Tmax) to identify heat waves for different European regions (see also C2-WP2). The results show a predictive skill for heat waves (e.g. for mean and maximum heat wave temperature) for the first years after initialisation.
Additionally, we analysed the predictive skill of several climate indices (ETCCDI) that are related to temperature extremes in the baseline1 ensemble. These indices include heating degree days, frost days and daily maximum summer temperature. All of these indices show high predictive skill for most of Europe, especially for lead years 2-5 (Fig. 2). The skill patterns are spatially robust, with higher skill scores for indices related to warm temperatures.
Regarding heavy precipitation, events that may cause floods in large European river catchments (Rhine, Elbe, Oder, Danube, Vistula) are analysed (see also C2-WP2). With this aim, we developed an index (based on 7day-running-mean precipitation and percentile exceedance) to derive flood events directly from regional precipitation data.
For agronomy, we investigate a climate index, which characterises the length of the growing season. The growing season starts when six consecutive days have temperatures above 5°C and ends when six days are below 5°C. For large parts of Europe, this index exhibits high positive skill scores in the baseline1 ensemble for lead years 2-5 (Fig. 3). Additionally, the productivity of several crops (wine, olives, fruit) is analysed using an index, which is based on precipitation, minimum/maximum temperature and soil parameters. First results show a high predictability in the baseline1 ensemble for nearly all lead times.
Overall, we found predictive skill over Europe on the decadal scale for several variables far beyond yearly mean temperatures. The results might be relevant for users outside the scientific community.
Institute for Meteorology and Climate Research (IMK-TRO), Karlsruhe Institute of Technology (KIT)
Prof. Dr. Joaquim Pinto
Institute for Meteorology and Climate Research (IMK-TRO), Karlsruhe Institute of Technology (KIT)
Julia Mömken
Feldmann, H. | Pinto, J.G., Laube, N., Uhlig, M., Moemken, J., Pasternack, A., Früh, B., Pohlmann, H., Kottmeier, C.
Moemken, J. | M. Reyers, B. Buldmann, and J.G. Pinto JG
Woollings, T. | C. Franzke, D.L.R. Hodson, B. Dong, E.A. Barnes, C.C. Raible and J.G. Pinto
Seregina, L.S. | R. Haas, K. Born, and J.G. Pinto
Reyers, M. | J.G. Pinto, and J. Moemken