Fostering Regional Decision Making by the Assessment of Uncertainties of Future Regional Extremes and their Linkage to Global Climate System Variability for China and East Asia
A collaborative project between the University of Birmingham and the University of Reading
- Dr GC Leckebusch, School of Geography, Earth and Environmental Science, University of Birmingham
- Dr K Hodges, Department of Meteorology, University of Reading
- Prof R Elliott, Department of Economics, University of Birmingham
FOREX general concept
Regional decision processes for the development of suitable adaptation strategies or the timely initiation of related mitigation efforts in East Asia will strongly depend on robust and comprehensive information about future near-term as well as long-term potential changes in the climate system.
Based on physical process understanding, it is important to quantify the regional effects of global or hemispheric scale phenomena for both, the scientific community to understand potential changes and the impact community (including decision makers) to act proactively in the most suitable manner.
The three most important aspects in this context are the provision of a) relevant, timely, and b) comprehensive information about potential changes as well as c) information about the quality of the confidence in the information. The latter can only be achieved by the successful quantification and communication of uncertainties inherent in the climate change information derived from available sources of modelling and understanding. The necessity of process validation in model (forecast) suites is therefore essential.
FOREX focusses on all three of these major aspects:
a) Relevant meteorological and climatological events: FOREX will investigate extreme meteorological events in China and East Asia with respect to the systems driving the major modes of variability of extreme precipitation and damaging wind speeds. For the rainy summer season (June/July) precipitation extremes related to the Meiyu/Changma/Baiu-Front will be investigated. Regional wind and precipitation maxima are related to local pressure minima (e.g., cyclones) embedded in the moisture convergence zone around the East & South China Sea, affecting China, Korean Peninsula and Japan. In winter the influence of Extra-tropical Cyclones (ETC) in the northern part of this region is of importance with increased extreme wind speeds especially over the Yellow Sea region under potential climate change conditions (e.g. Bengtsson et al., 2008). The other main feature of extreme conditions is constituted by the occurrence of Tropical Cyclones (TC, Typhoons) in the West-Pacific Basin affecting the entire East Asian domain with devastating impact due to rain and extremely strong winds.
b) Comprehensive information about potential changes will be derived by investigating the underlying physical systems, which result in extremes in meteorological variables (e.g. precipitation, wind) and the responsible low pressure systems in relation to the larger-scale synoptic conditions. This will go far beyond the variable-only-based perspective from UKCP09. In the first instance we will investigate the related regional extremes of precipitation and surface near wind speed (incl. gusts), and secondly we will investigate the dependency of the occurrence of these events on large-scale forcing factors acting on synoptic to hemispheric scales, thus linking regional extremes with the global climate system and its variability.
c) Assessment and diagnosis of uncertainties and their communication: FOREX will establish a unique two-sided perspective to assess uncertainties. We will focus on the physical hazard as well as the related extremes in the impact domain (Flooding, wind damage). For the assessment of uncertainties of the physical hazard we will apply state-of-the-art cyclone tracking and identification methods (e.g., Hodges, 1995) for the quantification of extreme ETCs and TCs in the various available ensemble model data sets (e.g. CMIP3/CMIP5/PPE) and link these to the occurrence of regional extreme wind and precipitation events. For the Impact domain, we will identify damage-relevant extreme wind events with the identification algorithm developed by Leckebusch et al. (2008) shown to provide suitable measures of potential damage out of single events for ETC and TC environments. With respect to uncertainties of impacts in the hydrological sphere, FOREX will apply state-of-the-art flood related indices of extreme precipitation to be derived from AOGCM model output (e.g., Sillmann et al. (2013a,b); Donat et al., 2014).
The development of comprehensive uncertainty information needs to consider the reasons for the uncertainty in information of potential future changes (scenario uncertainty, model uncertainty, method uncertainty, and process validation uncertainty). In the approach developed in FOREX all four of these uncertainties will be accounted for separately as well as in a convolved approach. The core of the methodology is the establishment of multi-model-sub-ensembles, as previously introduced by the applicants for extreme damage associated with ETCs (Donat et al., 2011) as the Multi-Model Combinatorics Approach (MMCA). This approach will be further developed and targeted to address the main objectives under investigation:
- evaluating the added value of incorporating alternative model ensembles in addition to CMIP5,
- comparing ranges of future changes between emission- and concentration-driven simulations,
- how to use historical observations to constrain the range of future projections?
The MMCA takes into account that multi-model ensemble (MME) studies are inherently affected by a certain level of arbitrariness. For example, the construction of an ensemble is determined by the availability of model simulations. As the individual simulations produce Anthropogenic Climate Change (ACC) signals with different magnitudes and even different signs the MME change will depend on the models included (cf. e.g., Donat et al., 2011).
In FOREX, the influence of different model combinations on the information of the physical hazard itself as well as on related impacts (e.g. loss potentials) is investigated systematically. The range of the resulting ACC signals, depending on the choice of the sub-ensembles, will allow us to draw conclusions on the effect of different sub-ensemble combinations and thus on the question, can alternative ensembles provide a better view of future risks. Additionally, this allows an estimation of the (un-)certainty of projected future changes. The idea behind this approach is to use all information included in the MME. The consideration of all possible model combinations can be seen as an “ensemble of (sub-) ensembles”. Basically, this idea of assessing the uncertainty of the projected changes is related to the principle of bootstrapping (Efron, 1979), which is applied to gain information about characteristics (e.g., quantiles) of an unknown theoretical distribution. However, whereas bootstrapping generally considers different limited samples to assess the characteristics of a basic population, the approach presented here accounts for all possible solutions that can be constructed from the ensemble of models.
Variables to be investigated by the targeted MMCA are the local wind speed and precipitation extreme characteristics (e.g., derived from Extreme Value Theory: percentile changes, return periods, intensity levels), the intensity-occurrence characteristics of the steering weather systems (Frontal structures, ETC, TC) themselves, and their links and dependencies to large-scale forcing mechanisms. The latter is of crucial importance for the development of constraints on the range of future projections. Investigation into the physical mechanisms forcing time-depending variability of the occurrence of extremes will increase our knowledge of which parts (sub-ensembles) of the targeted MMCA will be more reliable than others (process validation).
Processes diagnosed in observational datasets such as re-analyses (e.g., ERA-Interim, ERA20C, ERA-SAT, NCEP20CR, NCEPCFSR) and their representation in different available ensemble members, will allow for the assessment of potential predictability, which can be tested in the respective model suites (for seasonal-to-decadal timescales as well as long-term projections), to establish real predictability. Existing experiences from the German MiKliP project and collaboration with the MetOffice with respect to seasonal forecast skill of extremes will be used to establish strong links to WP4.2 (5-40 year time scale) in CSSP.
FOREXwill convolve various sources of uncertainty in one confidence statement such as demonstrated in the recently published work by the applicants in Held et al. (2013), taking into account different reasons for uncertainties (e.g., model uncertainties, scenario uncertainties, method uncertainty, impact uncertainties). The final design of the robust uncertainty estimates, securing the highest applicability of the results, will be developed in conjunction with partners from other WPs and regional China collaborators.
Dr GC Leckebusch