Towards sectoral and standardised vulnerability assessments: the example of heatwave impacts on human health | Umweltbundesamt

"Under climate conditions as projected by the model STAR, the mean values of impacts for NRW municipalities amount to 0.33, while the mean impacts for CCLM are slightly higher (0.36). Overall impacts are higher with CCLM assumptions, however the increase in impacts in relation to the baseline period is higher for the STAR model with 0.3 (0.25 with CCLM). Table 1 summarizes the results for all municipalities for both models. Under assumptions of the STAR model 13% of all municipalities have high to very high impacts, 71% display low to very low impacts while the remaining 16% have a medium level of impacts. In contrast to that, the results for the CCLM model classify 19% of all municipalities as highly to very highly impacted. A proportion of 69% falls into the classes low and very low with the remaining 12% displaying medium impacts. For a total of 274 municipalities the intensity of impacts is identical for both climate models; in 90 cases the results for the CCLM model are higher while in 32 cases the STAR model produces higher degrees of vulnerability. Altogether more municipalities are ranked as highly or very highly impacted under CCLM assumptions (78 municipalities). The smaller share of highly or very highly impacted municipalities under STAR (56 municipalities) correspond to those under CCLM except for one of the municipalities (Hagen), where impacts are projected under the model STAR and not under CCLM. Our results show an overall increase in impacts over time, especially so in the densely populated Ruhr area and in parts of the Rhine valley [...]. The spatial distribution of impacts as well as changes between the analysis periods show a distinct spatial pattern, which is apparent in the results of both climate models. With some exceptions, low-lying and densely populated areas are most vulnerable. A general gradient of impacts is apparent from densely populated areas towards those less densely populated, with the mountainous regions displaying the lowest levels of impacts. The metropolitan region is distinguishable, with higher levels of impacts compared to surrounding areas, however the southern municipalities along the river Rhine are distinctly less affected. The large cities of Cologne and Düsseldorf, for example, though very densely populated, show low impacts. Clusters of high to very high impacts emerge in the metropolitan Ruhr region, as well as in the North-East around the city of Bielefeld under both climate models. With exception of the city of Münster the northern lowlands exhibit very low levels of impacts. The mountainous areas in the southwest of the state constitute regions of very low impacts across both data sets, while dense settlements along the foothill regions of both Sauerland and Eifel, such as Aachen and Wuppertal display elevated
levels of impacts." (Lissner, Holsten, Walther, Kropp 2011: 11 f.)
"The cause-and-effect relationships translated into a quantitative representation via fuzzy logic are clearly determinable within our results. The maps in the decision tree [...] exemplify this using scenario data obtained from the STAR model. The resulting impacts can be traced back to the input values along the graph. In the cities of Cologne and Düsseldorf, for example, the low impacts can be clearly ascribed to the very low percentage of population ≥ 65 (0.2). While the potential UHI is very high, there are only very low numbers of especially sensitive age groups within both analysis periods. All other input variables here are very high, however, the minor number of elderly as an especially sensitive age group significantly reduces impacts. Due to the γ -value, the results show a slightly augmented sensitivity and consequent impacts of 0.36 (low), thus taking into account the excess heat exposure through the UHI. Though the exposure is very high in this region, the low sensitivity determines the outcome. In the case of Dortmund, where a medium number of elderly population is present, the intense UHI leads to a high local sensitivity, again exemplifying the effect of the applied γ -value: without introducing this compensation, the UHI would not be accounted for and consequent impacts would be lower. In both of these examples the results correspond for both applied climate models. The example of Münster demonstrates that the area of sealed surfaces reduces the potential UHI, even though there is a high population density. With additional lower levels of population ≥ 65 the sensitivity as well as the impacts remain at a medium level with both climate models. In municipalities in the mountainous Sauerland region, for example in Winterberg, a very high proportion of sensitive elderly population coincides with very low values for all other variables, resulting in very low levels of impacts. In several municipalities significant differences between the applied climate models can be observed in the results, as the models show different spatial characteristics (Fig. 6). Gevelsberg, situated within the metropolitan region, shows the highest difference between the models: impacts are very high under the CCLM model with a value of 1 and medium under STAR assumptions with a value of 0.4. This is due to the different projections of HWD, with STAR projecting less HWD than CCLM under the A1B emission scenario. Similarly, for the city of Wuppertal, also situated in the metropolitan region, the number of HWD determine the level of impacts with very high levels for CCLM (0.81) and medium levels for STAR (0.45). The high percentage of sensitive age groups in the region of Rheine in the north of NRW entails medium levels of sensitivity. With input from the STAR model the results show medium impacts as there are significant numbers of HWD projected for the region. Results from the CCLM model, however, project a low number of HWD, consequently, the impacts are low." (Lissner, Holsten, Walther, Kropp 2011: 12 f.)