EGU2020-1352
https://doi.org/10.5194/egusphere-egu2020-1352
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Representing the Urban Heat Island Effect in Future Climates

Annkatrin Burgstall1, Ana Casanueva2, Elke Hertig3, Erich Fischer4, Reto Knutti4, and Sven Kotlarski1
Annkatrin Burgstall et al.
  • 1Federal Office of Meteorology and Climatology MeteoSwiss, Zurich, Switzerland (annkatrin.burgstall@meteoswiss.ch)
  • 2Department of Applied Mathematics and Computer Science, University of Cantabria, Santander, Spain
  • 3Faculty of Medicine, University of Augsburg, Augsburg, Germany
  • 4Institute of Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland

An increasing fraction of people living in urban areas and the expected increase in long lasting heat waves highlight the important role of urban climates in terms of future climate change impacts, especially with relation to the heat-health sector. Due to the urban heat island (UHI) effect and its (generally) increased intensity particularly during nighttime, people living in urban areas happen to be more affected by heat-related discomfort and health risks than those in non-urban regions. In this contribution, temperatures of both rural and urban sites (station couples) in Switzerland and Southern Germany are analyzed, using (i) observed as well as (ii) bias-corrected and downscaled climate model data for daily minimum (tmin) and daily maximum temperature (tmax) to account for the UHI in future climates. As meteorological data are often restricted to locations of long-term measurements at rural sites only, they need to be transferred to urban sites first. For this purpose, the well-established quantile mapping technique (QM) is tested in a two-step manner. The resulting products are urban time series at daily resolution for tmin and tmax. By analyzing the temperature differences of the observed climate at rural sites and their respective urban counterparts and by assuming a stationary relationship between both, we can represent the UHI in future climates, which is quantified in terms of heat indices based on tmin and tmax (tropical nights, summer days, hot days).

The QM performance is evaluated using long-term weather station data of a Zurich station couple in a comprehensive cross-validation framework. Results reveal a promising performance in the present-day climate, given very low biases in the validation.

Applying the proposed method to the employed station couples, projections indicate distinct urban-rural temperature differences (UHI) during nighttime (considering the frequency of tropical nights based on tmin) compared to weak differences during the day (considering the frequency of summer days and hot days based on tmax). Moreover, scenarios suggest the frequency of all indices to dramatically rise at the urban site by the end of the century under a strong emission scenario (RCP8.5): compared to the rural site, the number of tropical nights almost doubles while the number of summer days reveals about 15% more days at the urban site when focusing on the station couple in Zurich and the late scenario period. The lack of nighttime relief, indicated by tmin not falling below 20°C (i.e. a tropical night), is especially problematic in terms of human health and makes the study of the urban climate in general and the UHI effect in particular indispensable.

How to cite: Burgstall, A., Casanueva, A., Hertig, E., Fischer, E., Knutti, R., and Kotlarski, S.: Representing the Urban Heat Island Effect in Future Climates, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-1352, https://doi.org/10.5194/egusphere-egu2020-1352, 2019

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 01 May 2020
  • CC1: Comment on EGU2020-1352, Clemens Schwingshackl, 05 May 2020

    Thank you for this very nice and interesting presentation, Annkatrin! I was wondering whether you have only applied traditional QM or also tested other methods such as quantile delta mapping (QDM). QM might sometimes give artificial trends (Maraun 2016, doi:10.1007/s40641-016-0050-x) and, thus, for studying extreme temperatures more advanced methods such as QDM might be better suited because they consider that with climate change temperatures might move beyond the observed range.

    • AC1: Reply to CC1, Annkatrin Burgstall, 05 May 2020

      Hi Clemens, thanks for your comment.

      The Quantile Mapping (QM) implementation we used is a variant of Empirical Quantile Mapping and based on the work of Rajczak et al. (2016, https://doi.org/10.1002/joc.4417) and Ivanov and Kotlarski (2017, https://doi.org/10.1002/joc.4870). It is based on 99 percentiles (1st to 99th percentile) of daily data and a linear interpolation for values between the two percentiles. For values outside the calibration range (i.e. smaller than the first and larger than the last percentile), the correction function (determined separately for each DOY with a moving window of 91 days) of the first and of the last percentile is applied, respectively. The treatment of these values is definitely a limitation of the QM implementation we employed. Thanks a lot for your suggestion to also consider other methods such as quantile delta mapping (QDM), which we surely keep in mind for further analyses.