EMS Annual Meeting Abstracts
Vol. 20, EMS2023-47, 2023, updated on 06 Jul 2023
EMS Annual Meeting 2023
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Deriving point-scale climate data for arbitrary locations – The “station transfer” method and an example application

Christoph Frei, Michael Begert, and Francesco A. Isotta
Christoph Frei et al.
  • Federal Office of Meteorology and Climatology, MeteoSwiss, Zürich-Airport, Switzerland (christoph.frei@meteoswiss.ch)

Many planning applications require statistical information about the climate at the scale of a point. In civil engineering, for example, the design of heating and cooling systems relies on estimates of the occurrence of cold/warm extremes at the location of the building. An emerging practice of climate service providers is to derive such information from interpolation-based or reanalysis-based climate datasets on a fine (km-scale) grid. Yet, such datasets are poorly suited for this purpose. The effective resolution is limited and, as a consequence, this data represents regional area-mean conditions, not the point scale. As a result, the frequency of extremes is underestimated, and cross-parameter relationships are compromised. Here, we propose and illustrate a statistical procedure to derive point-scale climate data, at any point in space, without the artefacts of traditional optimal-prediction-based interpolation.

The proposed method, denoted “station transfer”, derives a climate series for the target location by transferring the observations of a suitable station and applying an adjustment, so that the result is representative for the target location. The choice of station and the adjustment are fully data driven. Unlike conventional interpolation, where concomitant observations in the neighborhood are combined, the “station transfer” insists on a single station as the origin of an adjustment, which is less invasive and better preserves point-scale tail statistics and cross-parameter relationships. We introduce a stochastic model for the transfer where the adjustments and the areas of representativity are jointly estimated from the entire station network. The method is illustrated with an example application that predicts point-scale surface air temperatures over the territory of Switzerland, using data from 65 stations.

The development and experimental application of this presentation are preliminaries of an upcoming project, where MeteoSwiss will derive new building-design climate basics for the Swiss Society of Engineers and Architects. But “station transfer” may become a much more widely used complement to conventional gridding, because of the wide-spread request for point-scale and high-frequency climate information in civil planning. 

How to cite: Frei, C., Begert, M., and Isotta, F. A.: Deriving point-scale climate data for arbitrary locations – The “station transfer” method and an example application, EMS Annual Meeting 2023, Bratislava, Slovakia, 4–8 Sep 2023, EMS2023-47, https://doi.org/10.5194/ems2023-47, 2023.