Improving the urban surface representation in high-resolution reanalyses frameworks

In this work, we aim to include a computationally inexpensive enhanced urban and land surface representation in a reanalyses framework. First, we update the land use representation used in the ICON-LAM numerical weather prediction by employing the 100m CORINE data set with three urban land use classes instead of the older 300m GLOBCOVER with only one urban category. Then, we include an energetic surface modification for complex urban fabric as an increase in the heat capacity and area sensible heat flux.

Further, we use the Copernicus Land Surface Temperature (LST) satellite product available at 5-km 1-hour resolution in the context of data assimilation to correct surface temperatures. Specifically, we investigate the sensitivity on parameter settings within the LETKF-KENDA data assimilation scheme, i.e., horizontal localization length, adaptive inflation, coarsening factor, thinning, and different observational errors.
The latter two are of particular importance since we deal with spatially dense data, whose errors are spatially correlated. Thus, we include the observation error correlation in the KENDA scheme and compare the results to the standard version.

The results show that the LST assimilation improves the land temperature biases drastically, e.g. down from over 7K to 2.5K for the Berlin area. At the same time, it also seems to cause a degradation in 2-meter temperature biases. However, this effect seems to be related to their model equivalent calculation (especially during daytime) rather than the observation impact itself. The analysis increments do not exhibit a similar behavior across various cities in the Central European area.

The urban correction improves the 2-meter temperature representation, especially in the morning hours. We do not find the combined changes in surface temperatures due to LST assimilation and urban correction to drastically alter the Urban Heat Island representation for the two case study cities (Berlin and Cologne).