Targeted data selection for different urban climate analyses

Urban climate analyses have a wide range of different targets, including average values of different meteorological variables, percentiles and number of exceedances of certain threshold values. At least 30 years of data are required for this type of analysis. For large scale analysis model results of several ten kilometres resolution are either available for 30 years and more, or they may be calculated with a coarse resolution model in some foreseeable time. However, the urban effects may not present in these type of data. In any case, the coarse resolution data do not provide any inner-urban differences. These can only be found in model results of a resolution of well below 1 km. When aiming at differences in street canyons or next to buildings, the resolution should be very high and not exceed several meters. The high-resolution models that may be used to produce these type of data in a dynamic simulation are very resources consuming and the waiting time is enormous. A dynamic simulation of 30 years is much too time consuming. Therefore, only a few selected meteorological situations are dynamically calculated. These must be selected correctly in order to support the objective of the urban climate analysis. Average temperatures can then be derived by selecting several meteorological situations and using a statistical-dynamical downscaling approach. If, for example, extremes in meteorological variables are to be determined in the urban area (e.g. temperature or wind speed or humidity or radiation) the relevant meteorological situation can be determined from statistics of coarse resolution data. This is, however, not easy with a target such as the extreme intensity of the urban heat island, which depends on temperature, wind speed, humidity and radiation together. Here the combination of the variables that determines the extreme intensity must be taken into account.

For different temperature related variables and for determining the situation that leads to an extreme intensity of the urban heat island, the presentation will show ways for a targeted selection of the meteorological situations that are to be simulated with high-resolution models.