The effect of the CORINE land cover class application at mid-resolution WRF simulations on urban heat island

Abstract

<p>The CORINE land cover dataset provides a more realistic dataset for various scientific applications. In this study we incorporated almost all (38 instead of the original 44 classes, as water classes are not dividable in the WRF routines) CORINE classes to the WRF model and created 1-year long 5 km resolution simulations to check its effects. The simulated area covers the Central European region.</p><p>On a closer look, compared to the USGS dataset the first observable differences are the decreased number of non-irrigated croplands that are described as forests. The secondary difference comes in the coverage of urban areas. In some cases, a city previously covered by 2 grid points are now covered by 20, which causes a more pronounced urban heat island (UHI) effect.</p><p>Our goal is to analyse the effect of land cover differences on urban heat island intensity by only relying on the NoahMP provided heat flux calculations. For regional climate simulations the application of urban parameterisation could be more computationally demanding. Therefore, we try to assess the feasibility of UHI analysis at such a scale without urban parameterisation.</p><p>Urban areas are selected similarly to radar cell tracking, fitting a circle around the urban land cover classes and the rural regions are selected from a double size circle radius if there are no large altitude differences between the urban average altitude and the rural grid point. According to the preliminary results an annual average 1.5 °C UHI can be simulated without urban parameterisation. Principal component analysis shows that the main driver of the UHI is the annual variation of leaf area index in the rural regions. The secondary driver is either the precipitation or the snow cover, depending on the spring and wintertime weather.</p>