Semi-empirical modeling of urban heat island using climate classification and mobile measurements

According to climate change scenarios, frequency and intensity of heat waves are likely to increase in Europe during the twenty first century. Heat wave weather configuration, which exhibits notably low wind speed and low nebulosity, tends to favor the urban heat island phenomenon in urban context. The urban heat island effect, which is commonly defined as the synchronous screen-height air temperature difference between an urban area and a rural area nearby, impacts public health, energy consumption and outdoor thermal comfort. Therefore, this phenomenon needs to be taken into account during urban planning processes. However, while improvements of physically-based urban surface models lead to more accurate urban microclimate representation, decision support tools are needed to provide information on urban heat island for practical applications.

Existing simple schemes have studied the relationships between urban characteristics and urban heat island magnitude. However, these models are usually designed for one specific horizontal scale. Moreover, the information they provide may not consistently fit with the operational needs that urban planners meet. Besides of modeling issues, the above-mentioned definition of the urban heat island amplitude could be questioned. Indeed, the use of the adjectives « urban » and « rural » may be misleading, because they are generic and thus they can be interpreted differently regarding cultural background. In addition, these adjectives may be too simple to describe the complex layout of cities.

This work aims to introduce the first version of a semi-empirical climatic model, which targets two objectives, namely i) to model screen-height air temperature distribution at mesoclimatic scale and at microclimatic scale and ii) to be a decision support tool for urban planners, providing accurate and relevant information about urban heat island amplitude and allowing the test of urban heat island mitigation solutions. In order to perform this multi-scale study, the modeling approach adopts a climate classification developed by Stewart and called Local Climate Zone (LCZ). This classification possesses 17 different LCZ types, both urban and rural types. The methodological approach corresponding to this classification scheme consists in identify existing urban areas that correspond to a specific type of LCZ. A LCZ is defined as an urban area with a minimum diameter of 400 metres which demonstrates both i) a characteristic screen-height temperature regime and ii) uniform features in terms of urban morphology, land use, urban material and urban metabolism. One of the major advantages of this scheme is the possibility to redefine urban heat island magnitude in a more explicit and standardized way. Indeed, urban heat island amplitude can be defined as the air temperature difference between two specific LCZ that are significantly different in terms of morphology and land cover.

In order to elaborate this model, Stewart's classification has been applied in the Great Nancy Area, France. First, LCZ have been built using satellite data and GIS, and a map has been produced for the area of interest. Indicators regarding urban morphology, land use, urban material and urban metabolism have been calculated. Then mobile measurements have been planed to record the screen-height temperature regime. Routes have been designed to survey different LCZ types – urbanized or not. Traverses have been performed in the Great Nancy Area both during daytime and nighttime for a range of cloud cover and wind conditions. Data have been gathered at a three meters distance step by an instrumented vehicle. Measurements results exhibit a nocturnal mean air temperature difference of 2.0°C between LCZ 2 and LCZ 6 (calculated on 19 traverses), and a nocturnal mean air temperature difference of 4.4°C between LCZ 2 and LCZ D (calculated on 12 traverses). They also reveal that if some LCZ are homogeneous in terms of spatial distribution of screen-height air temperature, other LCZ are more heterogeneous.

The model have been developed based on mobile measurements campaigns and on LCZ map of Nancy. At mesoclimatic scale, each LCZ is considered as a simple element, and the Great Nancy Area is viewed as an assembly of contiguous LCZ. Each of them is associated with a specific air temperature pattern. The model uses downward radiative flux, wind speed and geographical data in order to deliver information about the evolution of the average screen-height air temperature. In addition, the analysis of the air temperature spatial disparities inside LCZ highlights possible positive or negative thermal anomalies at microclimatic scale.