**1. Introduction**

Climate change is expected to affect society substantially in terms of increased temperatures and precipitation, as well as more frequent weather extremes [1,2]. Sophisticated climate-change projections were carried out for many countries including the Netherlands. However, these projections mostly focus on rural areas, while policy makers are also interested in how they can make future cities more heat-resilient. The Netherlands has the challenge of realizing one million new houses by the year 2040, which mainly need to be created within the existing built environment. This study aims to clarify how climate change evolves in Dutch cities, and which urban morphological factors need to be modified in order to reduce heat stress.

Temperatures in cities are typically significantly higher for most of the day compared to their rural surroundings [3]. The heat stress due to these higher temperatures has a significant impact on health [4,5]. These elevated urban temperatures are known as the urban heat island (UHI), which is defined as the difference in 2 m of air temperature difference between a city and the surrounding countryside. Many factors contribute to a warmer urban environment. In general, the radiation balance of a city is altered due to its complex canyon structure, which results in a relatively small albedo, and thus, a large energy gain for the urban surface [6]. Furthermore, the energy balance of urban areas is different from rural areas due to the lower sky-view factor [3,7], enhanced impervious surfaces (initiating a higher sensible heat flux at the expense of the latent heat flux), and anthropogenic heat release [8]. Because of the relatively large heat conductivity and thermal capacity of the urban fabric, cities can store more heat during the daytime than rural environments. This energy is released after sunset creating the largest UHI in the evening or onset of the night. The elevated temperatures in urban areas affect human health and are a significant contributor to the excess mortality observed during heat waves [4,5]. High night-time temperatures have adverse effects on human health by preventing an adequate recovery from daytime heat and by inducing sleep deprivation [9]. The best known examples for western Europe were the high number of excess deaths in France during the heat waves of 2003 and 2006 [10,11]. In the Netherlands, the number of excess deaths was estimated to be between 1400 and 2200 in the summer of 2003 [12].

The UHI is difficult to quantify directly with observations, since the availability of long-term climatologic observational data in cities is limited. This is because official weather stations are preferably located in open rural or natural areas with the aim of obtaining a substantial spatial representativeness [13]. Consequently, long-term climatological observations to monitor urban heat are rarely available for cities. Other observational sources from urban areas such as amateur weather station data or field campaign data generally do not cover long periods, and the data often represent only small areas due to the high variability in temperature within the urban area setting [14,15]. To overcome this lack of observations, the UHI is often modeled using sophisticated atmospheric mesoscale models coupled to an urban canopy [16–19]. However, those models require large amounts of initial urban morphological data, which are often inaccurate and sometimes not available [20]. Another approach is to measure the UHI with remote sensing techniques [21,22]. However, correlations between remote-sensing-derived surface temperatures and air temperatures are weak, due to the absence of advection and flux divergence in the air volume [21]. In addition, the intervals of the revisiting times restrict a dynamical analysis of UHIs [22].

As an alternative approach, in this study, we diagnosed the daily maximum UHI by applying an empirical formula designed by Theeuwes et al. (2017) [20]. With this approach, we calculated urban minimum temperatures and compared them with rural minimum temperatures. The data needed in this equation consist of routine meteorological observations and morphological city characteristics. A novel concept is the application of this equation to determine the UHI for current and future climates in an urban agglomeration. As a test bed, we computed the UHI for the agglomeration of The Hague in the Netherlands (Figure 1). Previous research shows that cities of varying size in a maritime climate such as the Netherlands may experience severe heat stress [14]. This also holds for near-coastal cities like Rotterdam where nocturnal UHIs are up to 7 ◦C during clear heat-wave days [23]. Molenaar et al. [24] estimated that the number of hours of experienced heat stress, here defined as the hours with physiologically equivalent temperatures above 23 ◦C, will double from the current average of about 250 hours per year to slightly more than 500 hours per year under the warmest Royal Netherlands Meteorological Institute (KNMI) climate scenario.

Also a novel concept, we investigated the impact of construction plans on the magnitude of the UHI. In the future, the Dutch government will face the challenge of building an additional one million residences before the year 2040 [25]. A substantial part of this new construction has to be developed within existing urban areas. This means that the population density will increase, which will apply pressure to the heat resiliency of cities. For instance, Steeneveld et al. [14] found a good correlation

between the higher percentiles of UHI and population densities for Dutch cities of various sizes. Within the current research, we investigated the effect of different urban planning strategies on heat stress for current and future climate scenarios by creating heat maps with a resolution of 100 m. The urban planning strategies refer to the redevelopment and the construction of additional residences within the urban neighborhoods. The construction differs between high-rise buildings and low- and mid-rise buildings. The choice has consequences for the urban morphological characteristics of an urban area, as vegetation fractions and sky-view factors are important parameters in determining the UHI. The aim of this study was to estimate urban night-time temperatures for current and future climate in order to obtain measures which reduce the UHI.

This paper is organized as follows: Section 2 summarizes the modeling approach and introduces the climate scenarios and urban planning strategies we explored. Section 3 presents the results we obtained. Finally, in Sections 4 and 5, we reflect on our modeling approach and outcomes, and we draw conclusions.
