Advertisement

Environment Systems and Decisions

, Volume 38, Issue 2, pp 261–273 | Cite as

Numerical simulations to quantify the diurnal contrast in local climate trend induced by desert urbanization

  • Samy Kamal
  • Huei-Ping Huang
  • Soe W. Myint
Article

Abstract

The effect of urbanization on local climate is quantified by numerical simulations for five desert cities that represent a wide range of urban size, climate zone, and composition of land cover. Land-use land cover maps generated from Landsat data for 1985 and 2010, chosen as the start and end of a period of rapid urbanization, are used to constrain the surface boundary conditions for the numerical model. In this manner, this study focuses on the particular aspect of the effect of land-use changes on local climate. Within this scope, the results reveal a pattern of the climatic effect of desert urbanization with nighttime warming and weaker, but significant daytime cooling. This effect is confined to the urban area and is not sensitive to the size of the city or the detailed land cover types in the surrounding areas. The pattern is identified in both winter and summer. Exceptions to this pattern are found in a small number of cases when the noisiness of local circulation, specifically monsoon and land–sea breeze, overwhelms the climatic signal induced by land-use changes. The inter-cities’ differences in the temperature response to land-use change are also discussed.

Keywords

Urbanization Land-use change Sustainability Numerical modeling 

1 Introduction

The effect of land-use changes on climate has recently emerged as one of the major issues for environmental prediction and protection (e.g., Diffenbaugh 2009; Pielke et al. 2011; National Research Council 2012). As a special category of land-use change, urbanization over desert cities is of particular interest because it occurs over areas under a high level of environmental stress (UNEP 2006). Classical studies on the climatic effect of urbanization, especially the well-known paradigm of urban heat island (UHI, e.g., Oke 1982) which emphasizes nighttime warming, did not specifically consider the scenario of urban development over desert or semiarid areas. Existing studies that considered urbanization of desert cities have so far examined only a small number of cities (e.g., Brazel et al. 2000, 2007; Georgescu et al. 2009, 2011; Myint et al. 2013; Zheng et al. 2014 for Phoenix; Miller 2011 and Kamal et al. 2015 for Las Vegas, Lazzarini et al. 2013 for Abu Dhabi; Rasul et al. 2015 for Erbil). It remains to be determined whether the climatic effects of urbanization found in those case studies are universal or how they vary across space for desert cities. This study aims to advance the understanding on this issue with numerical simulations for five desert cities using the land-use maps from two contrasting eras separated by rapid urban developments.

The aforementioned studies on desert cities hint at a potentially common feature, namely, a weak daytime cooling (relative to the surrounding rural area) due to urbanization. In general, this feature is not unique to desert cities; daytime cooling was also found in some midlatitude non-desert cities such as Vancouver (Runnalls and Oke 2000), Indianapolis (Carnahan and Larson 1990), and Adelaide (Erell and Williamson 2007). Erell and Williamson (2007) noted that an increase in effective surface area (and effective thermal mass) due to urbanization acts to damp the diurnal cycle of temperature. If, for a particular city, this daytime cooling effect overwhelms other warming effects, a net daytime cooling could be observed. For desert cities, Kamal et al. (2015) suggest that there are common mechanisms that support daytime cooling by urbanization. First, urban-type impervious surfaces generally have a higher infrared emissivity than the background surfaces of desert or shrubland. Secondly, related to the argument of Erell and Williamson (2007), the build-up of urban structures helps increase the shadow effect and the effective surface area for infrared emission as compared to a flat surface covered by the same material. Thirdly, while the first two effects already favor daytime cooling, for desert cities there is no compensating warming effect produced by a decrease in evaporation which is common to non-desert cities (for which urbanization means replacing high vegetation coverage with urban impervious materials). Kamal et al. (2015) based their discussions on a set of numerical simulations for Las Vegas. Given that desert cities in the world vary widely in size and in the pre-urbanization land surface types, in this paper we will take a broader view to examine the robustness of daytime cooling in five desert cities with a wide range of size, location, and baseline climate.

Determining the climatic effect of urbanization for desert cities is challenging due to the lack of adequate observational data. Except for a few mega cities such as Phoenix and Las Vegas, the spatial and temporal coverage of meteorological observation over desert cities is poor. As an alternative, this work adopts the approach of numerical simulation using a meteorological model, the Weather Research and Forecasting (WRF) model (Skamarock et al. 2008), constrained by imposed pre- or post-urbanization land surface condition. Our confidence in the model lies in the fact that it has been cross-validated with observations in the aforementioned studies for Phoenix and Las Vegas. An example of model validation is given in Sect. 2.4.

The five cities chosen are (1) Las Vegas, USA, (2) Hotan, China, (3) Kharga, Egypt, (4) Beer Sheva, Israel, and (5) Jodhpur, India. Figure 1a–c shows the locations and the corresponding model domains for those cities. The details of the land-use data and the setup of numerical model are described in Sect. 2. Results from the simulations for both winter and summer, and day and night, are presented in Sect. 3 and further discussed in Sect. 4. Concluding remarks follow in Sect. 5.
Fig. 1

Locations and model domains for the five cities: a Beer Sheva, Israel (red) and Kharga, Egypt (blue). b Hotan, China (red) and Jodhpur, India (blue). c Las Vegas, USA. Three nested domains are used for each city. The urban area of each city is located within the innermost domain in each case. Land area is colored in brown

2 Methodology and data

2.1 Numerical model

Given the complexity of the many physical processes associated with urbanization and its climatic effects, we do not expect one model to cover all processes across multiple spatial and temporal scales. Instead, this study focuses on using a mesoscale meteorological model to quantify the response of near-surface air temperature to the long-term conversion from non-urban to urban type of land cover. It is understood that there are other urban effects, for example those related to anthropogenic heat and dust generation, which are not explicitly simulated by our model.

The core of the meteorological model is a set of numerical schemes for advancing the atmospheric variables in time. A short-term run would be similar to a typical weather prediction. For our purpose, the model is run for at least a season to produce the climate. The model is coupled to a module for surface energy balance and atmosphere–land exchanges, in which the surface fluxes (for sensible and latent heat, and radiative energy) are calculated. Those fluxes are incorporated into the process of forward integration of the atmospheric variables. In the procedure, land-use changes influence atmospheric temperature by altering the surface energy balance (e.g., Kamal et al. 2015), modifying the surface fluxes that feedback to the atmosphere. A schematic diagram that illustrates the key procedures in the model is shown in Fig. 2. For the analysis of model output, we focus particularly on 2 m air temperature, as is standard for many climate applications.
Fig. 2

A schematic diagram for the key procedures in the numerical simulation. The time integration for updating atmospheric variables is in the dashed box. The gray arrows indicate the parameterized module for calculating the surface fluxes, through which the influence of land-use changes enters the main procedure

The Weather Research and Forecasting (WRF) model (Skamarock et al. 2008) version 3.3.1 is used for the numerical simulations. We adopt the approach of dynamical downscaling (e.g., Leung et al. 2003; Mearns et al. 2012), namely running the model with imposed, time-varying, lateral boundary conditions assembled from large-scale observations. With multiple layers of nesting, a higher resolution can be used for the innermost model domain that encompasses the urban area of interest.

The majority of existing regional climate simulations use 15–30-km horizontal resolutions (e.g., Caldwell et al. 2008; Heikkila et al. 2011; Pan et al. 2011). For our purpose, a higher resolution is needed for the innermost model domain that covers the urban area. Three nested domains are used for each city, as shown in Fig. 1a–c. The cases for all cities except Las Vegas use 1-, 5-, and 25-km horizontal resolution for the innermost, intermediate, and outermost domains. A slightly coarser resolution of (3, 12, 48 km) is chosen for Las Vegas because it has a much larger urban area. These choices of horizontal resolution are comparable to those used by recent numerical studies on the climatic effect of urbanization (e.g., Kusaka et al. 2012; Georgescu et al. 2011; Kamal et al. 2015). All of our simulations use 28 vertical levels with the model top set to 50 hPa.

For non-urban processes, we adopt the physical parameterization schemes in WRF similar to that used by Kamal et al. (2015). The Kain–Frisch cumulus parameterization scheme is turned on for the outer and intermediate domains, but turned off for the innermost domain. (A discussion on this strategy can be found in Sharma and Huang 2012.) The treatments of urban processes rely on the urban canopy model (UCM) of Kusaka et al. (2001) and Kusaka and Kimura (2004), which is embedded in WRF through the Noah land surface model (Chen et al. 2006, 2011). The UCM includes a set of parameters that describe the geometry and thermal properties of urban landscape such as building height, roughness length above street canyon, urban fraction, and heat capacity of built structures. Since it is challenging to determine those parameters from archives of local governments of the five cities, we follow Kamal et al. (2015) by setting them to the default values which represent the average conditions of desert cities. For example, building height, urban fraction, and heat capacity of roof are set to 7.5 m, 0.9, and 1.0 × 106 J K−1m−3, respectively. While the UCM serves to support the calculations in the presence of extra urban-related parameters, the basic procedures for calculating the surface fluxes and land–atmosphere interaction are otherwise the same over urban and non-urban type of land.

An urban community consists of not only concrete and pavement, but also green areas such as city parks. While the local cooling due to greening is also potentially a significant factor in determining urban temperature in arid environment (e.g., Shashua-Bar et al. 2009), due to a lack of detailed data to constrain and incorporate such an effect in WRF model we did not include it in the numerical simulation. This point will be further discussed in Sect. 4.

2.2 Land-use data

For each city, two contrasting land-use maps are generated from Landsat satellite observations for 1985 and 2010, which represent the start and end of a period of rapid urban growth. The data for all five cities come from the same Landsat archive at USGS (landsat.usgs.gov). The five desert cities are selected as they are connected to adjacent periurban areas (urban fringe–desert interface) and their surrounding arid/semiarid environments, and they are subject to strong urbanization pressure in desert environments. The selection of the cities is further guided by: (1) significant population (> 100,000) at present day, (2) availability of cloud-free Landsat images, (3) sufficient separation between the urban area and the edge of the Landsat scene, and (4) avoidance of highly undulating terrain or mountainous landscape with elevations ± 100 m off the mean elevation of the urban core, so as to maintain consistent environmental conditions. For each city, two Landsat scenes covering 1985 and 2010 are processed. Landsat has six visible and near-infrared (VIS–NIR) bands with 30-m spatial resolution and one thermal band with 120-m resolution. All images were acquired at the correction level 1G, which includes geometric and radiometric corrections. Prior to image analysis, the VIS–NIR data were converted to reflectance following Chavez (1996), and the thermal band was converted to land surface temperature. We employed an object-based image analysis (OBIA) to classify each scene into 24 USGS Land-Use categories. They are in turn converted to the corresponding (and nearly identical) categories used in WRF. Since Landsat observations have a much higher resolution, a simple scheme of “choosing the majority” is adopted to weave multiple Landsat pixels into each grid box for WRF (Kamal et al. 2015).

Figure 3 summarizes the land-use maps on the WRF model grid for 1985 and 2010 for all five cities. Shown are the innermost model domains (see Fig. 1) for the respective cities. The major land-use categories that appear in Fig. 3 are summarized in Table 1. For our purpose, the default land-use maps in WRF are used for the intermediate and outermost domains. Although all five cities are located in desert regions, the detailed composition of land cover and the scenario of land-use change differ from city to city. For Kharga and Beer Sheva, urban land spreads from a concentrated small area in 1985 to multiple clusters in 2010. Hotan and Jodhpur are characterized by a large coverage of cropland, wetland, or mixed forest with the relatively small urban core (see inset for Hotan) embedded within it. This provides a contrast to Las Vegas which is surrounded mainly by shrubland. To incorporate the influence of the differing background land types, the innermost model domain is chosen to be significantly larger than the urban core itself.
Fig. 3

Land-use maps for the five cities in 1985 (left column) and 2010 (right column). From top to bottom: Las Vegas, Hotan, Kharga, Beer Sheva, and Jodhpur. Black indicates areas with urban build-up. The numbers in the color scale are the LU index in WRF as listed in Table 1. Latitude and longitude are marked at the margins of each map

Table 1

Major land-use categories for the five cities and their surrounding areas, and the corresponding LU indices in WRF

Land-use category name

LU index in WRF

Urban and build-up land

1

Dry cropland and pasture

2

Irrigated cropland and pasture

3

Shrubland

8

Evergreen broadleaf

13

Mixed forest

15

Water bodies

16

Wooden wetland

18

Barren or sparsely vegetated

19

Figure 4 summarizes land-use changes that occurred between 1985 and 2010 for the five cities. The most relevant categories of changes that involve urbanization (i.e., conversion of a non-urban type of land to urban land) are shown in red-hued colors. The remaining types of changes are shown in blue-green-hued colors. The major types of land-use changes that appear in Fig. 4 are summarized in Table 2. The detection of land-use changes is done by systematically screening and comparing the digitized data for the two eras for the same city.
Fig. 4

Maps of land-use changes that occurred between 1985 and 2010 for the five cities: a Las Vegas, b Hotan, c Kharga, d Beer Sheva, and e Jodhpur. White area is where no change occurred. The areas where urbanization occurred are shown in red-hued colors (category B–F in the color scale). Other types of land-use changes are shown in blue-green hued colors. A detailed description of the types of land-use changes is in Table 2. Latitude and longitude are marked at the margins of each panel

Table 2

Different types of land-use changes that occurred between 1985 and 2010 for the five cities and their surrounding areas

Type of land-use change

LU index in 1985

LU index in 2010

A

No change occurred

B

8

1

C

19

1

D

15

1

E

3

1

F

2

1

G

8

2

H

19

3

I

18

3

J

19

18

K

3

19

L

15

8

M

19

8

N

All other minor types of changes

The number in the second or third column is the LU index of WRF listed in Table 1. For example, type B is the conversion from shrubland in 1985 to urban land in 2010. The types of land-use changes related to urbanization are listed in boldface

The five cities differ significantly in size: The urban area of Las Vegas has a linear dimension of 50 km, while Kharga is only a few km across. The baseline climates of the five cities are significantly different: Hotan and Beer Sheva are continental and Mediterranean, respectively, while Jodhpur is influenced by Indian Monsoon in summer. Out of this wide range of background, we determine the common features of the climatic effect of urbanization.

2.3 Design of numerical experiment

To isolate the climatic effect of land-use changes, multiple sets of “twin experiments” are performed. Each pair of simulations is carried out by keeping all boundary conditions the same except changing the land surface boundary condition within the innermost model domain (as shown in Fig. 3) that covers the urban area. In this manner, the difference in the climate between the two runs can be attributed to the influence of land-use changes. The land-use maps for 1985 and 2010 are used to define the surface boundary conditions for the pair of runs. Otherwise, both runs are constrained in the fashion of climate downscaling by the same lateral boundary conditions. Specifically, four times daily meteorological variables from NCEP global analysis for 2010 are imposed to the lateral boundary of the outermost domain (see Fig. 1) through the duration of each run. By design, the “1985 run” is understood as not an attempt to fully reproduce the 1985 climatology, but as a perturbation to the 2010 run by only substituting the land surface boundary condition.

Since the baseline climatology differs significantly between winter and summer, two sets of twin experiments are performed for each city. The winter run is from October to January (preceded by a short spin-up period) using the lateral boundary condition from October 2009 to January 2010; summer run is from May to August and constrained by the lateral boundary condition from May 2010 to August 2010. Together, 20 seasonal runs (4 for each city) are performed with 123 days of simulation produced for each season and each city.

2.4 Model validation

A comprehensive validation for the WRF model at sub-mesoscale resolution and over desert cities remains challenging due to the lack of adequate observations. Among the five cities considered, only Las Vegas has some coverage of in situ observations by meteorological stations. Using an approach similar to Kamal et al. (2015), we choose the observations from three stations—McCarran Airport (36°05′N, 115°11′W), North Las Vegas Airport (36°12′N, 115°11′W), and Nellis Air Force Base (36°15′N, 115°02′W)—to cross-validate with WRF simulation. Figure 5 compares the 2 m air temperature as a function of the hour of the day from observation and simulation. Each curve of the diurnal cycle of temperature shown in Fig. 5 is averaged over 123 days from May to August 2010. The simulation captures the phase and amplitude of the observed diurnal cycle of temperature, but with a slight cold bias. However, the higher temperature from observation might be due in part to the fact that all three stations are located at airports with the land cover in surrounding area dominated by open concrete. Since the impact of urbanization will be extracted from the twin experiment by subtracting the “1985 run” from the “2010 run,” we expect this bias to further cancel out.
Fig. 5

Panels a, c, and e: The diurnal cycle of 2 m air temperature as a function of the time of the day from model simulation (red dashed) and observation (black solid). The locations are a North Las Vegas Airport, c Nellis Air Force Base, and e McCarran Airport. Each curve is constructed from the hourly observation or model output averaged over 123 days from May–August, 2010. Panels b, d, and f show the bias, defined as model minus observation, corresponding to panels (a), (c), and (e)

3 Results

3.1 Climatic effect at nighttime

Figure 6 shows the changes in 2 m air temperature in nighttime induced by land-use changes as deduced from the difference between the “2010 run” and “1985 run” for each pair of twin experiment. For brevity, shown are the difference maps at 2 AM local time which is representative of late night and early morning. The domains shown in Fig. 6 are the same as their counterparts in Fig. 4, i.e., they are the innermost domains for the WRF model.
Fig. 6

Change in 2 m air temperature at 2 AM local time, averaged over summer (left column) and winter (right column). Top to bottom: Las Vegas, Hotan, Kharga, Beer Sheva, and Jodhpur. The color scale, in °C, is shown at right. Latitude and longitude are marked at the margins of each panel

Nighttime warming akin to the classic UHI is found in all five cities over the grid points where urbanization occurred between 1985 and 2010. It is insensitive to the specific type of pre-urbanization land cover in 1985 from which the conversion to urban land occurred. For example, for Las Vegas a “ring” of elevated temperature is found to coincide with the new urban land converted primarily from the category of shrubland or barren. For Jodhpur, urban land emerged mainly at the expense of dry cropland and pasture. The feature of nighttime warming is qualitatively similar between winter and summer. The nighttime warming does not spread significantly beyond the vicinity of the new urban area, implying that the impact of urbanization on the 2 m air temperature is relatively local. The localness is related to the fact that, in the model simulation, the direct thermal response to land-use changes generally has a shallow vertical extent (e.g., Kamal et al. 2015). Since wind speed decreases toward the surface in the boundary layer, the shallow thermal response cannot be blown very far out of the location where land-use changes occur. The characteristic of nighttime warming is also not sensitive to the specific type of land cover in the surrounding areas of the urban core.

Although the cases with conversions between two non-urban land types are not of our main interest, they do occur at many grid points in the areas away from the cities, especially for Hotan. Unlike urban development, the non-urban types of land-use changes are not spatially concentrated, but are characterized by a complicated juxtaposition of multiple types of conversions. For Hotan, the areas with agricultural development (conversion from barren or wooden wetland to irrigated cropland) generally exhibit a nighttime cooling.

3.2 Climatic effect at daytime

Figure 7 shows the changes in 2 m air temperature at 2 PM, except for Beer Sheva for which 11 AM is chosen because the temperature response in the simulation peaks earlier in the day in the simulation. In contrast to UHI at nighttime, all cities exhibit daytime cooling over the grid points where urban land emerged between 1985 and 2010. Like the nighttime response, this feature is insensitive to the pre-urbanization land type and the size of the city (see the summary in Fig. 8). It is also found in both winter and summer.
Fig. 7

Change in 2 m air temperature at 2 PM (except 11 AM for Beer Sheva) local time, averaged over summer (left column) and winter (right column). Top to bottom: Las Vegas, Hotan, Kharga, Beer Sheva, and Jodhpur. The color scale, in °C, is shown at right. Latitude and longitude are marked at the margins of each panel

Fig. 8

A summary of the effect of urbanization on surface air temperature for four major types of conversions from non-urban to urban land cover. In each block with black border, the four squares represent four pre-urbanization land types of barren or sparsely vegetated (top left), shrubland (top right), dry cropland (lower left), and irrigated cropland (lower right) from which the urban land was developed. The four major columns from left to right are a daytime in summer, b nighttime in summer, c daytime in winter, and d nighttime in winter. The five major rows from top to bottom are the five cities as labeled at left. The area of the filled circle is proportional to the area where the specific type of land-use change occurred for the specific city. The largest circle (associated with Las Vegas) represents an area of 576 km2. The color scale, in °C, is shown at right

Over the non-urban areas in Beer Sheva and Jodhpur, some features of temperature changes shown in Fig. 7 cannot be clearly identified with the local land-use changes in Fig. 4. Those are meteorological noise due to the influence of regional circulation which overwhelms the climatic signal produced by land-use changes. For Jodhpur, the influence of Indian Monsoon in summer is hard to remove from the simulation for just one particular year. For Beer Sheva, with the proximity of the city to the Mediterranean Sea the urban effect is susceptive to the detail of land–sea breeze. Over those two cities, the strong regional circulation allows non-urban-related natural variability to be blown over the urban region and overlaps with the urban signal. The separation of the two would likely require longer integrations of the model and more sophisticated statistical analysis (e.g., by picking only the days with weak diurnal circulation to reduce non-urban influences), which is not pursued in this work. Since urbanization can also affect local circulation which in turn affects temperature (Kamal et al. 2015), a definitive quantification of the urban effect for these two cases would require using much longer simulations to potentially extract the signals in both temperature and wind. This remains a challenge due to the noisiness of wind field over urban areas (Kamal et al. 2015).

For the cities with relatively weak signals of urbanization-induced daytime cooling, such an effect could be more difficult to detect in observation since large-scale temperature trends (e.g., the well-known global warming associated with greenhouse gas forcing, IPCC 2013) and natural variability could overwhelm the urban signal in daytime. Moreover, enhanced nighttime warming due to the compound effect of urbanization and global warming could be partially carried over to daytime by the thermal inertia of the system. As an example, the observational study by Potcher and Itzhak Ben-Shalom (2013) indeed shows mainly a warming trend for Beer Sheva. It is useful to reiterate that this does not contradict our result from the twin simulations since they are designed to isolate the effect of urban land-use change while excluding other large-scale climatic influences. In comparison, for Las Vegas—the only mega city considered in this work—the rapid and extensive urbanization has led to a large enough signal of urbanization-induced daytime cooling that is identifiable from observation (Miller 2011).

3.3 Summary of the climatic effect of urbanization

Figure 8 summarizes the changes in 2 m air temperature induced by four leading types of conversion (four squares within each block) from non-urban to urban land cover. The figure demonstrates the change in temperature averaged over the grid points of WRF with the specific type of land-use change. Top to bottom are the five cities as labeled. The area of each circular disk is proportional to the area over which urbanization occurred. In Fig. 8, we have retained only the land-use change categories with a statistically significant signal (at > 95% level using standard t test on the “shift in the mean” against natural variability, e.g., Sardeshmukh et al. 2000). A specific square is left blank if the particular type of land conversion is too insignificant or if the corresponding temperature change is too small. A minor category not shown in Fig. 8 is the conversion from mixed forest to urban, which occurred in Jodhpur over a relatively small number of grid points. This case also exhibits nighttime warming and daytime cooling, for both winter and summer.

The summary in Fig. 8 reaffirms the robust features of nighttime warming and relatively weak daytime cooling due to urbanization over desert cities. Notably, this effect stays the same for the four cities that are significantly smaller than Las Vegas. For a quick reference, Tables 3 and 4 list the values of the changes in temperature as shown in Fig. 8. This generally reflects the localized nature of atmospheric response to small-scale land-use changes, at least in the WRF model. In other words, the local response in 2 m temperature does not significantly spill over to neighboring areas. As noted by Kamal et al. (2015), in WRF simulations the vertical extent of the atmospheric thermal response to a localized land-use change is generally shallow. The signal diminishes before it spreads vertically through the planetary boundary layer. This makes it difficult for the signal to disperse horizontally to areas far away from the city. Whether this characteristic is true in observation remains to be investigated.
Table 3

Changes in 2 m air temperature for summer as illustrated in Fig. 8

 

Nighttime

Daytime

C

B

E

F

C

B

E

F

Las Vegas

3.09

2.47

3.00

 

− 0.20

− 0.26

− 0.26

 

Hotan

  

2.46

   

− 0.13

 

Kharga

3.12

 

2.85

 

− 0.15

 

− 0.27

 

Beer Sheva

0.85

1.84

1.62

 

− 0.14

−0.03

− 0.12

 

Jodhpur

0.75

0.69

 

0.59

− 0.17

− 0.19

 

− 0.26

The four columns labeled C, B, E, and F correspond to the four kinds of conversion from non-urban to urban land coverage as defined in Table 2. A blank indicates that the specific type of land conversion did not occur or is insignificant for a given city

Table 4

Similar to Table 3, but for winter

 

Nighttime

Daytime

C

B

E

F

C

B

E

F

Las Vegas

1.95

2.12

1.68

 

− 0.12

− 0.21

− 0.69

 

Hotan

  

2.88

   

− 0.38

 

Kharga

3.38

 

2.99

 

− 0.65

 

− 0.47

 

Beer Sheva

1.56

2.00

1.76

 

− 0.12

− 0.20

− 0.29

 

Jodhpur

1.83

2.27

 

2.41

− 0.42

− 0.34

 

− 0.33

4 Discussion

The effect of urbanization on the local climate is generally not restricted to temperature. For example, Kamal et al. (2015) showed the weakening of terrain-induced diurnal circulation over Las Vegas due to urbanization. For Beer Sheva and Jodhpur, the two cities with diurnally or seasonally persistent local circulation, the wind effect might also be relevant. Due to noisiness of wind field over urban landscape, it is difficult to isolate this effect for those two smaller cities given the model resolution and duration of simulations considered in this work. It will be useful to clarify this effect in future work, especially considering that the change in the wind field can further affect temperature by modifying the ventilation effect (Kamal et al. 2015).

Our conclusions also potentially depend on the physical parameterization schemes in the model. The physical processes related to urban land-use changes are treated by the urban canopy model of Kusaka et al. (2001) and Kusaka and Kimura (2004) incorporated in WRF. Recently, more urban parameterization schemes of various levels of sophistication have been developed. An initial inter-comparison revealed non-trivial differences among their performances (Best and Grimmond 2015). Our five-city simulations, using a relatively basic, but widely adopted setting of the urban scheme, can potentially serve as a benchmark for testing increasingly complicated urban schemes in the model. A potential extension of this work is to consider an urban scheme that uses a finer classification of the land surface sub-types within the urban category (e.g., Li et al. 2013; Yang and Wang 2014). This would help more accurately incorporating the effect of urban greening into the simulation. While that effect is not yet fully included in the present work, given that urban greening generally leads to daytime cooling by shading and evapotranspiration (e.g., Shashua-Bar et al. 2009), it would not qualitatively alter our conclusion of daytime cooling, but might potentially further enhance that conclusion.

In our simulations, default values (given in Sect. 2.1) in the model are used for the basic parameters of urban structures, such as the average building height. Those values represent the typical conditions over an urban area. In general, those values can actually vary from city to city, and over time as a city undergoes development. The compromise we adopted was due to a lack of reliable observations for those parameters, especially for 1985, to constrain the model. Moreover, an accurate scheme for parameterizing the effects of detailed urban structures (such as degree of clustering of buildings, Myint et al. 2015) into a meteorological model is still under development. We anticipate future work using updated WRF model that incorporates those effects.

There are other processes broadly related to urbanization that would potentially affect climate, but are otherwise not treated in this study. Two such examples are the effect of anthropogenic heat, and the effect that arises from the changes in dust generation. A full coupling of those effects with a mesoscale meteorological model remains important future work. Another difficulty of incorporating those effects in the current simulation is the lack of reliable baseline data to constrain the boundary forcing for the model. For example, while we have reliable satellite observations of the land cover over the desert cities in 1985, it is more challenging to retrospectively estimate the anthropogenic heating or surface dust fluxes associated with those cities from that era. We hope that this discussion will inspire systematic collections of such data toward a more accurate simulation of urban climatic processes in the future.

5 Concluding remarks

From numerical simulations for five desert cities with a diverse range of size, baseline climate, and composition of land cover, a common pattern of nighttime warming and relatively weaker daytime cooling is found as a robust response of the local climate to urbanization. This climatic signal is confined to the areas where urbanization occurs. It is not sensitive to the type of background land surface that surrounds the urban area. It occurs in both summer and winter. In the majority of the cases, the signal is clear enough that a correspondence between land-use change and the induced change in 2 m air temperature is readily identified in the numerical simulations. While previous studies have found evidence of daytime cooling for several desert cities from observation, this work used specially designed numerical experiments to isolate and clearly attribute that effect as a robust response to desert urbanization.

By design, to isolate the urban effects, large-scale climate changes that are unrelated to urbanization are not included in our numerical experiments. A separation of the urban and non-urban effects from observation is generally difficult without dedicated long-term measurements. By numerical modeling, the framework used in this study can potentially be generalized to serve that purpose, if in the twin experiment perturbations are imposed to not only the land surface boundary condition, but also lateral boundary conditions that incorporate large-scale climate changes. This approach could also be used for projecting future climate changes in urban areas based on different scenarios of future urban developments.

Notes

Acknowledgements

This work was supported in part by National Aeronautics and Space Administration Grant NNX12AM88G. The authors appreciate useful comments from anonymous reviewers. Early results of this paper were reported in the first author’s PhD Dissertation (Arizona State University).

Compliance with ethical standards

Conflict of interest

The authors declare no conflict of interest.

References

  1. Best MJ, Grimmond CSB (2015) Key conclusions of the first international urban land surface model comparison project. Bull Am Meteor Soc 96:805–819CrossRefGoogle Scholar
  2. Brazel A, Selover N, Vose R, Heisler G (2000) The tale of two climates—Baltimore and Phoenix urban LTER sites. Clim Res 15:123–135. doi: 10.3354/cr015123 CrossRefGoogle Scholar
  3. Brazel A, Gober P, Lee SJ, Grossman-Clarke S, Zehnder J, Hedquist B, Comparri E (2007) Determinants of changes in the regional urban heat island in metropolitan Phoenix (Arizona, USA) between 1990 and 2004. Clim Res 33:171–182CrossRefGoogle Scholar
  4. Caldwell P, Chin H-NS, Bader DC, Bala G (2008) Evaluation of WRF dynamical downscaling simulation. Clim Change 95:499–521. doi: 10.1007/s10584-009-9583-5 CrossRefGoogle Scholar
  5. Carnahan WH, Larson RC (1990) An analysis of an urban heat sink. Remote Sens Environ 33:65–71CrossRefGoogle Scholar
  6. Chavez PS Jr (1996) Image-based atmospheric corrections: revisited and improved. Photogramm Eng Remote Sens 62:1025–1036Google Scholar
  7. Chen F, Tewari M, Kusaka H, Warner TT (2006) Current status of urban modeling in the community Weather Research and Forecast (WRF) model. In: Joint sixth symposium on the urban environment and ams forum: managing our physical and natural resources: successes and challenges, Atlanta, GA, USA, Amer. Meteor. Soc. CD-ROM. J1.4Google Scholar
  8. Chen F, Kusaka H, Bornstein R, Ching J, Grimmond CSB, Grossman-Clarke S, Loridan T, Manning KW, Martilli A, Miao S, Sailor D, Salamanca FP, Taha H, Tewari M, Wang X, Wyszogrodzki AA, Zhang C (2011) The integrated WRF/urban modelling system: development, evaluation, and applications to urban environmental problems. Int J Climatol 31:273–288CrossRefGoogle Scholar
  9. Diffenbaugh NS (2009) Influence of modern land cover on the climate of the United States. Clim Dyn 33:945–958. doi: 10.1007/s00382-009-0566z CrossRefGoogle Scholar
  10. Erell E, Williamson T (2007) Intra-urban differences in canopy layer air temperature at a mid-latitude city. Int J Climatol 27:1243–1255CrossRefGoogle Scholar
  11. Georgescu M, Miguez-Macho G, Steyaert LT, Weaver CP (2009) Climatic effects of 30 years of landscape change over the Greater Phoenix, Arizona, region: 1. Surface energy budget changes. J Geophys Res 114:D05110. doi: 10.1029/2008JD010745 CrossRefGoogle Scholar
  12. Georgescu M, Moustaoui M, Mahalov A, Dudhia J (2011) An alternative explanation of the semiarid urban area “oasis effect”. J Geophys Res. doi: 10.1029/2011JD016720 CrossRefGoogle Scholar
  13. Heikkila U, Sandvik A, Sorteberg A (2011) Dynamical downscaling of ERA-40 in complex terrain using the WRF regional climate model. Clim Dyn 37:1551–1564. doi: 10.1007/s00382-010-0928-6 CrossRefGoogle Scholar
  14. IPCC (2013) Climate change (2013) the physical science basis. In: Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds) Contribution of working group I to the fifth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, p 1535Google Scholar
  15. Kamal S, Huang H-P, Myint SW (2015) The influence of urbanization on the climate of Las Vegas metropolitan area: a numerical study. J Appl Meteorol Climatol 54:2157–2177. doi: 10.1175/JAMC-D-15-003.1 CrossRefGoogle Scholar
  16. Kusaka H, Kimura F (2004) Coupling a single-layer urban canopy model with a simple atmospheric model: impact on urban heat island simulation for an idealized case. J Meteorol Soc Jpn 82:67–80CrossRefGoogle Scholar
  17. Kusaka H, Kondo H, Kikegawa Y, Kimura F (2001) A simple single-layer urban canopy model for atmospheric models: comparison with multi-layer and slab models. Bound Layer Meteorol 101:329–358CrossRefGoogle Scholar
  18. Kusaka H, Hara M, Takane Y (2012) Urban climate projection by the WRF Model at 3 km horizontal grid increment: dynamical downscaling and predicting heat stress in the 2070’s August for Tokyo, Osaka, and Nagoya Metropolises. J Meteorol Soc Jpn 90B:47–63. doi: 10.2151/jmsj.2012-B04 CrossRefGoogle Scholar
  19. Lazzarini M, Marpu PR, Ghedira H (2013) Temperature-land cover interactions: the inversion of urban heat island phenomenon in desert city areas. Remote Sens Environ 130:136–152CrossRefGoogle Scholar
  20. Leung LR, Mearns LO, Giorgi F, Wilby RL (2003) Regional climate research: needs and opportunities. Bull Am Meteorol Soc 84:89–95CrossRefGoogle Scholar
  21. Li D, Bou-Zeid E, Barlage M, Chen F, Smith JA (2013) Development and evaluation of a mosaic approach in the WRF-Noah framework. J Geophys Res Atmos 118:11918–11935. doi: 10.1002/2013JD020657 CrossRefGoogle Scholar
  22. Mearns LO, Arritt R, Biner S, Bukovsky MS, McGinnis S, Sain S, Caya D, Correia J, Flory D, Gutowski W, Takle ES, Jones R, Leung R, Moufouma-Okia W, McDaniel L, Nunes AMB, Qian Y, Roads J, Sloan L, Snyder M (2012) The North American regional climate change assessment program: overview of phase I results. Bull Am Meteorol Soc 93:1337–1362CrossRefGoogle Scholar
  23. Miller JA (2011) Urban and regional temperature trends in Las Vegas and southern Nevada. J Ariz Nev Acad Sci 43:27–39CrossRefGoogle Scholar
  24. Myint SW, Wentz EA, Brazel AJ, Quattrochi DA (2013) The impact of distinct anthropogenic and vegetation features on urban warming. Landsc Ecol 28:959–978CrossRefGoogle Scholar
  25. Myint SW, Zheng B, Talen E, Fan C, Kaplan S, Middel A, Smith M, Huang H-P, Brazel A (2015) Does the spatial arrangement of urban landscape matter? Examples of urban warming and cooling in Phoenix and Las Vegas. Ecosyst Health Sustain 1:art15. doi: 10.1890/EHS14-0028.1 CrossRefGoogle Scholar
  26. National Research Council (2012) Urban meteorology: forecasting, monitoring, and meeting user’s needs. The National Academies Press, Washinton, p 176Google Scholar
  27. Oke TR (1982) The energetic basis of the urban heat island. Q J R Meteorol Soc 108:1–24Google Scholar
  28. Pan L, Chen S, Cayan D, Hart M, Zhang Q, Liu Y, Wang J (2011) Influence of climate change on the California and Nevada regions revealed by a high-resolution dynamical downscaling study. Clim Dyn 37:2005–2020. doi: 10.1007/s00382-010-0961-5 CrossRefGoogle Scholar
  29. Pielke RA Sr, Pitman A, Niyogi D, Mahmood R, McAlpine C, Hossain F, Goldewijk KK, Nair U, Betts R, Fall S, Reichstein M, Kabat P, de Noblet N (2011) Land use/land cover changes and climate: modeling analysis and observational evidence. Wiley Interdiscip Rev Clim Change 2:828–850CrossRefGoogle Scholar
  30. Potchter O, Itzhak Ben-Shalom H (2013) Urban warming and global warming: combined effect on thermal discomfort in the desert city of Beer Sheva, Israel. J Arid Environ 98:113–122CrossRefGoogle Scholar
  31. Rasul A, Balzter H, Smith C (2015) Spatial variation of the daytime surface urban cool island during the dry season in Erbil, Iraqi Kurdistan, from Landsat 8. Urban Clim 2(Part 2):176–186CrossRefGoogle Scholar
  32. Runnalls KE, Oke TR (2000) Dynamics and controls of the near-surface heat island of Vancouver. British Columbia. Phys Geogr 21:283–304CrossRefGoogle Scholar
  33. Sardeshmukh PD, Compo GP, Penland C (2000) Changes of probability associated with El Nino. J Clim 13:4268–4286CrossRefGoogle Scholar
  34. Sharma A, Huang H-P (2012) Regional climate simulation for Arizona: impact of resolution on precipitation. Adv Meteorol. doi: 10.1155/2012/505726 CrossRefGoogle Scholar
  35. Shashua-Bar L, Pearlmutter D, Erell E (2009) The cooling efficiency of urban landscape strategies in a hot dry climate. Landsc Urban Plan 92:179–186CrossRefGoogle Scholar
  36. Skamarock WC, Klemp JB, Dudhia J, Gill DO, Barker DM, Duda MG, Huang XY, Wang W, Powers JG (2008) A description of the advanced research WRF version 3, NCAR Tech. Note NCAR/TN-475 + STR. National Center for Atmospheric Research, Boulder, p 125Google Scholar
  37. United Nations Environment Programme (UNEP) (2006) Global deserts outlook, archived at www.unep.org/pdf/Global_Deserts_Outlook.pdf, p 164
  38. Yang J, Wang ZH (2014) Physical parameterization and sensitivity of urban hydrological models: application to green roof systems. Build Environ 75:250–263CrossRefGoogle Scholar
  39. Zheng B, Myint SW, Fan C (2014) Spatial configuration of anthropogenic land cover impacts on urban warming. Lands Urban Plan 130:104–111CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2017

Authors and Affiliations

  1. 1.Department of OceanographyNaval Postgraduate SchoolMontereyUSA
  2. 2.School for Engineering of Matter, Transport, and EnergyArizona State UniversityTempeUSA
  3. 3.School of Geographical Sciences and Urban PlanningArizona State UniversityTempeUSA

Personalised recommendations