A Regional Climate Simulation Study using WRF Model over the Arabian Peninsula for Temperatures Difference between (2006-2010) and (2056-2060)

Copyright: © 2015 Alsarraf H. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. A Regional Climate Simulation Study using WRF Model over the Arabian Peninsula for Temperatures Difference between (2006-2010) and (20562060)


Introduction
Global models can characterize the large scale weather systems that affect the Arabian Peninsula, but not the fine scale processes associated with local terrain and land cover. To capture these smaller scale features, a simulation at sufficiently high spatial resolution of the local complex terrain and the heterogeneous land surfaces is necessary [1]. Investigation of the climate change issue began with Global Climate Models (GCMs), but recently climate scientists took advanced a step by using high resolution dynamical downscaling models to investigate climate change in particular regions [2]. Using a higher resolution regional model to examine a regional will provide an enhanced results in terms of interactions between land, atmosphere, and ocean; convection and microphysics; and cloud, aerosol, and radiation [3]. In this study, we used the Weather Research and Forecasting (WRF) model as a Regional Climate Model (RCM), to dynamically downscale GCM output to the Arabian Peninsula, and especially, Kuwait using three nested domains at grid resolutions of 48 km, 12 km, and 4 km. The main motivation of this study is to use a high resolution regional model to investigate the changes in summer surface Temperatures in the future.
To address these, we will use output from a GCM to provide lateral forcing for WRF runs for two five-year periods, fifty years apart, to reveal signals of climate change. The first five-year period was from 2006-2010, while the second five-years were from 2056-2060. The purpose is to investigate the changes in temperatures in the first five years at the beginning of the 21 st century comparing it with conditions fifty years later according to the model.

Global and regional surface temperature change
Changes in extreme temperatures over a long time period have been analysis by Yan et al. [4]. They determined that not only low temperatures have been decreasing and high temperatures have been increasing during the last decades, but also that there have been earlier changes in these extremes [5,6]. For example, between the period of 1900-2002, the surface temperature increased linearly by 0.069°C per decade. A high increase in temperature took place between 1910 and 1945 and from 1970 to the present, when the temperature increased by 0.16°C and 0.17°C per decade, respectively, which is about three times more than the increase during the past century [7,8]. Trenberth et al. [9] also showed that the global mean surface air temperature has risen around 0.74°C over the last 100 years , and that the rate of increase has grown from 0.07°C per decade over the last 100 years to 0.13°C per decade over the last 50 years. Surface air temperatures have increased more over land (0.27°C per decade) than over the ocean (0.13°C per decade) meaning that increases in the frequency of extremely high average temperatures are to be expected [10]. The global climate variability and the changes in the Indian

Abstract
The purpose of this study is to examine the impact of climate change on the changes on summer surface temperatures between present (2006-2010) and future (2056-2060) over the Arabian Peninsula and Kuwait. In this study, the influence of climate change in the Arabian peninsula and especially in Kuwait was investigated by highresolution (48, 12, and 4 km grid spacing) dynamic downscaling from the Community Climate System Model (CCSM 4) using the WRF (Weather Research and Forecasting) model. The downscaling results were first validated by comparing NCEP model outputs with the observational data. The global climate change dynamic downscaling model was run using WRF regional climate model (WRF) simulations (2006)(2007)(2008)(2009)(2010) and future projections (2056-2060). The influence of climate change in the Arabian Peninsula can be projected from the differences between the two period's model simulations. The regional model simulations of the average maximum surface temperature in summertime predicted an increase over the next 50 years. monsoon do affect the climate in the Arabian Peninsula which led into significant changes in the climate over a period of decades [11]. A study by Almazroui et al. [12], described about the recent changes that had occurred in the Arabian Peninsula, and how these changes have brought changes in the population growth, industrial development and expansion of irrigated agriculture. In the Arabian Peninsula 80% of the area is occupied by Saudi Arabia [13]. The climatic datasets from its 27 ground observations that were analyzed for the period 1978-2009, described the observed state and change of the present climate.
The data showed that the spatial distribution of mean annual temperature for the Arabian Peninsula, averaged for the period 1979-2009 was highest over the Rub Al-Khali, and was equally high temperatures in the southwest of the Peninsula, mainly along the Red Sea coast. The observation from periods 1978-2009 was broken down into two periods, the first half-period (1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993), and the second half-period (1994-2009). The temperatures observation from 1978-2009 indicates an increase in maximum temperature by 0.71°C, Mean temperature by 0.60°C, and minimum temperature by 0.48 °C, overall temperatures are increasing significantly in the region [14,15]. The average daily JJA 2 m temperatures observation from the World Meteorological Organization (WMO) at Kuwait international airport from  shows an increase in 2 m temperatures by 1.2°C from 1960 to 1967, and increase from 1968 to 1990 by 2°C [16]. In the summer, the average daily high temperature ranges from 42 to 46°C. Dust emission normally decreases Tmax and Tmin values by between 0.5°C [17], however the increase in temperatures can be an indications of decreasing dust emissions and an increase of the incoming shortwave radiation in the region.

Experimental Design
In this part of the study, a GCM was used to provide lateral forcing for WRF runs for two five-year time periods, fifty years apart, to allow any climate change signal to be revealed. The first five years run were in 2006-2010 and second five years are to run were in 2056-2060. The purpose is to investigate surface maximum temperatures comparing the present simulations with future simulations.
The Weather Research and Forecasting (WRF) Model was used to dynamically downscale the results of the CCSM4, to get higher resolution for analysis of data attained from study of the Middle East. The simulation of the WRF regional climate model was run using the same three nested domains at resolutions of 48, 12, and 4 km as for the previous reanalysis runs to keep simulated states close to the driving state at large scales, while still generating small-scale features ( Figure  5). The goal of downscaling is to initiate mesoscale meteorological     features consistent with the large-scale state simulated using the global model [18].
For dynamical downscaling, data generated from 5 years of current climate simulations (2006-2010) using CCSM4 under the RCP8.5 scenario and 5 years simulations of future climate patterns (2056-2060) were used. The WRF runs were initialized at 0000 UTC, April 27 and ended on October 3 of each year covering the summer months (May-September). The WRF was run daily providing data every three hours a day.
Applying the WRF model for five years throughout the summer only is to ensure that we would capture the changes in maximum temperature. Five years at the beginning of the century will provide us with a reasonable estimate of the present climate in order to be able to compare it with future climate.
The second WRF run simulated weather patterns for a five year period in the middle of the century (2056-2060) during the summertime. These two five year periods (2006-2010 and 2056-2060) were chosen because going 50 years out helps assure the longterm signal will be clearly distinguished from shorter-term climate variability. We will present the differences between these two periods to investigate climate changes in the region. The WRF was configured to perform long simulations and fully represent the climate system's response to climate change forcing.
The WRF model has been improved over earlier versions (MM5 and MM4) the previous years, and used to simulate regional climate model in different regions. Because WRF is non-hydrostatic model is very flexible to use in multiple physical parameterization and had the capability to simulate as RCM due to the scale ranging [19]. Downscaling reduce error in climatological means temperatures, wind victors, moister, and precipitations [20]. The process of downscaled global climate model will reduce the bias and provide more realistic simulations in summertime in the Arabian Peninsula in temperatures, Mean sea level pressure, and wind speed [21].
One useful capability of the WRF is its flexibility in choosing different dynamical and physical schemes [22]. The physical parameters that been used in this study captured the physical and the dynamical climate processes in the Arabian Peninsula more realistic (Table 1). We used the WRF downscaling model that had been used to analyze desert areas within the US for better results. We also tested the parameters in the verification process by comparing WRF outputs that is forced by NCEP with observation data.

Model Evaluation
In this study the WRF model was downscaled from the National Centers for Environmental Prediction (NCEP)-NCAR Global Reanalysis Products (NNRP) to evaluate climate variability for three years 1997, 2000, and 2008. We illustrated climate variability by testing WRF as a Regional Climate model simulation in different time scales, hourly, daily means, and monthly means, in summertime from May to September. The WRF regional model performance did show the summertime variability, and the behavior of specific phenomena. The dynamical downscaling model (WRF) was applied for the years 1997, 2000, and 2008, in which NCEP/NCAR reanalysis data were used to provide initial and lateral boundary conditions. The WRF runs were initialized at 0000 UTC, April 27 and ended on October 3 of each year giving complete coverage of the summer season (May through September) with output data eight times per day. The three-hourly values were then used to compute daily averages and monthly averages for further analysis. Daily maximum and minimum temperatures were the maximum and minimum, respectively, of the three-hourly temperatures and, as such, may not represent the actual maximum and minimum temperatures simulated by the model.
The WRF high resolution simulation results forced by NCEP reanalysis are compared with data coming from the local surface observations, using the nearest grid point of the model for the years 1997, 2000, and 2008. The NCEP runs were verified with the Kuwait International Airport (29°N, 48°E) observation data. The Kuwait airport is also an aviation station which provides METARs every 30 minutes for the World Meteorological Organization (WMO). The observation data at Kuwait International Airport are taken every half an hour, and analyzed to give hourly, daily, monthly, and yearly averages. The data are well organized for mesoscale phenomena and climatology studies. In this study we focused on the maximum temperature at 2 meters. Temperatures were measured at 2 m above the surface taking both the maximum and minimum temperature daily. This is very valuable data which can be compared with the 2 m maximum temperature data that has been simulated by the NNRP-forced WRF.

Variable
Value Description  The verification is calculated in Table 2, which shows the averages and standard deviation for years 1997, 2000, and 2008 for the months of June, July, and August. In Table 2, its showing that averages of the KIA observations and the NCEP 12km domain are very close, but the 48 km and the 4 km domain NCEP model showing lower values compared to KIA observations. The calculation of standard deviation in Table 2 shows a close value between KIA observations and 4 km domain NCEP model. The model tends to give slightly lower averages values than the KIA observations. The correlation between the model and the observation showing a positive linear relationship, while the Root-Mean-Square-Error (RMSE) and the Mean-Absolute-Error (MAE) calculation showing that the smallest magnitude of error at 12 km grid resolution ( Table 2).

GCM forcing
Global climate modeling used to capture large scales of atmospheric phenomena and the interactions with land, and ocean. To be able to decrease the global climate model biases and illustrate regional topography in better fine scale, scientist must use the downscaling climate model with finer resolution that has the capability to identify regional and local phenomena [23]. In Figure 6, the left column shows the difference of the mean maximum temperature in May, June, July, August, and September between present (2006-2010) and the future (2056-2060) in the Middle East from the Community Climate System Model (CCSM4) simulation. The right column in Figure 6 shows difference between the same periods of simulation from the WRF-RCM (D01) 48 by 48 km grid resolution forced by CCSM4. It is clear to see that the WRF-RCM simulation is different than the CCSM4 simulation, with the WRF-RCM showing more detail in the temperature because the high resolution regional model will take into account urban, microscale, and mesoscale, and local land/sea breeze circulation.

Dynamic downscaling model
The observations and the historical simulation of global climate models indicate an increase in global temperature. Examine regional temperature also shows an increase temperature which is expected [24]. In most deserts environment temperatures are expected to increase up to 2.0°C, with higher increases in summertime. The increase in temperature in summertime will address an issue with water resource in the future. Simulating the temperature difference between present and future in WRF in the Arabian Peninsula using 3 downscaled domains will expect an increase in temperatures.    In May, the mean maximum temperature showed an increase of 1°C in the Arabian Peninsula, but the northwest of the Arabian Peninsula showed an increase of 2-3°C. In June (Figure 7b), the results showed a similar trend in the increase of surface temperature in the Arabian Peninsula, over the Arabian peninsula an increase by 1°C, except the northwest regions in the Arabian peninsula were its indicating an increase of 2°C. July, which is considered to be the hottest month in the Arabian Peninsula (Figure 7c), showed that Tmax will increase by 2-3°C in the Northern region of Arabian Peninsula, central region an increase by 1-2°C, and in southeast regions of the Arabian peninsula an increase of 1-2°C, were the southwest is an increase of 1°C.
In August (Figure 7d    in Figure 9a in May shows an increase by 1°C in Kuwait, eastern Saudi Arabia (ESA), Qatar, and UAE. In Figure 9b, showing a similar trends in June as seen in May. In July the D03 indicating an increase by 2°C in Kuwait, and an increase of 2-3°C in southern Iraq. The model simulation indicates an increase by 1°C in ESA, Qatar, and UAE. In July (Figure 9c

Summary and Conclusion
The WRF predicted during the whole summer an average increase between 0.5-2.5°C in Tmax in the next fifty years in the Arabian peninsula between the present (2006-2010) and the future (2056-2060) using the three high resolutions downscaling domains D01, D02, and D03 ( Table 3). As in the study by [25] model projections results shows an increase by 3.1°C under B2 scenario , and an increase by 4°C under A2 scenario by the end of the century. The three domains simulations indicate similar results in terms of increasing in temperatures with some more details in the fine domains. The simulations results from the three different fine scale resolutions didn't show big differences in temperature differences. The increase in Temperatures during summer time in the Middle East was expected mainly due to a global increase in temperature that had been investigated and predicted by many scientists using the GCM and by investigating regional temperature trends by using the downscaling RCM in different regions.
The increase in temperatures in the future in the desert of the Arabian Peninsula will modify the soil moister and albedo values in the future. The decreases of albedo values will leads into increasing the shortwave radiations observed by the surface. The rise of surface temperature will increase the soil evaporation rates in the regions; this will be results in an increase of the evapotranspiration. The future increase of the evaporation will affect groundwater, and surface water in the desert, these consequences will harm the water resource. The surface evaporation is projected to increase by 6.5% (20.8 mm /y) Under A2 scenario and an increase by 15.2% (48.9 mm / y) Under B2 scenario using PRECIS by the end of the century [25]. The increase of 1°C in the Arabian Peninsula would raise the evapotranspiration.
The increase of 1°C in the Arabian Peninsula would raise the evapotranspiration by 1-4.5%, and with an increase of 5°C will increase the evapotranspiration by 6-19.5% [26]. Correlating both studies results into this study of WRF temperatures simulations indicating that an increase between 0.5-2.5°C in the Arabian Peninsula the next fifty years can affect the water resources in region.
The climate change impact on temperatures in the future can alter the temperature contrast in the region. Further action can take place in the Land/sea breeze circulation in the region, with an increase of inland temperature, and the gulf maintain its normal temperature could lead into intensifying the breeze in summertime. Other scenario is if both increased in the same rate there will be no further change in terms of intensifying or weakening in the future [26][27][28][29].
The second part of this research will be using high resolution WRF climate model to examine the impact of climate change on future dust storms frequent in summertime in Kuwait. The study will investigate the changes in the summertime Pressure gradient in the Arabian Peninsula and how will it affect the wind speed in Kuwait by midcentury.