This updated analysis of the status of the climate of the Arabian Peninsula (comprising Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, United Arab Emirates and Yemen) is important for several application-oriented sectors, such as water resources, agriculture, power generation, biodiversity, migration and food security. Having up-to-date climatic information for a region enhances the effectiveness of any assessments for climate impact studies, especially in relation to global warming and changes in climate (IPCC, 2007). This is particularly prevalent for, area-wise, the largest country in the Peninsula (Saudi Arabia), encompassing the world's largest sand desert, the Rub Al–Khali (Edgell, 2006; Bishop, 2010). A hindrance in obtaining climatic information for the Arabian Peninsula is the sparseness and/or unavailability of surface observation datasets for most of the countries. Thus, the utilization of available gridded datasets is invaluable, and appropriate under these circumstances.
Table I. Surface observation station information with rainfall over Saudi Arabia, including WMO code, longitude (°E), latitude (°N) and altitude (m).a
Rainfall (annual; mm)
In the first column, the serial number refers to station numbers in Figure 1. In the second column, the asterisk (*) marked after the station name indicates that the data are available only from 1985. All other datasets are for the duration 1978–2009. The last row provides the all-station average which is the country annual normal rainfall (mm).
This article addresses this aspect, and an attempt is made to provide an inclusive reference document presenting the status of the climate using several gridded datasets, and at the same time discussing the analysis with reference to the available ground observation datasets, where applicable. In particular, two main climatic parameters, namely rainfall and temperature, are studied in detail for the Arabian Peninsula as well as for Saudi Arabia because these two variables are important measures of climate in a region (Elagib and Mansell, 2000; Lazaro et al., 2001; Moonen et al., 2002; Islam et al., 2010). Rainfall is important for the natural replenishment of aquifers for fresh renewable water. Population growth, industrial development and expansion of irrigated agriculture have increased dramatically over the past century, raising demand for water supply. Furthermore, large-scale environmental challenges, such as the recent droughts in the Middle East (including the Arabian Peninsula), are straining the already scarce water resources (Ragab and Prudhomme, 2000; Şen et al., 2011). Both rainfall and temperature are key controllers of agricultural production, and in the Peninsula they have a distinctive important impact on wheat yield; there is also a concern that the currently low wheat yields are compromising the availability of water, which is considered to be of critical importance to the populace in Saudi Arabia (Ragab and Prudhomme, 2000).
The present-day climate of the desert and semi-desert areas is known to have changed on various temporal (interannual, interdecadal, multidecadal and interseasonal) and spatial scales (particularly for rainfall), which represents a major challenge for the climate forecasting and modelling of these climatic variables in these areas (Lioubimtseva, 2004). Generally, the climate of the Arabian Peninsula is characterized by hot dry summers and mild wet winters, as with the Mediterranean climate (El-Sabbagh, 1982; Ragab and Prudhomme, 2000). In the Arabian Peninsula, some areas receive their total annual rainfall in only a few days, from intense bursts of rain over a short duration such as 3 to 5 h in a day. A climatological study of rainfall and temperature variability is needed to better understand and to provide a general context for extreme rainfall events such as those occurred in Jeddah, Saudi Arabia on 25 November 2009 and on 26 January 2011 (Almazroui, 2011b, 2012a). In the recent past, above normal temperatures have also been noticed in the Arabian Peninsula, particularly since late 1990s (Almazroui, 2012b).
Nasrallah and Balling (1996) analysed the temperature over the Arabian Peninsula using a gridded dataset (5° latitude × 10° longitude for the duration 1891–1990, and 2.5° × 2.5° for the duration 1879–1992). It was reported that during this period, the temperature in the Arabian Peninsula increased by 0.63 °C. The decadal variations of the mean temperature on the seasonal basis over Bahrain for the period 1947–2005 were studied by Elagib and Abdu (2009). A positive trend of 0.16 °C per decade during the dry season was detected. Rehman (2010) analysed daily temperature and rainfall data for one station (Dhahran) in Saudi Arabia, for the period 1970–2006, and reported a rise of 0.4 °C (0.6 °C) and 0.5 °C per decade in the annual minimum (maximum) and mean temperature, respectively. They also reported a rise of 0.6 mm per decade for rainfall.
A monthly rainfall dataset derived from 29 stations in Saudi Arabia was used by Alyamani and Şen (1993) to study seasonal rainfall variability for a 15-year duration. Maximum rainfall over the southwestern region was noticed in January, along with other maxima in February and October in the central provinces. A 30 station-based annual rainfall dataset for the southwestern region of Saudi Arabia for the period 1970–1992 was studied by Abdullah and Al-Mazroui (1998). It was concluded that rainfall distributions fit better with normal and gamma distribution functions than with an exponential distribution, depending upon the selected domain inside the region. El-Seed (1995) also used the normal distribution for the study of annual rainfall probability over Saudi Arabia using different rainfall thresholds. Similarly, a 63 station-based rainfall study of southwestern Saudi Arabia for the period 1971–1990 was studied by Subyani (2004) to assess seasonal and regional variations in rainfall.
By the year 2050, the land areas may warm by as much as 1.68 °C over the Arabian Peninsula (IPCC, 2001). Projections for the future climate of the Arabian Peninsula (for the 2050s relative to 1961–1990) were studied by Kotwicki and Al Sulaimani (2009), along with the past and present-day climate. It was determined that extremes for both temperature and rainfall seem to be increasingly likely in the Arabian Peninsula, re-emphasizing the importance of a detailed assessment of the recent past climate of the region. Accordingly, the cumulative aim of this article is to provide updated information about the status of the climate of the Arabian Peninsula using gridded datasets, and for Saudi Arabia using both gridded and ground observation datasets, in particular during recent decades.
2. Data and methodology
In this article, we analyse the climatological spatiotemporal distributions of rainfall and temperature (and their trends) on an annual basis for the Arabian Peninsula and Saudi Arabia for a 32-year period (1978–2009). We used the gridded datasets to analyse the recent climatic conditions prevailing over the Arabian Peninsula, taken from the CPC Merged Analysis of Precipitation (CMAP; Xie and Arkin, 1997), Climatic Research Unit (CRU; New et al., 2000; Mitchell and Jones, 2005) and the Tropical Rainfall Measuring Mission (TRMM; Kummerow et al., 2000), simultaneously including surface observations over Saudi Arabia collected by the Presidency of Meteorology (PME). To cover the Arabian Peninsula (including its surroundings), the longitudinal and latitudinal boundaries in this study are taken as 20°–65°E and 5°–37°N, respectively (Figure 1). Given the significantly diverse perceived implications of the station-based climatological information, no attempt is made to re-grid or to transform the observed PME station data onto the grids, mainly to provide a multilevel comparison and to avoid rescaling. Furthermore, rainfall amounts < 1 mm are not displayed.
Figure 1 displays the location of Saudi Arabia in the Arabian Peninsula, along with the topography; rainfall in this arid environment is governed by its topography (Lioubimtseva, 2004). The latitudinal and longitudinal values of the 27 observed data stations in Saudi Arabia are also marked (red dots) with numbers. A list of all the stations from which the data are used in this study is provided in Table I. Table I shows annual rainfall for all stations used in this study as well the country annual rainfall average, which is computed using simple arithmetic mean. As can be seen from the table annual rainfall varies from 25.0 mm (Station 8) to 229.1 mm (Station 23), and that the country-wide average is 92.7 mm. A north–south oriented mountain range, with heights reaching 1500 m or above, is located in the southwest of the country, bordering the Red Sea; these mountain ranges pass through Saudi Arabia and on into western Yemen, and they slope gradually downwards to the east. This mountain range plays an important role in uplifting the Indian monsoon and moisture transpiring from the Red Sea (Chakraborty et al., 2006). The region encompassing this mountain range is also a reason of heavy rainfall occurring (Almazroui, 2006, 2011a).
The TRMM higher resolution (0.25° × 0.25°) gridded rainfall dataset is available for analysis, but only exists since 1998. This analysis used the TRMM 3B42 product, which is derived from merging the TRMM Microwave Imager (TMI) radiometer and the PR instrument data with other microwave (MW) sources. The relatively coarse resolution (2.5° × 2.5°) gridded rainfall dataset is available from the CMAP (both over the land and the adjacent bodies of water) and from the CRU (0.5° × 0.5°) (although for land only), both available up to 2009. The standard CMAP gridded dataset used in this study consists of monthly averaged precipitation values, obtained from five types of satellite estimates, such as GPI, OPI, SSM/I scattering, SSM/I emission and MSU, for the period 1979 onwards. The CRU TS3.1 dataset is derived from the gridded climate datasets, interpolated using the original information from the stations since 1901. In the interpolation records, the anomaly with respect to the average for the period 1961–1990 is calculated. The anomaly is interpolated using thin-plate splines as a function of latitude/longitude. The CRU dataset gridded values are obtained by applying a smooth fitting (in 3D space) to the available surface station observations (New et al., 2000). For the gridded datasets, we define very dry, dry and wet regions based on 40–80 mm, 80–150 mm and > 150 mm annual mean rainfall totals, respectively.
Commonly used statistical techniques are employed to estimate the linear trends and percentage differences using the datasets mentioned above (see, for instance, Wilks, 2006). The observed monthly temperature and rainfall datasets were used to estimate annual trends through regression methods. For the temperature variable, maximum, mean and minimum temperatures are considered separately. For detailed documentation of any changes in climate, a trend analysis is performed, using the observed datasets, by dividing the total period (1978–2009) into two halves: the first half is from 1978 to 1993 and the second is from 1994 to 2009.
3.1. Evaluation of gridded datasets
Before obtaining the climatology from the CMAP, CRU and TRMM datasets, the gridded rainfall and temperature are validated using the available observed station datasets of Saudi Arabia (Figure 2). It is found that there is strong significant correlation (0.86–0.99) between the country average gridded and the observational monthly rainfall (Figure 2(a)). Similar to rainfall, there is also a strong significant correlation between the gridded and the observational monthly temperature (Figure 2(b)). This suggests that there is a strong agreement between the gridded and the observational datasets.
3.2. Mean rainfall climatology
Figure 3 displays the mean annual rainfall for the Arabian Peninsula obtained from the CMAP, CRU and observed datasets for the period 1979–2009. Over the Peninsula, between approximately 15° and 30°N, we note a drop in the annual rainfall for the gridded datasets, from approximately 250 mm to approximately 80–100 mm (Figure 3(a) and (b)). The lowest mean annual rainfall (0–40 mm) occurs over the sand desert area to the west of the Arabian Peninsula, which is mainly over Egypt extending up to northwestern Saudi Arabia. The rainfall over the world's largest sand desert (Rub Al-Khali) is also low (below 60 mm). Across the Arabian Peninsula, the mean annual rainfall is less than 200 mm, thus classifying the climatic conditions of the Peninsula as semi-arid, arid and hyper-arid (Edgell, 2006). However, the rainfall in the southwest of the Peninsula ranges from 200 to 400 mm. The presence of the semi-permanent Hadley cell (high pressure) contributes toward reducing the rainfall amounts (on an annual basis) between 15° and 30°N. Nevertheless, a moderately heavy rainfall (80–150 mm) region exists in the middle-to-north of Saudi Arabia, and heavier rainfall (>150 mm) is evident in the extreme southwestern corner of the Arabian Peninsula (i.e., in Yemen) that extends up to southwestern Saudi Arabia and (to some extent) into southern Oman. A comparison with the observed dataset for Saudi Arabia only (Figure 3(c)), indicates that the coarse resolution CMAP data only moderately captures the unique characteristics of the southwestern region rainfall patterns, whereas the finer-resolution CRU dataset captures them very well. These results for heavy rainfall in the southwest of Saudi Arabia are consistent with the available reported information (e.g., Abdullah and Al-Mazroui, 1998; Subyani, 2004).
Figure 4 displays the short-term annual rainfall climatology over the Arabian Peninsula for the period 1998–2009; this time period is chosen due to the high-resolution TRMM rainfall data being available. Similar to the period 1979–2009 discussed earlier, heavy mean annual rainfall (>150 mm) occurs periodically over the southwestern region of the Arabian Peninsula, which includes the southwest of Saudi Arabia, whereas moderately heavy rainfall (80–150 mm) in the central parts of Saudi Arabia is also noticeable. The dry (0–40 mm) and moderately dry (40–80 mm) zones to the west of the Arabian Peninsula and over the Rub Al-Khali are also evident in this short-term climatology. Therefore, it is concluded that both the short-term and the long-term mean annual rainfall climatology fairly represent the dry region over Egypt and the moderately dry region over the Rub Al-Khali, with a moderately wet region over the southwest of the Peninsula. Thus, the results produced by the gridded datasets vis-à-vis the observed rainfall patterns over Saudi Arabia encourage the use of these datasets for studying the rainfall characteristics of the Peninsula.
The average rainfall for Saudi Arabia over different time scales is obtained from both the gridded (TRMM, CMAP and CRU) and the observational datasets, and are summarized in Table II. As can be seen from the table, only TRMM shows a relatively large bias (25.04%), compared to the other gridded datasets. This might be due to the shorter period of time for the TRMM data analysed, or it could be a false detection by its sensor and/or algorithm. Furthermore, it is also expected that the gridded datasets may have some limitations in estimating the values for the low-rain months (e.g., June).
Table II. Rainfall biases obtained for the TRMM, CMAP and CRU datasets with respect to the station data
3.3. Mean temperature climatology
On an annual basis, the climatological behaviour of the mean, minimum and maximum temperatures over the Arabian Peninsula is presented in this section, using the same two periods as discussed earlier. Descriptions of maximum and minimum temperatures are relevant for the study of extremes (Wilks, 2006).
Figure 5 displays the spatial distribution of mean annual temperature for the Arabian Peninsula, averaged for the period 1979–2009 using the CRU and the observed datasets. The gridded CRU dataset indicates that the average annual temperature is highest over the Rub Al-Khali, at around 27–30 °C, although equally high temperatures are also noticeable in the southwest of the Peninsula, mainly along the Red Sea coast (Figure 5(a)). The mean temperature over the major portions of the Arabian Peninsula is in the range 24–30 °C, whereas it is relatively lower (<21 °C) in the north. No observational data site exists over the Rub Al-Khali for verification of the result obtained from the gridded dataset, however, the available observed data over Saudi Arabia supports the temperature patterns discussed above (Figure 5(b)).
Figure 6 displays the spatial distribution of mean annual minimum temperature for the Arabian Peninsula, averaged for the period 1979–2009 using the CRU and the observed datasets. The northwestern and southwestern regions of the Arabian Peninsula experience temperatures as low as 6–15 °C (Figure 6(a)). The southeastern parts, on the other hand experience a relatively higher mean minimum temperature of 21–24 °C, which is mainly over the Rub Al-Khali. This is the highest temperature zone in the Arabian Peninsula, as we have seen for the mean temperature (see Figure 5(a)). Again, the distribution of minimum temperature over Saudi Arabia as obtained from the gridded CRU dataset is fully supported by the observed dataset (Figure 6(b)).
Figure 7 displays the spatial distribution of annual mean maximum temperature for the Arabian Peninsula, averaged over 1979–2009 using the CRU and the observed datasets. In the middle of the Arabian Peninsula, the maximum temperature peaks, indicating the impact of the land–ocean contrast. The highest regional temperature (33–36 °C) is evident to the southwest of the Arabian Peninsula, and slightly lower temperatures are evident along the Red Sea coast and over the Rub Al-Khali (Figure 7(a)). Relatively moderate temperatures, of between 27 and 33 °C, are obtained for most other parts of the Peninsula. Inside Saudi Arabia, the behaviour of maximum temperature obtained from the gridded data well represents the characteristics obtained from the observed data (Figure 7(b)).
The distribution patterns for the annual average mean, minimum and maximum temperatures obtained from the gridded datasets fairly represent the climatological characteristics of the Arabian Peninsula, being similar to those obtained from the observed dataset for Saudi Arabia. The biases for the monthly, seasonal and annual temperatures obtained from the CRU vis-à-vis the station data are presented in Table III. There is good agreement between the CRU and observational datasets in most of the months, however, the largest overestimation for minimum temperature is around 0.5 °C in January. For maximum temperature, the results show that the CRU underestimates the observed values in all months, where the largest bias is around − 1.54 °C in July. On the annual scale, the CRU overestimates minimum temperature by about 0.17 °C, whereas it underestimates by about 1.22 °C (0.52 °C) for maximum (mean) temperature.
Table III. Temperature biases obtained from the CRU datasets, against the station values, averaged for the period 1978–2009
3.4. Rainfall trends
In this section, the temporal distribution of the observed rainfall over Saudi Arabia is discussed using regression trend analysis.
Figure 8 displays the time sequences of the observed annual rainfall, averaged over Saudi Arabia during the period 1978–2009. The 32-year analysed time period is described in three phases: the entire period (1978–2009), the first half (1978–1993), and the second half (1994–2009). The analysis showed that there was an insignificant increasing trend in the first half of about 13.9 mm per decade, where the average rainfall was 93.6 mm with a standard deviation of 29.7 mm. Importantly, as can be seen from the figure, the trend analysis of the second half shows a significant decreasing trend (at 99% level) of about 47.8 mm per decade, with average rainfall of about 88.8 mm and standard deviation of about 31.9 mm (Figure 8). A decrease of mean annual rainfall by 10 mm from the period 1966–1978 to 1979–1991 for a station (Riyadh) in central Saudi Arabia is also reported by Qureshi (1994). However, the entire period also shows a decreasing trend of 6.2 mm but this is statically insignificant, whilst the average rainfall of the entire period (1978–2009) was found to be 92.7 mm and its standard deviation value was 30.4 mm. A similarly insignificant decreasing trend was found for the smoothed series for the entire period using the Gaussian filter (Figure 8). The relatively large standard deviation value for all periods indicates large interannual variability in the rainfall over Saudi Arabia. Note that the large decreasing rate in the rainfall of the recent past (1994–2009) over Saudi Arabia is a warning of the possible impact of climate change in the country, suggesting further study of station data at the daily scale. Such a fully updated review of the climate status of Saudi Arabia and its different parts could significantly enhance climate change assessment studies related to various socioeconomic application-oriented activities.
As noted in Figure 8, the highest annual rainfall occurred during the year 1982, amounting to 171.5 mm, whereas the next highest rainfall occurred during the year 1997, amounting to 165.1 mm. The gridded datasets display the two broad dense spatial distribution patterns in these 2 years, confirming that heavy rainfall occurred across the country (Figure 9). During 1982, heavy rainfall occurred in the central parts of Saudi Arabia, extending to the Arabian Gulf, whereas during 1997 it occurred over the southwest of Saudi Arabia. During these heavy rainfall years, the dry zone over Saudi Arabia remains similar to the rainfall climatology (see Figures 3 and 4), whilst the moderately dry zone is shifted somewhat to the south of the Rub Al-Khali in the CMAP. The country average rainfall amounts for Saudi Arabia were 171.5 (165.1), 184.4 (167.7) and 105.9 (119.9) mm for the observed, CMAP and CRU data in 1982 (1997), respectively. From Figure 8, the two minimum rainfall years may also be noticed. During the year 2007, the mean annual rainfall was 52.8 mm, whereas during the year 2009, it was only 40.7 mm. These are relevant in identifying the meteorological drought of those years (based on rainfall amounts and patterns only).
3.5. Temperature trends
In this section the trend analyses for the maximum, mean and minimum temperatures are discussed, based on the observed dataset for Saudi Arabia.
Figure 10 displays the time sequences for the annual maximum, mean and minimum temperatures for Saudi Arabia along with their linear trends, based on the observed data during the period 1978–2009. The maximum temperature (on average 31.72 °C) increases at a rate of 0.71 °C per decade, with R2 = 0.60, and with a standard deviation value of 0.83 °C. The mean temperature (on average 24.67 °C) increases at a rate of 0.60 °C per decade, with R2 = 0.62, and with a standard deviation value of 0.66 °C. The minimum temperature (on average 17.56 °C) increases at a rate of 0.48 °C per decade, with R2 = 0.60, and with a standard deviation value of 0.52 °C. The maximum temperature increase rate is about 0.11–0.12 °C higher than the increase rate of the mean and the minimum temperatures. The relatively small standard deviation values and higher R2 are indicative of the lower interannual variability in the temperature compared to the rainfall values. These linear rising trends are in general agreement with the results reported by Rehman (2010) and Narsallah and Balling (1996), given the statistical uncertainty caused by the available time period for which the data are analysed (Liebmann et al., 2010).
Figure 11 displays the spatial distributions for the mean annual temperatures in 1982 and 1999, which are the lowest and highest mean annual temperature years, respectively, during the period 1978–2009 (see Figure 10). During the lowest mean temperature year (1982), the mean annual temperature drops to 23.18 °C, whereas during the high mean temperature year (1999), mean annual temperature rises to 25.45 °C (see Figure 10). Because mean, minimum and maximum temperatures are linearly related to each other, only the mean temperature for the lowest and highest mean temperature years are displayed for illustrative purposes (Figure 11). It is clear that, during the lowest temperature year (1982), temperatures between 21 and 27 °C are evident across Saudi Arabia but that the areas experiencing temperatures as low as between 15 and 18 °C are in the northwest and southwest of the Arabian Peninsula (Figure 11(a)). In contrast, during the highest mean temperature year (1999), major parts of the Saudi Arabia/Arabian Peninsula experienced temperatures between 24 and 30 °C (Figure 11(b)). In this year, however, the low-temperature zone (15–18 °C) is confined to only the southwestern corner of the Peninsula.
The annual mean rainfall climatology obtained from the gridded CMAP, CRU and high-resolution TRMM datasets shows the drop of rainfall between 15° and 30°N, which covers major parts of the Arabian Peninsula. These gridded datasets clearly indicate that the southwest heavy rainfall region is similar to that obtained from the observed dataset, justifying the use of these gridded datasets. They also reveal the drier zone over Egypt that extends into northern Sudan and into northwestern Saudi Arabia. Furthermore, the moderately dry zone is evidenced over the Rub Al-Khali. Overall, the observed data show that rainfall over Saudi Arabia is decreasing, and that the decrease rate is faster and significant in the recent past (1994–2009), compared to the entire analysis period (1978–2009), whereas it was increasing in the 1980s to the mid-1990s.
Similar to rainfall, the temperature climatology indicates that the warmer areas are in the Rub Al-Khali and Sudan, whereas the colder ones are in the northwestern and southwestern parts of the Arabian Peninsula. The overall temperature over Saudi Arabia is increasing significantly. The lowest temperatures in any given year occur over the far northwestern and southeastern parts of the country. In contrast, the highest temperatures in any given year are over the Rub Al-Khali; these extend northward and westward within the country. However, the variations in rainfall and temperature over the Arabian Peninsula, regardless of duration, have not yet been fully assessed from the above discussion. In this connection, the changes in rainfall and mean temperature for the decades 1980–1989, 1990–1999 and 2000–2009, with respect to the duration, 1979–2009 are discussed next.
Figure 12 displays the spatial distribution of rainfall change (%) for three decades (1980–1989, 1990–1999 and 2000–2009, each with respect to 1979–2009) obtained from the CRU data. The Δ statistic values were calculated as follows:
Here, Δ1980s stands for the 1980–1989 decade, and Vbase stands for the 1979–2009 period average. A similar definition was used for the 1990–1999 and 2000–2009 decades. During the decade 1980–1989 over the Arabian Peninsula, there are two regions where the relative percentage change in rainfall is positive, that is the areas are wetter (above average). One is in the north/northwestern region and the other is in the centre extending to the Arabian Gulf (Figure 12(a)). In this decade, the drier areas (below average) in the Arabian Peninsula are over Yemen and over a belt inclined southeast-to-northeast from Oman to Kuwait through the Gulf countries (extending westward into Iran). The dryness in the southwest of the Peninsula extends westward into Sudan and southward into Somalia. In the same regions during the 1990–1999 decade, the opposite situation occurs vis-à-vis the northwestern and southwestern regions as well as Sudan and southwestern Iran (Figure 12(b)), except that in the central and eastern regions (now extending into eastern Oman), the higher rainfall is more pronounced. During the 2000–2009 decade, yet another picture emerges; in contrast to the previous decade (1990–1999), the regions of eastern Saudi Arabia including Gulf countries, eastern Oman and southwestern Iran become drier, but the increase in rain continues in the southwestern areas of the Peninsula, in particular over Yemen (this also now extends into Eritrea). As for the northwestern parts of the Peninsula, the picture reverses once more and this region again experiences an increase in rainfall, as evident in the 1980–1989 decade (Figure 12(c)). This behaviour is indicative of a great deal of variability in the rainfall patterns on the decadal scale; this knowledge is critical for developing effective long-term planning strategies related to addressing the impacts of change in climate.
Figure 13 displays the spatial distribution of mean temperature change ( °C) for the decades 1980–1989, 1990–1999 and 2000–2009, each with respect to 1979–2009, obtained from the CRU dataset. Equation (1) is used here with appropriate modifications. During the 1980–1989 decade, there are clear indications of cooler mean temperatures over the Arabian Peninsula and the entire analysis domain, except for a slight warming evident over a portion of eastern Sudan (Figure 13(a)). During the 1990–1999 decade, the mean temperature over the Peninsula increased relative to 1980–1989. Hence, the cooling of the previous decade was reversed during 1990–1999, with the mean temperature for this decade at ± 0.1 °C with respect to the base period (1979–2009), except for an low-temperature belt (0.1–0.2 °C) inclined southeast-northwest over Saudi Arabia (Figure 13(b)). In contrast, warmer temperatures are evident over the entire Arabian Peninsula during 2000–2009 (Figure 13(c)). Similar to the rainfall patterns, a very different picture emerges for the Peninsula and the entire analysis domain for the 2000–2009 decade compared to the 1980–1989 decade. This indicates that temporal scale variability can be significantly high for both rainfall and temperature in a particular region, and thus, precaution must be exercised and the temporal scale must be considered in employing these parameters in any long-term planning designed to address the impacts of climate change.
The CRU, CMAP and TRMM gridded datasets are used to present the spatiotemporal distributions of rainfall and temperature over the Arabian Peninsula for the period 1979–2009 on annual and decadal bases. These three different datasets are compared with the observed data for Saudi Arabia and good agreement among the gridded and station datasets was found. The differing spatial resolution of these datasets has allowed for a multi-scale intercomparison of rainfall and temperature over the Peninsula for the first time.
The southeastern region of the Arabian Peninsula, encompassing the Rub Al-Khali, received the lowest annual rainfall throughout the analysis period. The highest rainfall is observed over the southwest of the Arabian Peninsula. The temporal analysis of the rainfall obtained from the surface observed data over Saudi Arabia indicates a large interannual variability in the rainfall over the region for the study period (1978–2009). The decadal percentage changes calculated for rainfall further confirm this finding. A linear decreasing trend of 6.2 mm per decade is found in the observed rainfall for Saudi Arabia during the study period (1978–2009), whilst there is a statistically significant decreasing trend of as much as 47.8 mm per decade for the recent past (1994–2009). It is interesting to note that rainfall increased in the southern Peninsula and along the Red Sea coast, whilst it decreased over most parts of Saudi Arabia during the 2000–2009 decade, compared to 1980–1989. The highest mean temperatures are observed over the Rub Al-Khali, whereas the lowest are observed over northwestern and southwestern parts of the Arabian Peninsula. The interannual variations in temperature over Saudi Arabia are found to be relatively small; however, the decadal variations in temperature are relatively large. For Saudi Arabia, the linear trend for maximum temperature is an increase of 0.71 °C per decade, which is higher than that of mean temperature (0.60 °C per decade) and minimum temperature (0.48 °C per decade), respectively (for the period 1978–2009). The mean temperature over the entire Peninsula increased during the 2000–2009 decade relative to 1980–1989. With these findings, the next step is to utilize climate models in order to simulate the Arabian Peninsula's current and future climate as well as to study weather extremes; these are needed if climate change impact studies for the region are to deliver reliable data for planners.
This study was sponsored by King Abdulaziz City for Science and Technology (KACST), Riyadh, Saudi Arabia, under Strategic Priorities for Environmental Technology Program (Grant No. 8-ENV125-3). The Presidency of Meteorology and Environment (PME) in Saudi Arabia is acknowledged for providing the observational datasets used in this study. The CMAP (http://www.esrl.noaa.gov/psd/data/gridded/), the CRU (http://www.cru.uea.ac.uk) and the TRMM (http://trmm.gsfc.nasa.gov) are also acknowledged for providing the gridded datasets.