Journal of Hydrology: Regional Studies 12 (2017) 122–135

Contents lists available at ScienceDirect

Journal of Hydrology: Regional Studies

journal homepage: www.elsevier.com/locate/ejrh

Assessment of the consistency among global precipitation products

T
over the United Arab Emirates
⁎
Youssef Wehbea,c, , Dawit Ghebreyesusa, Marouane Temimia, Adam Milewskib,
Abdulla Al Mandousc
a
Department of Chemical and Environmental Engineering, Khalifa University of Science and Technology, Masdar Institute, Masdar City, P.O. Box
54224 Abu Dhabi, United Arab Emirates
b
Department of Geology, Water Resources and Remote Sensing Laboratory (WRRSL), University of Georgia, P.O. Box 1868, Athens, GA, USA
c
National Center of Meteorology and Seismology (NCMS), P.O. Box 4815, Abu Dhabi, United Arab Emirates

AR TI CLE I NF O AB S T R A CT

Keywords: Study region: United Arab Emirates.

United Arab Emirates Study focus: Numerous global precipitation products have been developed and calibrated.
Precipitation products However, their performance over arid regions such as the United Arab Emirates (UAE) has been
Consistency revealing notable differences triggered by a combination of climatic and terrestrial attributes.
Topography
The objective of this study is to cross-validate and analyze the consistency of four global
Land cover
precipitation products from the GPCC, TRMM, WM, and CMORPH datasets over the UAE using a
Calibration
network of 53 rain gauges from 2000 to 2010. The spatial analysis of their consistency versus
topography and land cover is expected to reveal the factors affecting the country’s rainfall
regime. The study also identifies and calibrates the best statistically performing precipitation
product as an essential climatic input for monitoring, forecasting, and modeling hydrologic
applications over the UAE.
New hydrological insights for the region: The UAE, similar to other dryland environments lacking
adequate hydrologic monitoring networks, presents a unique area to evaluate satellite remote
sensing products and the erratic spatiotemporal nature of precipitation in diverse environments.
Statistical analyses indicate that the TMPA V7 precipitation products record the highest overall
agreement with the observational network. Within the UAE, areas that receive high rainfall and
fall within the vegetated highlands (e.g., > 250 m), provide the most promise for incorporating
satellite precipitation into hydrologic monitoring, modeling, or water resource management.

1. Introduction

In the last few decades, the Middle East has undergone substantial development that led to the establishment of large cities such as
Dubai and Abu Dhabi in the United Arab Emirates (UAE), among others, with a considerable increase in water demand. The lack of
renewable water resources in arid regions such as the UAE requires accurate monitoring of the scarce rainfall events. The challenge
exists in the sporadic nature of rainfall events across the UAE with significant variability in space and time. Increased understanding
of precipitation variability over arid regions should lead to a better understanding of recharge regimes, mitigation of the risk of flash
floods, and more optimized management of available resources. This has become particularly important where rain gauges or ground-

⁎
Corresponding author.
E-mail addresses: ywehbe@ncms.ae (Y. Wehbe), dtghebreyesus@masdar.ac.ae (D. Ghebreyesus), mtemimi@masdar.ac.ae (M. Temimi),
milewski@uga.edu (A. Milewski), amandoos@ncms.ae (A. Al Mandous).

http://dx.doi.org/10.1016/j.ejrh.2017.05.002
Received 6 December 2016; Received in revised form 7 May 2017; Accepted 8 May 2017
Available online 15 May 2017
2214-5818/ © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

based radar are generally sparse, mainly because such regions are remote and stretch over large domains with low population
densities, diminishing the need to deploy a network of rain gauges and automated weather stations. Therefore, estimates of
precipitation have relied on other alternatives including satellite measurement, numerical models, or hybrid approaches that
integrate both.
Global precipitation products from remote sensing observations and/or numerical models offer an operational and systematic
monitoring of rainfall events in arid regions. Currently, remote sensing precipitation products are widely available covering the entire
globe with high spatial (e.g., 0.1 arc-degree) and temporal resolutions (e.g., 30 min) (Awange et al., 2016). However, remotely sensed
products are very susceptible to a wide range of errors. Other factors such as cloud top reflectance, thermal radiance, orbital drift and
retrieval algorithms are amongst the main sources (Kummerow et al., 2004). Moreover, short life span and sensor degradation cause
data inhomogeneity (Kummerow et al., 2004). Infrared sensors do not directly measure the precipitation. Instead, they capture the
cloud top brightness as a proxy for rainfall and may not be accurate (Kalinga and Gan, 2010), whereas microwave sensors estimate
relatively accurate rainfall but with a lower temporal resolution (Dinku et al., 2007; Joyce et al., 2004; Pan et al., 2010).
Compounding the problem further, previous studies have shown that there is significant inconsistency among the products over the
globe which makes their assessment over arid regions particularly important where the error can be larger than the observation. In
situ observations are not without their own limitations in certain environments (e.g., arid, complex terrain). Errors are found in these
environments given the uneven distribution of observational networks (Jeffery, 2006; Hughes, 2006; Beesley et al., 2009; Nicholson,
2013). The topography of the region also affects the representativeness of the rain gauge. For instance, rain gauges on mountainous
regions are not representative of the region’s precipitation due to the terrain complexity (Peng et al., 2014). Such deficiencies of rain
gauges observations lead to a rise in the popularity of remote sensing-based products (Sawunyama and Hughes, 2008; Li et al., 2009).
A common way to improve the estimation is by integrating both technologies in a comprehensive system that minimizes the effect of
their drawbacks (Boushaki et al., 2009).
One of the most commonly used precipitation products, the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite
Precipitation Analysis (TMPA), provides near global products at a spatial resolution of 0.25° and a temporal scale of three hours
(Huffman et al., 2007). However, several studies demonstrated that TRMM products exhibit significant systematic bias. For example,
the TRMM 3B42 product overestimated precipitation over the Zambezi River Basin (Cohen Liechti et al., 2012) and the Tibetan
Plateau in China (Yin et al., 2008), but underestimated it in Nepal (Islam, 2009). Moreover, TRMM 3B43 monthly product failed to
show significant linear correlation with the rain gauges in eastern Kyrgyzstan (Karaseva et al., 2012). TRMM products were found to
be well in line with the rain gauge observations in west Africa according to Jobard et al. (2011). Recently, Milewski et al. (2015)
assessed the four TMPA products: 3B42: V6, V7temp, V7, RTV7 using a rain gauge network in Northern Morocco. They found the
latest product, 3B42 V7, to be the most consistent with the gauge observations based on spatially interpolated Pearson correlation
coefficient results, while recording overestimations across all four products in arid environments. Building upon the success of
TRMM, the Global Precipitation Measurement mission, launched in February 2014, is providing higher temporal (30 min) and spatial
(0.1 arc-degree) resolution precipitation estimates through the IMERG product (Integrated Multi-satellitE Retrievals for GPM). The
GPM IMERG product intercalibrates, merges, and interpolates GPM constellation satellite microwave precipitation estimates with
microwave-calibrated infrared estimates, and rain gauge analyses to produce a higher resolution and more accurate product
(Huffman et al., 2014) The GPM core satellite estimates precipitation from two instruments, the GPM Microwave Imager (GMI) and
the Dual-Frequency Precipitation Radar (DPR). More importantly for this study, the GPM radar has been upgraded to two frequencies,
adding sensitivity to light precipitation. These sensors are upgrades over the TRMM sensors and promise the ability of increased
accuracy in the aforementioned terrains.
The Climate Prediction Center MORPHing technique (CMORPH) produces global precipitation analysis at high spatial and
temporal resolutions with exclusive reliance on low orbiter satellite microwave observations. The daily CMORPH product
overestimated precipitation when evaluated against the rain gauge observations in the Zambezi River Basin (Cohen Liechti et al.,
2012), in West Africa (Jobard et al., 2011), and in Indonesia (Vernimmen et al., 2012). It also showed significant underestimation
when examined over the Ethiopian river basins (Bitew and Gebremichael, 2011; Romilly and Gebremichael, 2011). CMORPH was
found to perform well over the US, Australia, and Northwestern Europe according to Ebert et al. (2007).
Satellite-based products have shown uncertainty throughout the world. They seem to be time and space sensitive as their
reliability varied over different regions and during different seasons. This creates a potential bias in water resource assessments and
investigations when integrating such products in hydrological models or studies. Significant disagreement with the rain gauges was
noticed in arid regions compared to mid-latitude regions in the world (Fekete et al., 2004). Assessment of these precipitation products
was rarely performed over arid and semi-arid regions despite their abundance (arid areas cover almost a quarter of the world). Fekete
et al. (2004) stated that precipitation estimates over arid and semi-arid areas require significant improvements. The uncertainty in
estimating precipitation is causing significant fluctuations in the hydrologic processes, such as runoff, due to the nonlinear
relationship. The PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks) product
developed by Sorooshian et al. (2014) and the CMORPH product were found to overestimate the precipitation over the arid regions of
northwest China, while the TRMM daily product 3B42 performed relatively better with acceptable precision, but a correction factor
was suggested before application (Yang and Luo, 2014). Recently, Liu (2015) compared TRMM, GPCC (Global Precipitation Climate
Center), and WM (Willmott et al., 2001) rainfall products overland on the global scale. The products showed climatology agreement
in terms of spatial, distribution, zonal means as well as seasonal variations, but large discrepancies were found in light rain.
Precipitation retrieval over arid regions has been challenging because of uncertainties in estimating land surface properties,
mainly land surface emissivity. Yan and Weng (2011) demonstrated lower accuracy in the Global Forecast System (GFS) system over
arid regions. Also, Zheng et al. (2012) showed that over the desert, there is a cold and significant bias during daytime in estimating

123
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

the Land Surface Temperature (LST) using satellite. The main cause of this error is the uncertainty in the assumption of the emissivity
of the surface properties, attributed to the lack of knowledge of soil types and their dielectric properties over arid regions. This could
be reinforced by accurate measurement of the surface emissivity over arid regions. Moreover, the new vegetation-dependent
formulations of momentum and thermal roughness lengths could be used to reduce the uncertainty of the satellite products in arid
regions (Zheng et al., 2012). The technological advancement of sensors and their capabilities prepared the groundwork for such
improvements of global precipitation products which may make remote sensing-based ones a viable option (Boushaki et al., 2009;
Milewski et al., 2015). However, such advancement first requires an accurate assessment of the available products over arid regions
and the examination of their inconsistency with respect to land surface properties.
This study puts a specific focus on the United Arab Emirates (UAE) and analyzes the consistency of precipitation products by inter-
comparing available retrievals and cross-examining them against in situ observations. The goal is to understand the spatial and
temporal variability in the uncertainty of precipitation retrievals within the context of aridity, topography, and land cover across the
UAE. Four global products were selected based on the dual criteria of their significant overlapping coverage period (2000–2010) and
their distinct mechanisms for precipitation measurement, including: ground-based (WM) merged with model simulations (GPCC),
microwave remote sensing (CMORPH), and blended microwave-radar retrievals (TRMM). An observation network of 53 overland
rain gauges across the UAE were analyzed to quantitatively assess the accuracy of the products and understand the discrepancies. The
determination of the inconsistency among the products and their bias variability in time and space should lay the framework for the
development of a blended precipitation product that accounts for their individual limitations.

2. Description of study area and data

2.1. Study area

The statistical analysis between the satellite-based precipitation products and the in situ observations was performed over the
UAE, which is generally a semi-arid to arid area located in the northwestern part of the Arabian Peninsula (22°N–26°N, 51°E–57°E)
(Fig. 1). The country has a coastal stretch of 650 km along the Arabian Gulf and 90 km along the Gulf of Oman. The topography of the
area is mostly flat in the coastal and western parts with Al Hajar Mountain chain in the northeastern part of the country reaching an
altitude of 1800 m (see Fig. 1a) according to the 30-m resolution ASTER digital elevation model (Abrams, 2000). Consistently, a 30-m
resolution land cover map over the UAE (see Fig. 1b) was derived from the National Geomatics Center of China global land cover
product with an overall accuracy of 85% (Jun et al., 2014). The western part of the country is mostly sand dunes and merges into the
Empty Quarter of Saudi Arabia. The land cover is uniform, being predominantly classified as bare land (rural) in the western part of
the country, and artificial surfaces (urban), particularly in the emirates of Dubai and Abu Dhabi, with some dense shrub land to the
east of Dubai (see Fig. 1b). The climate of the UAE is generally dry and hot with the average maximum temperatures above 40 °C in
during summer (May–August). The highlands in the northern part of the country experience relatively cooler temperatures with near
freezing temperatures during the winter. Humidity can reach as high as 95% during late summer due to humid southeasterly winds.

2.2. Datasets

Four global precipitation data products were used in this study, namely, the Global Precipitation Climatology Centre (GPCC),
Willmott and Matsuura (WM), TRMM, and CMORPH precipitation products. The products were derived independently from a
combination of in situ observations, remote sensing data, and numerical weather prediction models. A mask was applied to the
studied products to extract estimates over the study domain.

2.2.1. GPCC precipitation product

The GPCC data product is proposed by the World Metrological Organization (WMO). The dataset was amassed from available rain
gauges from approximately 48,000 land stations and 6000–7000 ocean stations. The product has a spatial resolution of 0.5° at a
monthly time step and includes data from low noise infrared radiometer for measuring the atmospheric and surface precipitation
(Schneider et al., 2011).

2.2.2. WM precipitation product

The Willmott and Matsuura product was developed by the Department of Geography at the University of Delaware. This station-
based time series was drawn primarily from recent versions of the Global Historical Climatology Network (GHCN version 2) and the
Global Surface Summary of Day (GSOD) archive from a total of 16,500 stations. The data have a spatial resolution of 0.5° with a
global coverage. Time series data from 1900 up to 2010 is available at a cumulative monthly interval (Willmott et al., 2001).

2.2.3. TRMM 3B42 precipitation product

A wide range of TRMM products are available depending on user application and interest. The extent of the coverage is on the
mid-latitude from 50°N to 50°S and 180°W to 180°E. The satellite contains a microwave scanner and infrared imager to estimate the
precipitation based on atmospheric properties (e.g., cloud top temperature, ice particles, brightness temperature). For this study, the
daily 3B42 product with a spatial resolution of 0.25° by 0.25° was selected (Huffman et al., 2010).

124
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Fig. 1. Rain gauge network versus topography (a) and land cover (b).

2.2.4. CMORPH precipitation product

The Climate Prediction Center morphing method (CMORPH) products are derived from low-orbit microwave satellites, including
SSM/I, AMSU-B, and AMSR-E, and propagated using the data obtained from an IR of a geostationary satellite with a spatial coverage
of 60°N-S (Joyce et al., 2004). In the latest CMORPH Version 1.0, satellite estimates are bias-corrected using daily rain gauge analysis
(Xie et al., 2011). Three spatial and temporal resolutions are available: 8 km–30 min, 0.25°-3 hourly, and 0.25°-daily. In this study,
the 0.25°-daily bias-corrected Version 1.0 CMORPH data were used.

2.2.5. Rain gauge data

The rain gauge measurements were obtained from daily ground observations recorded by the UAE National Center of Meteorology
and Seismology (NCMS) over the time period of 2003–2010, inclusive. Their readings have not yet been made publically available,
and therefore were not incorporated in both the gauge-based WM and gauge-simulated GPCC products. The rain gauges are
adequately distributed across the country (see Fig. 1). It should be noted that the 53 overland stations were exclusively used in this
study.

3. Methodology

The daily TRMM 3B42 (version 7) and CMORPH products were accumulated to a monthly temporal resolution, while spatial

125
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

resampling of the GPCC and WM products to a 0.25° spatial resolution was carried out through global least square surface fitting. The
2000–2010 (inclusive) time span was selected as an overlapping temporal coverage across all four products for their intercomparison,
while varying time periods in the range of 2003–2010 were subsetted for the comparisons with the network observations based on the
available record of each station. On average, a 6-year coverage period was used for the validation.
Root Mean Square Error (RMSE) and relative mean BIAS (rBIAS) methods of evaluation were implemented with the observed
value representing the rain gauge measurement (see Eqs. (1) and (2)). The assumption that the rain gauges are the true measurements
may not be ideal but it is widely assumed in various studies. The RMSE and rBIAS reflect the average error and degree of over- or
underestimation of the product in estimating the rainfall.
n
∑i=1 (yoi − yest i)2
RMSE =
n (1)
n
∑i=1 (yoi − yest i)
rBIAS = n × 100
∑i=1 yoi (2)

where, yesti and yoi are the estimated (product) and observed rainfall, respectively, at rain gauge i. n is the number of observations.
Another essential statistical tool for analyzing the performance of a remote sensing precipitation retrieval is the Probability of
Detection (POD). The POD measure reflects the instrument’s efficacy of capturing the rain events regardless of their magnitude. Also,
its counterpart, the False Alarm Ratio (FAR) describes the fraction of the false events out of the total detection by the instrument.
Daily observations of the TRMM and CMORPH products were used for the POD and FAR (see Eqs. (3) and (4)). The Mean Relative
Difference (MRD) is another metric that serves the precision analysis of the precipitation products among each other (see Eq. (5)). It
shows clearly how individual product estimations are related to the ensemble of the other products’ estimations.
rainy events detected by both rain gauge and product
POD =
rainy events detected by rain gauge (3)

events only detected by the product

FAR =
total events detected by the product (4)

yi − yrest
MRDi =
y (5)

The Pearson correlation (r) (Benesty et al., 2009) and Nash-Sutcliffe Efficiency (NSE) coefficients (Moriasi et al., 2007) were also
used in relating the observed values and product estimates (see Eqs. (6) and (7)). The Pearson correlation reflects the statistical
association between variables, and can range between −1 to 1, where 1 is total positive linear correlation, 0 is no linear correlation,
and −1 is total negative linear correlation. The Nash–Sutcliffe efficiency can range from −∞ to 1, where the closer the efficiency is
to 1, the more accurate the estimate. The obtained station correlations with each product were spatially interpolated using the global
method of kriging. It is expected that variations in elevation and land cover across the country will impact the spatial distribution of
the agreement. Finally, regression analysis was carried out for the TRMM 3B42 daily product for its correction against the daily
recorded data over the UAE.
n
∑i=1 (yo i − yo)(yest i − yest)
r= n n
∑i=1 (yo i − yo)2 × ∑i=1 (yest i − yest)2 (6)
n
∑i=1 (yo i − yest i)2
NSE = 1 − n
∑i=1 (yoi − yo)2 (7)

4. Results and discussion

4.1. Consistency of products with rain gauges

Fig. 2 shows the RMSE and rBIAS boxplots of the products versus the station observations. The lowest average RMSE of 10.5 mm
was found with the TRMM product, indicating a higher agreement with in situ observations over all stations, followed by 12.2 mm
with the GPCC product and 12.4 mm with the CMORPH product, while the WM product recorded the highest average RMSE of
15.9 mm. The correlations were in line with the previous RMSE results in terms of the agreement ranking with the network
observations: TRMM, GPCC, CMORPH, and WM in descending order. However, the RMSE results simply indicate the error magnitude
without describing its relative significance to the size of the actual amount. Having a lower RMSE does not guarantee a product to be
applicable which suggests examining the rBIAS. According to the rBIAS evaluation criteria there are three categories for the
performance of precipitation products: underestimation for rBIAS < −10%, overestimation for rBIAS > 10%, and approximately
equal i.e. acceptable range for −10% < rBIAS < 10% (Brown, 2006). The rBIAS results showed an overestimation of the observed

126
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Fig. 2. The RMSE and rBIAS boxplots of the products against the network observations.

rainfall values across the GPCC, WM, and TRMM products with similar averages of around 35% (0.35), while the CMORPH product
showed an underestimation with an average of −41% (−0.41). Also, the CMORPH product showed the most significant relative bias
spatial variation ranging from −2.93 in the internal western area of Owtaid, Abu Dhabi to 0.89 in the northeastern area of Umm al-
Qaiwain, indicating a high level of stochastic uncertainty across the stations.
The r and NSE coefficient boxplots are shown in Fig. 3. A correlation of a precipitation product versus rain gauge observations is
considered statistically significant if the coefficient is greater than 0.7 (Condom et al., 2011). The highest average correlation of 0.75
was recorded with the TRMM product ranging from 0.57 in Al Gheweifat station to 0.93 in Falaj Al Moalla station (in addition to five
outliers below 0.5), followed closely by averages of 0.74 with the GPCC product and 0.71 with the CMORPH product, while the WM
product recorded the lowest correlation of 0.57. Similarly, the highest average NSE value (0.47) was recorded by the TRMM and
GPCC products with an equal maximum of 0.81 and minimums of −0.88 and −1.1, respectively. The CMORPH product followed
with an average NSE of 0.31, while the WM product recorded a null average NSE. Given the difference in spatial scales, the frequently
low expected NSE values (more than 40% below 0 for each product) were excluded from further analyses.
Generally, both satellite products (TRMM and CMORPH) showed high ranges of POD values across the UAE. The average POD of
the CMORPH product was found to be 0.66, which was higher than the 0.5 POD of the TRMM product (Fig. 4). The average FAR
values were found to be 0.62 and 0.67 for the TRMM and CMORPH, respectively. The FAR results closely matched the 0.62 and 0.66
FAR values obtained over neighboring Saudi Arabia by Kheimi and Gutub (2012). When revising the computation with a 3 mm daily
rainfall threshold, the FAR improved to 0.39 and 0.24 for TRMM and CMORPH, respectively. Alternatively, the obtained FAR values
were significantly higher than those reported by Su et al. (2008) over the Plata Basin (0.1–0.3). More importantly, they found the FAR
was increasing with increasing thresholds; opposite to the result found here, which may be attributed to the contradicting wet and dry
climatology of the domains. Hence, on average, more than 39% of the reports by TRMM and more than 24% of the reports by
CMORPH were viewed to be false.
Spatially, the products were more accurate in capturing inland rainfall, while their estimations for coastal locations were found to be the
less accurate. The coastal location could have negatively impacted the predictability of the rainfall, especially when microwave data are used
in the retrieval as in the case of TRMM estimates. Rain gauge observations are often considered to be very accurate and regarded as the actual
measurements. Despite these considerations, there are several studies that show the ineffectiveness of the rain gauge measurements,
particularly, in hyper-arid areas where with recurring sand storms, they are often filled with sand. Another important factor is the evaporation
since there is a time gap between the time of the event and the time of the reading which doesn’t account for evaporative losses. Considering
the associated inaccuracies, it is essential to check the consistency of the products among each other for further analysis.

Fig. 3. Pearson correlation and Nash-Sutcliffe Efficiency (NSE) coefficients.

127
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Fig. 4. POD and FAR results for TRMM and CMORPH satellite products.

4.2. Intercomparison of the products

A calibration of TRMM with rain-gauge measurements over Saudi Arabia for the period 1998–2009 resulted in a 0.9 correlation
coefficient with a 99% significance level on a monthly scale, indicating the reliability of TRMM usage in water-related studies in the
region (Almazroui, 2011). EL Kenawy and Mccabe (2015) studied rainfall climatology, anomalies, and trends over Saudi Arabia using
gauge based products, namely, APHRODITE, GPCC, PRINCETON, UDEL, CRU and PREC/L. Their work suggested that the monthly
GPCC showed the most accurate rainfall estimates over the regional scale of Saudi Arabia. In this context, strong agreement is
expected between TRMM and GPCC over the neighboring UAE.
Fig. 5 shows the scatterplots and obtained correlations between the products at gauge locations throughout the study period of
2000–2010. The TRMM and GPCC products showed the highest agreement with an overall correlation coefficient of 0.6, which
exceeded, by far, the remaining correlations of 0.23 and 0.1 between TRMM and CMORPH and TRMM and WM products,
respectively. The relationship between TRMM and GPCC products was found to be significant ranging from 0.6–0.971 on a station-
per-station basis. Moreover, another satellite product, CMORPH, was found to be relatively highly correlated with GPCC except over
the Dubai and eastern UAE stations where the correlation is very low from 0.41 to 0.277. The correlation between the two ground
measurement products GPCC and WM was not significant across most of the stations. Also, the correlation between the two satellite
products TRMM and CMORPH was found to be inconsistent across the cities ranging from 0.328 around Fujairah to as high as 0.878
in Ras Al Khaimah.
The mean relative difference (MRD) was computed as a spatial average representing six distinct Emirates and cities across the
UAE, namely, Abu Dhabi, Dubai, Al Ain, Fujairah, Ras Al Khaimah, and Sharjah. Table 1 shows the MRD of the products across the
UAE.

Fig. 5. Product scatterplots and correlations across the UAE gauge network.

128
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Table 1
Mean Relative Difference results averaged over different cities.

GPCC WM TRMM CMORPH

Abu Dhabi −1.41 0.45 4.12 −3.23

Dubai −6.82 −6.1 −2.4 15.1
Al Ain 0.98 1.91 −0.41 −2.49
Fujairah −3.12 −4.62 −4.98 12.61
Ras Al Khaimah −2.12 2.88 1.51 −2.27
Sharjah 0.65 3.47 0.55 −4.67

In Abu Dhabi, Al Ain, Ras Al Khaimah, and Sharjah the products tend to be precise with relatively small variation in their
estimation. CMORPH performed poorly in the four cities reaching as low as 4.67 below the average, however, the highest is obtained
by TRMM for Abu Dhabi and WM for the remaining three. MRD of the three products, with the exception of CMORPH, showed good
precision with the highest deviation occurring in Ras Al Khaimah. However, the introduction of the CMORPH product produced
much uncertainty in the MRD of the products especially for the cities of Dubai, Fujairah and Sharjah. The worst precision in the
products occurred in Dubai and Fujairah.
The discrepancies between the precipitation products is primarily attributed to their distinct observation techniques, retrieval
mechanisms, and calibration approaches. First, the sensors used in the retrieval of precipitation products are on different satellite
platforms with different overpass times and observation geometries such as satellite, zenith, and solar angles. Second, the retrieval
mechanisms depend on the wavelengths and frequencies which exhibit varying sensitivity to water vapor. Microwave-based retrieval
uses a single channel (e.g. the 22 GHz onboard AMSR-E and AMSR2) or multiple frequencies that sound the atmosphere to determine
water vapor profiles. The microwave-based retrieval algorithms rely on overland ice-scattering techniques, hence the warm-rainfall
processes without significant ice over the UAE result in the underestimation of the gauge-adjusted microwave algorithms in the
CMORPH product (Nesbitt et al., 2008). The infrared-based retrieval utilizes the top-of-cloud temperature which indicates their
vertical development. It is expected that the difference in the physics involved in the precipitation retrievals using microwave (active
or passive) and infrared signals should lead to discrepancies in the obtained products. On the other hand, the TMPA research product
(e.g., 3B42.V7) algorithm merges ground observations (e.g, gauge-based GPCC) with both microwave and infrared satellite
observations. The variable number of stations used per grid in the GPCC product impacts its quality. The improved performance of
the GPCC product over the WM product (both gauge-based) is due to its higher gauge density than that of the GHCN (v2), especially
over northeastern UAE (Schneider et al., 2014).

4.3. Spatial distribution of rainfall across the UAE

The average annual rainfall distribution across the UAE derived from the rain gauges showed a consistent spatial pattern with that
of the GPCC, TRMM, and CMORPH – derived rainfall distributions, with the northeastern part of the country recording the higher
amounts and the western part recording the lowest (see Fig. 6). Conversely, the rainfall distribution derived from the WM product
showed higher rainfall amounts in the west and minimum rainfall in the northeastern section. This is quite opposite to the reality
reflected by the rain gauges, where the western part receives the least rainfall. The higher rainfall recorded along the windward sides
of the mountainous northeastern tips of the country is primarily attributed to the approaching local microscale weather processes
triggering isolated orographic rainfall events (Niranjan Kumar and Ouarda, 2014).
The satellite-based precipitation products TRMM and CMORPH and the GPCC gauge-based product showed a near similar rainfall
distribution pattern except in some coastal parts of the country (see Fig. 6). The TRMM distribution indicated the peak rainfall
occurring in the coastal (Arabian Gulf) areas of the country with its paramount in the northern regions. The variation was found to be
the highest for the CMORPH product ranging from 30 to 120 mm per year across the country. In contrary, the lowest range was
presented by the WM product from 70 to 106 mm per year.

4.4. Variation of spatially interpolated correlations with elevation and land cover

The variation in the agreement between the products and the gauge observations was examined by spatially interpolating the
station-based correlations over the UAE study area as presented in Fig. 7. As expected, the TRMM product maintained the highest
spatial correlations across the study area since it recorded the highest station correlations as previously denoted, followed by the
GPCC, CMORPH, and WM products, with the exception of CMORPH performing better than GPCC over the western part. The
correlations varied from 0.17 to 0.88, with the higher values in the northeastern highlands (> 250 m) and lower values in the
western flat (< 250 m), coastal, and inland areas, consistently for all four products. This characteristic is opposite in part to the
TMPA-gauge correlation variations reported by Milewski et al. (2015) over Morocco. They recorded a decrease in correlation with an
increase in elevation due to the high sensitivity of the passive microwave sensors to snow content, which is absent in the case of the
UAE; however, they too reported high correlations in areas with elevations less than 1000m, which is generally the case of the present
topography.
The relatively low correlations with the TRMM product recorded by stations in the predominantly bare western part of the
country (see Fig. 1b), such as Al Jazeera (0.38), Owtaid (0.69), Mukhairiz (0.61), Harnim (0.7), Bu Hanrah (0.7) and Razeen (0.68)

129
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Fig. 6. Average annual rainfall (mm) distribution over the UAE derived from rain gauges and precipitation products (2003–2010).

can be attributed to the low or almost absent rainfall events coupled with the notable false alarm ratio (FAR) of 0.39 for the TRMM
product. Also, the retrieval algorithm may not be successful at filtering out any emissivity over the bare sandy lands (Almazroui,
2011; Chen et al., 2013; Grody and Weng, 2008). Higher significant correlations, exceeding 0.8, were consistently recorded across the
products at stations such as Khazna, Jabal Hafeet, and Al Malaiha (northeast) where a dense mixture of forested and cultivated land
cover exists (see Fig. 1b). Wulfmeyer et al. (2014) and Milovac et al. (2016) used the Weather Research and Forecasting (WRF) model
to study the impact of land use change on convection and precipitation over arid regions, including Oman. They concluded that large
arid plantations could lead to convection initiation and enhance rainfall likelihood, however, the magnitude of such events remains
governed by synoptic conditions in the region. Thus, planation placement has the potential not only to mitigate climate change
through storage of atmospheric CO2, but also to foster precipitation in arid regions and reverse desertification. The coupled result of

130
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Fig. 7. Station-product correlation variation across the UAE.

higher rainfall rates and gauge correlations over such regions in the UAE corroborates the speculation of rain enhancement by the
introduction of vegetation and forested land to the currently predominant (∼90%) bareland cover.

4.5. TRMM 3B42 calibration over the UAE

The result of the relative bias (rBIAS) analysis strongly supports the need for a correction procedure, even for TRMM (the most
highly correlated product). Given the convenient temporal and spatial scale and strong correlation with the ground readings, the
TRMM 3B42 product was identified as the most suitable remote sensing precipitation product for the UAE. However due to the
overestimation of the product, it should be corrected using statistical methods before application. This conclusion strongly supports
the recommendation made by Yang and Luo (2014) after evaluating satellite precipitation products in an arid region of China. Since
the TRMM 3B42 product showed strong correlation coefficients, reaching 0.93 with the station observations, linear regression would
be the best option to adjust the overestimation of the product. For each gauge, 70% of the daily records were used to calibrate the
retrievals, while the remaining 30% were used to validate the correction. A linear regression equation has the form y = c + mx (see
Eq. (8)), where the m and c represent the slope and constant, respectively.
TRMMcorrected = c + m × TRMMuncorrected (8)

The correction factors c and m for each station are shown in Table 2 below:
By referring to the last row of Table 2, the averaged TRMM overestimation correction over the UAE can be expressed as:
TRMMcorrected = −0.03 + 0.9 × TRMMuncorrected (9)

The validation of the obtained correction was verified by comparing the TRMM 3B42 averages over the UAE to the spatially
averaged monthly gauge observations before and after correction over the period of 2003–2010. Fig. 8 shows the scatter plots of the
monthly averages before and after the application of linear correction. The correlation coefficient improved from an insignificant
0.51 value to a significant 0.77 value. Overall, the correction performed well on the scale of the UAE; however, locally it may not be
as effective, especially around stations with insignificant slope coefficients (Table 2). Almazroui (2011) calibrated the TRMM 3B42
over the neighboring country of Saudi Arabia using monthly rain gauge observations for the period of 1998–2009. They found that
the rainfall over Saudi Arabia is recommended as being the TRMM recorded value multiplied by 0.93 subtracted from 0.04, which are
reasonably close to correction factors of 0.9 and 0.03 found in the present study (see Eq. (9)).
It is worth noting that future work should focus on the analysis of the consistency between precipitation products and other
satellite-based proxies like soil moisture (Fares et al., 2013; Temimi et al., 2014), land surface temperature (Notarnicola et al., 2012),

131
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Table 2
Correction factors per station for the TRMM 3B42 product versus daily observations.

Station Lat (oN) Lon (oE) c (mm) m (mm−1) Statistical Significance

'Abu Dhabi' 24.48 54.33 −0.01 0.89 S

'Al Faqa' 24.72 55.62 −0.06 0.85 S
'Al Jazeera B.G' 23.29 52.29 −0.03 0.90 S
'Al Shiweb' 24.78 55.80 −0.05 0.88 S
'Al Arad' 23.84 55.52 −0.02 1.00 S
'Al Foah' 24.33 55.81 0.00 0.97 S
'Al Gheweifat' 24.12 51.12 −0.06 0.77 I
'Al Khazna' 24.12 55.11 −0.03 0.79 I
'Manama' 25.39 56.01 −0.05 0.88 I
'Mezyed' 24.03 55.85 −0.01 0.99 S
'Masafi' 25.45 56.12 −0.03 0.96 S
'Mezaira' 23.15 53.78 −0.06 0.94 S
'Madinat zayed' 23.68 53.70 −0.03 0.74 I
'Makassib' 24.67 51.82 −0.05 0.86 S
'Harnim' 22.97 54.30 −0.02 0.79 I
'Hatta' 24.81 56.14 −0.03 0.83 S
'Jabal Hafeet' 24.06 55.78 −0.06 1.00 S
'Falaj Al Moalla' 25.34 55.85 −0.03 1.00 S
'Dalma' 24.49 52.29 −0.05 0.97 S
'Das Island' 25.15 52.87 −0.02 0.89 S
'Dhudna' 25.51 56.32 −0.06 0.93 S
'Bu Hanrah' 23.51 54.53 −0.03 0.94 S
'Al Qlaa' 24.16 52.98 −0.05 0.99 S
'Owtald' 23.40 53.10 −0.02 0.93 S
'Khatam AlShaklah' 24.21 55.95 −0.05 0.91 S
'Qarnen' 24.94 52.85 −0.02 0.88 S
'Mukhariz' 22.93 52.88 −0.06 0.87 S
'Razeen' 23.68 54.75 0.02 0.86 S
'Raknah' 24.35 55.71 0.01 0.99 S
'Saih Al Salem' 24.83 55.31 −0.05 0.98 S
'Sham' 26.03 56.09 −0.02 0.94 S
'Sir BaniYas' 24.32 52.60 −0.06 0.90 S
'Um Azimul' 22.71 55.14 0.02 0.99 S
'Jabal Mebreh' 25.65 56.13 −0.05 0.80 I
'Umm Al Quwaln' 25.53 55.66 0.01 0.81 I
'Tawiyen' 25.56 56.07 −0.05 0.86 S
'Swiehan' 24.47 55.33 0.01 0.89 S
'Al Ain' 24.22 55.79 −0.01 0.98 S
'Al Qattara' 24.27 55.76 −0.05 0.90 S
'Sir Bu Nair' 25.22 54.23 −0.02 0.88 S
'Um Ghafa' 24.07 55.93 −0.05 0.78 S
'Damsa' 24.18 55.41 −0.02 0.80 S
'Al Aryaro' 24.31 54.17 −0.06 0.89 S
'Abu Al Abyad' 24.21 53.79 0.02 0.93 S
'Ras Ghanadah' 24.84 54.77 0.01 0.94 S
'Al Rowdah' 24.10 55.54 −0.05 0.84 I
'Yasat' 24.17 51.99 −0.02 0.93 S
'Abu Al Bukhoosh' 25.50 53.15 −0.06 0.91 S
'Ras Musherib' 24.33 51.15 0.02 0.90 S
'Jabal Jais' 25.95 56.17 −0.05 0.99 S
'Al Wathbah' 24.18 54.70 0.01 0.80 I
'Al Malaiha' 25.13 55.89 −0.05 0.81 I
'Al Quaa' 23.39 55.42 0.01 0.90 S
Coefficient Averages −0.03 0.90

S: Significant linear coefficient (m) at the 5% significance level.

I: Insignificant linear coefficient (m) at the 5% significance level.

evaporation (Rabin et al., 2014), or land surface emissivity (Norouzi et al., 2015). In the absence of a dense vegetation cover in arid
regions which could impact the retrieval of surface parameters, such comparison should be relevant. Precipitation products can also
be compared with the Gravity Recovery and Climate Experiment (GRACE) observations (Swenson, 2012) to assess the agreement
between recorded rainfall and change in water storage (Ghebreyesus et al., 2016).The comparison should be carried out over a larger
geographic domain given the coarse resolution GRACE water storage anomalies.

132
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Fig. 8. Scatter plots of the monthly spatial averages (mm) of the TRMM 3B42 product against network observations before and after correction.

5. Conclusion

The rain gauge-based products (e.g., GPCC, WM) performed poorly with the lack of adequate field observation, as is the case with
the WM data set. However, GPCC measurements were improved for the UAE, primarily in the northeastern quadrant, given the
inclusion of additional data resources (∼5:1 ratio) as compared to the WM product (Schneider et al., 2014). The satellite-based
products (e.g., TMPA 3B42v7, CMORPH) have their advantages in certain areas across the globe, however TMPA performs the best
over the UAE. This is most likely attributed to the incorporation of the GPCC data into the research product as well as the additional
multi-satellite retrievals as compared to the CMORPH product.
The elevated (e.g., > 250 m) and forested areas in the northeastern part of the UAE showed better agreement between the
products and the station observations as opposed to the western bare lands and coastal locations. These results were attributed to the
(i) local orographic rainfall events in the mountains coupled with convection initiation from vegetated land-atmospheric feedback
and (ii) emissivity filtering-based anomalies in the retrieval algorithms over bare and sandy land in Southwestern UAE, particularly in
the case of the TRMM product. Although the result of the TRMM correction showed a significant improvement in the monthly
correlation over the UAE from 0.51 to 0.77, it was not a perfect fit over the entire country. This is most likely the result of a number of
concurrent factors such as data scarcity (Southwest UAE – Fig. 7), anomalous climatic events in 1999 (not discussed in this study),
difference in spatial extents between satellite products and observation network, land surface emissivity bias in deserts, bright band,
ground clutter, or the attenuation of the PR reflectivity, among other factors. However, the density of the rain gauge network
(scarcity of data) is a key factor which determines the accuracy of the results. Thus, it is recommended to apply a bias correction
before the use of the products to systematically correct the error and provide more accurate precipitation estimates.

Funding

This study was conducted within the framework of a project funded by Masdar Institute of Science and Technology, entitled,
“Hydrological Modeling and Remote Sensing for Flood Risk Assessment in the UAE” under grant number SSG2015-000060.

Authors contributions

Youssef Wehbe and Dawit Ghebreyesus carried out the methodology and wrote the article; Prof. Marouane Temimi directed the
progress and analyzed the results; Prof. Adam Milewski revised figures, provided writing assistance, helped evaluate the results, and
proof-read the article; Dr. Abdulla Al Mandous revised the manuscript and ensured data quality control of the rain gauge
measurements.

Acknowledgements

We thank the anonymous reviewers for their careful reading of our manuscript and their many insightful comments and
suggestions.
The authors would like to thank the UAE National Center of Meteorology and Seismology (NCMS) for the rain gauge observations
used in this study. The authors are also thankful to the providers of the CMORPH (CPC MORPHing technique) High-resolution
precipitation product (60S-60N) and the Global (land) precipitation and temperature product from Wilmott & Matsuura, University of
Delaware. The Tropical Rainfall Measurement Mission (TRMM) 3B42 data were provided by the NASA/Goddard Space Flight Center
and archived at the NASA GES DISC. The Global Climatology Precipitation Center (GPCC) data were obtained from the NOAA/OAR/
ESRL PSD, Boulder, Colorado, USA.

133
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

References

Abrams, M., 2000. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER): data products for the high spatial resolution imager on NASA's
Terra platform. Int. J. Remote Sens. 21, 847–859.
Almazroui, M., 2011. Calibration of TRMM rainfall climatology over Saudi Arabia during 1998–2009. Atmos. Res. 99, 400–414.
Awange, J.L., Ferreira, V.G., Forootan, E., Khandu, Andam-Akorful S.A., Agutu, N.O., He, X.F., 2016. Uncertainties in remotely sensed precipitation data over Africa.
Int. J. Climatol. 36, 303–323.
Beesley, C., Frost, A., Zajaczkowski, J., 2009. A comparison of the BAWAP and SILO spatially interpolated daily rainfall datasets. In: 18th World IMACS/MODSIM
Congress. Cairns, Australia. pp. 17.
Benesty, J., Chen, J., Huang, Y., Cohen, I., 2009. Pearson correlation coefficient. Noise Reduction in Speech Processing. Springer.
Bitew, M., Gebremichael, M., 2011. Assessment of satellite rainfall products for streamflow simulation in medium watersheds of the Ethiopian highlands. Hydrol. Earth
Syst. Sci. 15, 1147–1155.
Boushaki, F.I., Hsu, K.-L., Sorooshian, S., Park, G.-H., Mahani, S., Shi, W., 2009. Bias adjustment of satellite precipitation estimation using ground-based measurement:
a case study evaluation over the southwestern United States. J. Hydrometeorol. 10, 1231–1242.
Brown, J.E.M., 2006. An analysis of the performance of hybrid infrared and microwave satellite precipitation algorithms over India and adjacent regions. Remote Sens.
Environ. 101, 63–81.
Chen, S., Hong, Y., Cao, Q., Gourley, J.J., Kirstetter, P.E., Yong, B., Tian, Y., Zhang, Z., Shen, Y., Hu, J., 2013. Similarity and difference of the two successive V6 and V7
TRMM multisatellite precipitation analysis performance over China. J. Geophys. Res.: Atmos. 118.
Cohen Liechti, T., Matos, J., Boillat, J.-L., Schleiss, A., 2012. Comparison and evaluation of satellite derived precipitation products for hydrological modeling of the
Zambezi River Basin. Hydrol. Earth Syst. Sci. 16, 489–500.
Condom, T., Rau, P., Espinoza, J.C., 2011. Correction of TRMM 3B43 monthly precipitation data over the mountainous areas of Peru during the period 1998–2007.
Hydrol. Processes 25, 1924–1933.
Dinku, T., Ceccato, P., GROVER-Kopec, E., Lemma, M., Connor, S., Ropelewski, C., 2007. Validation of satellite rainfall products over East Africa's complex
topography. Int. J. Remote Sens. 28, 1503–1526.
EL Kenawy, A.M., Mccabe, M.F., 2015. A multi-decadal assessment of the performance of gauge-and model-based rainfall products over Saudi Arabia: climatology,
anomalies and trends. Int. J. Climatol.
Ebert, E.E., Janowiak, J.E., Kidd, C., 2007. Comparison of near-real-time precipitation estimates from satellite observations and numerical models. Bull. Am. Meteorol.
Soc. 88, 47–64.
Fares, A., Temimi, M., Morgan, K., Kelleners, T.J., 2013. In-situ and remote soil moisture sensing technologies for vadose zone hydrology. Vadose Zone J. 12.
Fekete, B.M., Vörösmarty, C.J., Roads, J.O., Willmott, C.J., 2004. Uncertainties in precipitation and their impacts on runoff estimates. J. Clim. 17, 294–304.
Ghebreyesus, D.T., Temimi, M., Fares, A., Bayabil, H.K., 2016. A multi-satellite approach for water storage monitoring in an arid watershed. Geosciences 6, 33.
Grody, N.C., Weng, F., 2008. Microwave emission and scattering from deserts: theory compared with satellite measurements. IEEE Trans. Geosci. Remote Sens. 46,
361–375.
Huffman, G.J., Bolvin, D.T., Nelkin, E.J., Wolff, D.B., Adler, R.F., Gu, G., Hong, Y., Bowman, K.P., Stocker, E.F., 2007. The TRMM multisatellite precipitation analysis
(TMPA): quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeorolo. 8, 38–55.
Huffman, G.J., Adler, R.F., Bolvin, D.T., Nelkin, E.J., 2010. The TRMM multi-satellite precipitation analysis (TMPA). Satellite Rainfall Applications for Surface
Hydrology. Springer.
Huffman, G.J., Bolvin, D.T., Braithwaite, D., Hsu, K., Joyce, R., Xie, P., 2014. NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for
GPM (IMERG).
Hughes, D., 2006. Comparison of satellite rainfall data with observations from gauging station networks. J. Hydrol. 327, 399–410.
Islam, D., 2009. Analysis of TRMM data in monitoring rainfall over mountainous regions. SMRC Rep. 27.
Jeffery, S., 2006. Error Analysis for the Interpolation of Monthly Rainfall Used in the Generation of SILO Rainfall Datasets. Technical report, the Queensland
Department of Natural Resources.
Jobard, I., Chopin, F., Bergès, J.C., Roca, R., 2011. An intercomparison of 10-day satellite precipitation products during West African monsoon. Int. J. Remote Sens. 32,
2353–2376.
Joyce, R.J., Janowiak, J.E., Arkin, P.A., Xie, P., 2004. CMORPH: a method that produces global precipitation estimates from passive microwave and infrared data at
high spatial and temporal resolution. J. Hydrometeorol. 5, 487–503.
Jun, C., Ban, Y., Li, S., 2014. China: open access to Earth land-cover map. Nature 514 434–434.
Kalinga, O.A., Gan, T.Y., 2010. Estimation of rainfall from infrared-microwave satellite data for basin-scale hydrologic modelling. Hydrol. Process. 24, 2068–2086.
Karaseva, M.O., Prakash, S., Gairola, R., 2012. Validation of high-resolution TRMM-3B43 precipitation product using rain gauge measurements over Kyrgyzstan.
Theor. Appl. Climatol. 108, 147–157.
Kheimi, M.M., Gutub, S., 2012. Assessment of remotely-sensed precipitation products across the Saudi Arabia region. AGU Fall Meeting Abstracts 0448.
Kummerow, C., Poyner, P., Berg, W., Thomas-Stahle, J., 2004. The effects of rainfall inhomogeneity on climate variability of rainfall estimated from passive microwave
sensors. J. Atmos. Oceanic Technol. 21, 624–638.
Li, L., Hong, Y., Wang, J., Adler, R.F., Policelli, F.S., Habib, S., Irwn, D., Korme, T., Okello, L., 2009. Evaluation of the real-time TRMM-based multi-satellite
precipitation analysis for an operational flood prediction system in Nzoia Basin, Lake Victoria, Africa. Nat. Hazard. 50, 109–123.
Liu, Z., 2015. Evaluation of precipitation climatology derived from TRMM multi-satellite precipitation analysis (TMPA) monthly product over land with two gauge-
based products. Climate 3, 964–982.
Milewski, A., Elkadiri, R., Durham, M., 2015. Assessment and comparison of tmpa satellite precipitation products in varying climatic and topographic regimes in
morocco. Remote Sens. 7, 5697–5717.
Milovac, J., Branch, O.-L., Bauer, H.-S., Schwitalla, T., Warrach-Sagi, K., Wulfmeyer, V., 2016. High-Resolution WRF Model Simulations of Critical Land Surface-
Atmosphere Interactions Within Arid and Temperate Climates (WRFCLIM). High Performance Computing in Science and Engineerinǵ 15. Springer.
Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L., 2007. Model evaluation guidelines for systematic quantification of accuracy in
watershed simulations. Trans. Asabe 50, 885–900.
Nesbitt, S.W., Gochis, D.J., Lang, T.J., 2008. The diurnal cycle of clouds and precipitation along the Sierra Madre Occidental observed during NAME-2004:
Implications for warm season precipitation estimation in complex terrain. J. Hydrometeorol. 9, 728–743.
Nicholson, S.E., 2013. The West African Sahel: a review of recent studies on the rainfall regime and its interannual variability. ISRN Meteorol. 2013.
Niranjan Kumar, K., Ouarda, T., 2014. Precipitation variability over UAE and global SST teleconnections. J. Geophys. Res.: Atmos. 119.
Norouzi, H., Temimi, M., Prigent, C., Turk, J., Khanbilvardi, R., Tian, Y., Furuzawa, F., Masunaga, H., 2015. Assessment of the consistency among global microwave
land surface emissivity products. Atmos. Measur. Tech. 8, 1197.
Notarnicola, C., Caporaso, L., DI Giuseppe, F., Temimi, M., Ventura, B., Zebisch, M., 2012. Inferring soil moisture variability in the Mediterrean Sea area using infrared
and passive microwave observations. Can. J. Remote Sens. 38, 46–59.
Pan, M., LI, H., Wood, E., 2010. Assessing the skill of satellite-based precipitation estimates in hydrologic applications. Water Resour. Res. 46.
Peng, B., Shi, J., Ni-Meister, W., Zhao, T., JI, D., 2014. Evaluation of TRMM Multisatellite Precipitation Analysis (TMPA) products and their potential hydrological
application at an arid and semiarid basin in China. Selected Top. Appl. Earth Observ. Remote Sens. IEEE J. 7, 3915–3930.
Rabin, R.M., Temimi, M., Stepinski, J., Bothwell, P.D., 2014. Monitoring surface dryness using geostationary thermal observations. Remote Sens. Lett. 5, 10–18.
Romilly, T., Gebremichael, M., 2011. Evaluation of satellite rainfall estimates over Ethiopian river basins. Hydrol. Earth Syst. Sci. 15, 1505–1514.
Sawunyama, T., Hughes, D., 2008. Application of satellite-derived rainfall estimates to extend water resource simulation modelling in South Africa. Water SA 34, 1–9.
Schneider, U., Becker, A., Finger, P., Meyer-Christoffer, A., Rudolf, B., Ziese, M., 2011. GPCC Full Data Reanalysis Version 6.0 at 0.5: Monthly Land-surface

134
Y. Wehbe et al. Journal of Hydrology: Regional Studies 12 (2017) 122–135

Precipitation from Rain-gauges Built on GTS-based and Historic Data. http://dx.doi.org/10.5676/DWD_GPCC. FD_M_V6_050.
Schneider, U., Becker, A., Finger, P., Meyer-Christoffer, A., Ziese, M., Rudolf, B., 2014. GPCC's new land surface precipitation climatology based on quality-controlled
in situ data and its role in quantifying the global water cycle. Theor. Appl. Climatol. 115, 15–40.
Sorooshian, S., Hsu, K., Braithwaite, D., Ashouri, H., 2014. NOAA Climate Data Record (CDR) of Precipitation Estimation from Remotely Sensed Information Using
Artificial Neural Networks (PERSIANN-CDR), Version 1, Revision 1. Available at gis.ncdc.noaa.gov/geoportal/catalog/search/resource/details.page.
SU, F., Hong, Y., Lettenmaier, D.P., 2008. Evaluation of TRMM multisatellite precipitation analysis (TMPA) and its utility in hydrologic prediction in the La plata basin.
J. Hydrometeorol. 9, 622–640.
Swenson, S., 2012. GRACE Monthly Land Water Mass Grids NETCDF Release 5 0. Ver. 5.0. PO. DAAC, CA, USA.
Temimi, M., Lakhankar, T., Zhan, X., Cosh, M.H., Krakauer, N., Fares, A., Kelly, V., Khanbilvardi, R., Kumassi, L., 2014. Soil moisture retrieval using ground-based L-
band passive microwave observations in northeastern USA. Vadose Zone J. 13.
Vernimmen, R., Hooijer, A., Aldrian, E., Van Dijk, A., 2012. Evaluation and bias correction of satellite rainfall data for drought monitoring in Indonesia. Hydrol. Earth
Syst. Sci. 16, 133–146.
Willmott, C.J., Matsuura, K., Legates, D., 2001. Terrestrial air temperature and precipitation: monthly and annual time series (1950–1999). Center Clim. Res. 1.
Wulfmeyer, V., Branch, O., Warrach-Sagi, K., Bauer, H.-S., Schwitalla, T., Becker, K., 2014. The impact of plantations on weather and climate in coastal desert regions.
J. Appl. Meteorol. Climatol. 53, 1143–1169.
Xie, P., Yoo, S.-H., Joyce, R., Yarosh, Y., 2011. Bias-corrected CMORPH: a 13-year Analysis of High-resolution Global Precipitation. Availeble online: http://ftp.cpc.
ncep.noaa.gov/precip/CMORPH_V1.0/REF/EGU_11462 (accessed on 11 December 2014).
Yan, B., Weng, F., 2011. Effects of microwave desert surface emissivity on AMSU-A data assimilation. IEEE Trans. Geosci. Remote Sens. 49, 1263–1276.
Yang, Y., Luo, Y., 2014. Evaluating the performance of remote sensing precipitation products CMORPH, PERSIANN, and TMPA, in the arid region of northwest China.
Theor. Appl. Climatol. 118, 429–445.
Yin, Z.-Y., Zhang, X., Liu, X., Colella, M., Chen, X., 2008. An assessment of the biases of satellite rainfall estimates over the Tibetan Plateau and correction methods
based on topographic analysis. J. Hydrometeorol. 9, 301–326.
Zheng, W., Wei, H., Wang, Z., Zeng, X., Meng, J., Ek, M., Mitchell, K., Derber, J., 2012. Improvement of daytime land surface skin temperature over arid regions in the
NCEP GFS model and its impact on satellite data assimilation. J. Geophys. Res.: Atmos. 117.

135