rainfall analysis for the northern wadis of united arab ... · coastal zones, and drainage basins....

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Rainfall Analysis for the Northern Wadis of United ArabEmirates: A Case Study

Mohsen Sherif1; Salem Akram2; and Amapr Shetty3

Abstract: The rainfall distribution in space and time, its consistency, probability of occurrences, and drought conditions in three selectedwadis in the northern area of the United Arab Emirates �UAE� have been investigated. Like other arid and semi-arid regions, rainfall inthe selected wadis is limited, scattered, and random. The standard deviation and coefficient of asymmetry of the rainfall patterns arerelatively high. The three wadis have similar rainfall patterns based on the monthly, annual, and 24-h maximum rainfalls. Empiricalequations were developed to estimate the annual and 24-h maximum rainfalls. A probability analysis is carried out for the annual and 24-hmaximum rainfalls and probability graphs are developed with a confidence level of 95%. Intensity–duration–frequency curves aredeveloped for the three wadis. A conceptual model is used in drought characterization. The available rainfall data of the three wadis areanalyzed for drought intensity, duration, and frequency. Based on the drought characterization, the study region is classified as “arid.” Thepresent study constitutes the first comprehensive analysis for rainfall in UAE.

DOI: 10.1061/�ASCE�HE.1943-5584.0000015

CE Database subject headings: Rainfall; Droughts; United Arab Emirates; Case reports.

Introduction

The United Arab Emirates �UAE� lies in the southeastern part ofthe Arabian peninsula between latitudes 22° 40� and 26° 00�North and longitudes 51° 00� and 56°00� East. It is bounded fromthe north by the Arabian Gulf, on the east by the Sultanate ofOman and the Gulf of Oman, and on the south and the west by theKingdom of Saudi Arabia, Fig. 1. The total area of the UAE isabout 83,600 km2. Most of the land is desert and is characterizedby the predominance of Aeolian landform system. Its geomorpho-logic features include mountains, gravel plains, sand dunes,coastal zones, and drainage basins.

The land of UAE is divided into two distinct zones: the largerlow-lying zone and the mountains zone. The first covers over90% of the country’s area, extending from the northwest to theeastern part of the country where it is truncated by the mountainszone �Al Hammady 2003�. The low-lying zone ranges in altitudefrom sea level up to 300 m. Its major part is characterized by thepresence of sand dunes which rise gradually from the coastalplain reaching their highest elevation of 250 m above sea level�asl�. Along the coast of the Arabian Gulf, the low-lying land ispunctuated by ancient raised beaches and isolated hills which mayreach up to 40 m asl in some locations �Baghdady 1998�.

The UAE is known for its arid conditions. A long hot summer

1Professor of Water Resources, Civil and Environmental EngineeringDept., College of Engineering, UAE Univ., P.O. Box 17555 Al Ain, UAE.

2Director, Water and Dams Division, Ministry of Environment andWater, P.O. Box 1509, Dubai, UAE.

3Hydrologist, Water and Dams Division, Ministry of Environment andWater, P.O. Box 1509, Dubai, UAE.

Note. This manuscript was submitted on November 29, 2006; ap-proved on August 12, 2008; published online on February 18, 2009.Discussion period open until November 1, 2009; separate discussionsmust be submitted for individual papers. This paper is part of the Journalof Hydrologic Engineering, Vol. 14, No. 6, June 1, 2009. ©ASCE, ISSN
1084-0699/2009/6-535–544/$25.00.
JOU

J. Hydrol. Eng., 2009,

and short mild winter characterize its climate. The mean annualtemperatures are approximately uniform throughout the countrywith slight local variations. The principal rain in UAE falls be-tween November and March, with maximum intensity duringFebruary and March. Rainfall is very scarce, random, and infre-quent. The mean annual rainfall is about 110 mm with extremevariability in space and time.

Rainfall and runoff deficits have been observed in UAE andother neighboring Gulf Cooperation Council Countries, includingBahrain, Kuwait, Qatar, Saudi Arabia, and Sultanate of Oman �AlRashed and Sherif 2000�. The annual rainfall in the Arabian Gulfregion has dropped significantly since 1998. Based on the recordsof the last 8 years, the annual rainfall in UAE is in the order of40 mm only. Therefore, more reliance has been placed on ground-water resources that are mostly nonrenewable and on the desali-nation of seawater.

The recharge from the bed of ephemeral wadis and subsurfaceflow in valley bed in mountainous areas constitute an importantelement for the proper management of water resources in arid andsemi-arid regions �Khazei et al. 2003�. The successful implemen-tation of water resources management requires knowledge ofspace, time, and frequency distribution of rainfall and its charac-teristics. Apart from the quantum of rainfall, its distribution inspace and time plays a vital role in the planning and managementof water resources.

Risk analysis of rainfall events and design of water structuresare generally based on extreme precipitation events. Therefore,evaluation of rainfall extremes through intensity–duration–frequency curves and intensity–duration–area–frequency curveshas been considered by many researchers for several decades�Eaglson 1970; Chow et al. 1988; Willems 2000; Veneziano andFurcolo 2002; Veneziano et al. 2006, and others�.

This paper presents the first comprehensive study for thecharacterization and analysis of rainfall in the northern area ofUAE. To that end, three main catchments were selected, namely
Wadi Bih, Wadi Tawiyean, and Wadi Ham. Empirical relations
RNAL OF HYDROLOGIC ENGINEERING © ASCE / JUNE 2009 / 535

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

for the estimation of annual rainfall and maximum daily rainfallwith different return periods are developed. Intensity–durationcurves are presented and a drought analysis is conducted. Al-though rainfall and occasional water floods may not be of signifi-cant contribution to the water budget in UAE, their assessmentand analysis are vital for design of dams and mitigation of pos-sible flood hazardous.

Physical Settings of the Selected Catchments

The three investigated catchment areas of Wadi Bih, WadiTawiyean, and Wadi Ham are located in the northern part of theUAE, Fig. 2. These areas were selected based on the availabilityof relatively long and accurate rainfall records. However, itshould be noted that the annual rainfall in the northern areas isrelatively higher than in the middle and southern regions of thecountry. Many detention and retention dams have been con-structed across the main wadis within the selected areas for sur-face water harvesting and groundwater recharge.

Fig. 1. Location map

Wadi Bih, in Ras Al Khaimah Emirtae, is a large southwesterly

536 / JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / JUNE 2009


flowing wadi complex comprising parts of the Jibal Ruusmountains of Oman in its eastern section and alluvial plains to thewest and southwest around Ras Al Khaimah Emirate. The waditopography consists of three major land forms namely, mountains,alluvial terrace, and low land. The maximum elevation of thecatchment is 2,087 m above mean sea level �amsl�. The catch-ment area of Wadi Bih is about 483 km2. Wadi Tawiyean, also inRas Al Khaimah Emirate, is arising in the Jibal Ruus mountainsand flowing west to the coastal plain. The boundary of the water-shed is in common with southern boundary of Wadi Naqab. Thewadi topography consists of mountains, alluvial terrace, and lowland of three distinct land forms. The maximum elevation of thecatchment is 1,527 m above mean sea level. The catchment areaof Wadi Tawiyean dam is about 198 km2. Wadi Ham, in FujairahEmirates, rises in the mountains at an elevation 1,109 m amslimmediately south and south east of the Musafi, draining south-eastward into the Gulf of Oman between Fujairah and Kalbha.The wadi complex comprises parts of Musafi Mountain in itsnorthwestern portion and alluvial plains to the east around Fu-jairah. The catchment area of this wadi is approximately 195 km2

United Arab Emirates
of the
�Sherif et al. 2005; Halcrow 1984�.

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Rainfall Pattern and Characterization

Rainfall forms the input of all hydrological studies. Apart fromthe quantum of rainfall, its time distribution plays a critical role inthe planning and management of water resources. Understandingrainfall pattern of a given catchment is fundamental for anyproper runoff–flood investigations. The analysis of daily, monthly,annual rainfall, and 1-day maximum annual rainfall is useful topredict rainfall amount in a season for desired recurrence intervalwith a certain level of confidence.

Methods of calculating average areal rainfall include, but arenot limited to, the arithmetic average of gauged quantiles, theisohyetal map method, and the Thiessen method �Singh 1992�.The number of rain gauges available in the selected wadis is notsufficient to develop isohyetal maps. The Thiessan polygonmethod was used in this investigation to calculate the averageareal precipitation over the selected catchements.

Observations of rainfall within and around the selected basinswere used. The Ministry of Environment and Water, MEW �for-merly, Ministry of Agriculture and Fisheries� has been responsiblefor the hydrmeteorological network in the United Arab Emirates�UAE� since 1971. Three stations of daily data for Wadi Ham,three stations for Wadi Tawiyean, and three stations for the WadiBih were used. Fig. 2 presents the locations of these rain gauges.The length of data records, Thiessen weightage, and rain gaugesconsidered for the three wadis are given in Table 1. No record ofprecipitation is available for the Oman region in the case of WadiBih. One should also recognize the difference between the meanannual rainfall in the three wadis based on the records of22–25 years and the mean annual rainfall in UAE based on therecords of the last 8–10 years. The average rainfall has decreasedsignificantly over the last 10 years. In addition, the average an-nual rainfall in the three wadis is relatively high as compared to

Fig. 2. Location map of the three selected wadis and available raingauges

other areas in the country �MAF 2001�.

JOU


Statistical Analysis

The random variability of hydrologic variables such as precipita-tion and streamflow has been recognized for centuries. The use ofstatistics in hydrology provides information about various param-eters and distribution of random variables of rainfall analysis todesign and operate water systems. Hydrologic data are mostlyavailable as samples of limited sizes. Statistics enables the extrac-tion of needed information from the available data and character-ization of hydrologic random variables needed for rainfall andrunoff analyses.

Because of the high randomness and inconsistency in rainfallpattern and intensity, it might not be proper to develop waterutilization plans and policies based on statistical analysis of shortand medium rainfall records. Nevertheless, such analysis can beused in the design of storage dams and mitigation of flash floods.

In data analysis, certain calculations are usually made to de-termine some of the basic inherent properties of the data. Theinferences include information about the central tendency, range,distribution within the range, variability around the central ten-dency, degree of uncertainty, and frequency of occurrence of val-ues �Viessman and Lewis 2003�. The standard deviation measuresthe dispersion of sample values around the mean. The skewnesscoefficient measures the asymmetry of the frequency distributionof the data. It has an important meaning as it gives an indicationof the symmetry of the distribution of the data. Symmetricalfrequency distributions have very small or negligible sampleskewness coefficient �Cs�, whereas asymmetrical frequency dis-

Table 1. Rain Gauge and Thiessen Weights

WadiArea�km2�

Raingauges

Area�km2�

Tieissenweights

Period�year�

Ham 195 Masafi 64.35 0.33 25

Bithna 113.10 0.58

Farah 17.55 0.09

Tawiyean 198 Tawiyean 138.60 0.70 23

Sinah 15.84 0.08

Khatt 9.90 0.05

Dibba 33.66 0.17

Bih 464 Bih 394.40 0.85 22

Sham 37.12 0.08

Burayarat 32.48 0.07

Table 2. Statistical Value of Monthly Rainfall Distribution in Wadi Ham

Month

Rainfallrange�mm�

Mean�mm�

% ofnormalR–fall

S.D.�mm�

October 0–74 6 4 15

November 0–33 6 4 10

December 0–178 16 11 37

January 0–111 19 13 26

February 0–198 41 27 61

March 0–129 35 24 47

April 0–54 11 7 16

May 0–28 3 2 8

June 0–8 1 1 2

July 0–72 5 3 15

August 0–23 2 2 5

September 0–20 2 2 5


14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

tributions have either positive or negative coefficients. A smallvalue of Cs often indicates that frequency distribution of thesample may be approximated by the normal distribution functionas Cs=0 for this function. The kurtosis coefficient measures thepeakedness or the flatness of the frequency distribution near itscenter. The positive value of an excess coefficient indicates that afrequency distribution is more peaked around its center than thenormal distribution. The negative value of an excess coefficientindicates that a given distribution is more flat around its centerthan the normal.

Statistical analyses of monthly and annual mean rainfall forthe three selected wadis were carried out. The statistical results ofthe rainfall in the three wadis are presented in Figs. 3–5 andTables 2–6. Annual rainfall in Wadi Ham ranges from7.6 to 505.8 mm indicating a wide variation from one year toanother. Average rainfall estimated based on the records of thelast 25 years is 151 mm with a standard deviation of 126.8 mm, akurtosis coefficient of 1.36, and a coefficient of asymmetry of

Fig. 3. Distribution of m

Fig. 4. Distribution of mea



1.18, Table 5. The probabilities of occurrence of 75 and 50% ofthe average rainfall are estimated as 51 and 64%, respectively. Itimplies that the distribution of rainfall in Wadi Ham is highlyscattered.

Rainfall events in Wadi Ham are observed between Januaryand December but are mostly encountered during the months ofFebruary and March. More than 50% of annual rainfall normallyoccurs during these two months. Monthly rainfall values rangefrom 0 to 198 mm, the mean monthly varies from 1 to 41 mmwith a standard deviation varying between 2 and 61 mm, Table 2.The monthly standard deviation exceeds the monthly average pre-cipitation revealing that the year-to-year monthly variation in pre-cipitation is quite extreme in the area. The 24-h maximum rainfallvalues range from 2 to 137 mm and the mean is 50 mm with astandard deviation 34 mm. The kurtosis coefficient is 0.69 andcoefficient of asymmetry is 1.06, Table 6. The probabilities ofoccurrence of 75 and 50% of mean 1-day maximum rainfall are57 and 79%, respectively.

nual rainfall, Wadi Ham

al rainfall, Wadi Tawiyean

ean an

n annu

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Annual rainfall in Wadi Tawiyean catchment area ranges from4 to 430 mm. Based on the rainfall records of the last 23 years,the average annual rainfall is estimated as 129 mm with a stan-dard deviation of 115 mm, a kurtosis coefficient of 0.46, and acoefficient of asymmetry of 0.95, Table 5. The probabilities ofoccurrence of 75 and 50% of the average rainfall are estimated as57 and 72%, respectively. Again the rainfall is distributed be-tween January and December but mostly encountered during thetwo months of February and March. Monthly rainfall valuesrange from 0 to 156 mm and mean monthly varies from 0 to36 mm with variation of the standard deviation from 0 mm,where mean monthly rainfall is zero, to 45 mm, Table 3. Themonthly standard deviation exceeds the monthly average precipi-tation indicating that the year-to-year monthly variation in pre-cipitation is quite high in the area. The 24-h maximum rainfallvalues range between 2 and 86 mm with a mean value of 36 mmand a standard deviation of 23 mm. The kurtosis coefficient is−0.69 and the coefficient of asymmetry is 0.46, Table 6. Theprobabilities of occurrence of 75 and 50% of mean 1-day maxi-mum rainfall are 55 and 77%, respectively.

Table 3. Statistical Value of Monthly Rainfall Distribution in WadiTawiyean

Month


Mean�mm�

% ofnormalR-fall

S.D.�mm�

October 0–41 4 3 10

November 0–97 11 8 23

December 0–156 20 15 34

January 0–96 22 16 27

February 0–130 29 22 40

March 0–150 36 25 45

April 0–37 7 5 11

May 0–3 0 0 1

June 0–5 0 0 1

July 0–79 4 3 17

August 0–20 1 1 4


Fig. 5. Distribution of m

JOU


The annual rainfall in Wadi Bih ranges from 6 to 414 mm.Based on the records of the last 22 years, the average rainfall isestimated as 119 mm with a standard deviation of 109 mm, akurtosis coefficient of 0.53, and a coefficient of asymmetry of0.89, Table 5. The probabilities of occurrence of 75 and 50% ofthe average rainfall are estimated as 51 and 67%, respectively.Like the other two wadis, this implies that the distribution ofrainfall in Wadi Bih is scattered and random. Rainfall events arealso observed between January and December with about 50% ofthe annual rainfall occurring in February and March. Monthlyrainfall values range between 0 and 176 mm and the meanmonthly varies from 0 to 34 mm with a variation of the standarddeviation from 0, where the mean rainfall is zero, to 50 mm,Table 4. The monthly standard deviation exceeds the monthlyaverage precipitation. The 24-h maximum rainfall values rangebetween 3 and 105 mm with a mean of 33 mm and a standarddeviation of 24 mm. The kurtosis coefficient is 3.26 and the co-efficient of asymmetry is 1.50, Table 6. The probabilities of oc-currence of 75 and 50% of mean rainfall are 51 and 68%,respectively.

Probability is a constant characterizing a given set of objects

Table 4. Statistical Value of Monthly Rainfall Distribution in Wadi Bih

Month


Mean�mm�

% ofnormalR–fall

S.D.�mm�

October 0–27 1 1 6

November 0–38 3 3 9

December 0–176 24 20 42

January 0–90 23 19 26

February 0–160 30 24 41

March 0–157 34 28 50

April 0–23 4 4 7

May 0–1 0 0 2

June 0 0 0 0

July 0–22 2 2 7

August 0 0 0 0


nnual rainfall, Wadi Bih
ean a

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

or incidents in a particular period. The probability can be conve-niently estimated if the length of available data is sufficient. Theavailable rainfall data in the three wadis are used to fit probabilitydistribution which is then used to extrapolate design events fromthe recorded events either graphically or by estimating the param-eters of frequency distribution.

The 24-h maximum and the mean annual rainfall data for thethree selected wadis were analyzed for estimating the expectedrainfall at different return periods using Gauss, Lognormal,Gumble, Pearson Type-III, and Weibull distribution formula�Singh 1992�. The formula proposed by Weibull was found to betheoretically suitable for plotting the annual maximum series. Itprovides estimates that are consistent with experience �Hann andShapiro 1967�. Weibull formula can be mathematically expressedas

Fa = �m/N + 1�100 �1�

where N=total number of data items; m=number of items ar-ranged in descending order of magnitude; and Fa=plotting posi-tion.

In this study, Weibull formula was used to estimate the returnperiods and the 95% confidence interval was also drawn to indi-cate the likely range of the true value of the quantile. Usually, anerror of 5% is considered acceptable �Singh 1992; Helsel andHirsch 2000�.

Rainfall for different probabilities of 1-day annual maximumand annual mean rainfall were determined from probabilitygraphs and corresponding values were deduced. The estimatedannual rainfall and 24-h maximum rainfall for the correspondingreturn periods are provided in Tables 7 and 8.

In Wadi Ham, the estimated annual rainfall for a 2-year returnperiod is 112 mm and for 100-year return period is 691 mm�Table 7�. The model for the estimation of annual rainfall for thecorresponding return period is

RaH = 147.20 ln�TH� + 16.402 �2�

where RaH=annual rainfall in Wadi Ham and TH is the returnperiod. This equation can be expressed in a standard form as

TH = e��P−16.402�/147.20� �3�

Table 5. Statistical Value of Annual Rainfall Distribution

Wadi


Mean�mm�

S.D.�mm� K

Ham 7.6–505.8 151 126.8

Tawiyean 4–430 129 115

Bih 6–414 119 109

Table 6. Statistical Value of 1-Day Annual Maximum Rainfall Distribut

Wadi


Mean�mm�

S.D.�mm� Ku

Ham 2–137 50 34

Tawiyean 2–86 36 23 −

Bih 3–105 33 24



where P=annual precipitation in millimeters. The observed val-ues are within the 95% confidence interval of the fitted valueswith a coefficient of determination of 0.98. In other words, 98%of the recorded annual rainfall is included in the estimated rain-fall. Also, the estimated 24-h maximum rainfall for a 2-year re-turn period is 44 mm and for 100-year return period is 202 mm�Table 8�. The model for the estimation of the 24 h maximumrainfall for the corresponding return period in Wadi Ham is givenas

R24H = 40.472 ln�TH� + 15.562 �4�

where R24H=24-h maximum rainfall in Wadi Ham.The estimated annual rainfall of Wadi Tawiyean for the 2-year

return period is 116 mm and for 100-year return period is670 mm. The model for the estimation of annual rainfall in WadiTawiyean for the corresponding return period is

RaT = 139.52 ln�TT� + 20.301 �5�

where RaT=24-h maximum rainfall and TT=return period �inyears� for Wadi Tawiyean. Eq. �5� can also be expressed in astandard form as

TT = e��P−20.301�/139.55� �6�

where P=annual precipitation in millimeters. The observed val-ues are within 95% confidence interval of the fitted values with acoefficient of determination of 0.95. On the other hand, the esti-mated 24-h maximum rainfall for the 2-year return period is30 mm and for the 100-year return period is 135 mm. The modelfor the estimation of the 24-h maximum rainfall for the corre-sponding return period is

R24T = 27.044 ln�TT� + 10.91 �7�

where R24T=24-h maximum rainfall in Wadi Tawiyean. The ob-served values are within the 95% confidence interval of the fittedvalues with a coefficient of determination of about 0.91, i.e., 91%of the observed 24-h maximum rainfall is included in the esti-mated range of rainfall.

Coefficient ofasymmetry

Probability ofoccurrence

of 75%average


of 50%average

1.18 51 64

0.95 57 72

0.89 51 67

Coefficient ofasymmetry


of 75%average


of 50%average

1.06 57 79

0.46 55 77

1.50 51 68

urtosis

1.36

0.46

0.53

ion

rtosis

0.69

0.69

3.26

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

In Wadi Bih, the estimated annual rainfall for the 2-year returnperiod is 97 mm and for the 100-year return period is 614 mm.The model for the estimation of annual rainfall in Wadi Bih forthe corresponding return period is given as

RaB = 132.37 ln�TB� + 4.8307 �8�

where RaB=24-h maximum rainfall and TB=return period �inyears� in Wadi Bih. This equation can be expressed in a standardform as

TB = e��P−4.8307�/132.37� �9�

The observed values are within the 95% confidence interval ofthe fitted values with a coefficient of determination of about 0.95.The estimated 24-h maximum rainfall for the 2-year return period

Table 7. Estimated Return Period of Annual Rainfall

Returnperiod

Estimated rainfall �mm�

Ham Tawiyean Bih

2 112 116 97

5 247 246 218

10 350 344 310

25 486 474 431

50 588 572 523

100 691 670 614

Table 8. Estimated Return Period of 24-h Maximum Rainfall

Returnperiod

Estimated rainfall �mm�

Ham Tawiyean Bih

2 44 30 26

5 81 54 53

10 109 73 73

25 146 98 100

50 174 117 120

100 202 135 140

Fig. 6. Smoothened inten

JOU


is 26 mm and for 100-year return period is 140 mm. The modelfor the estimation of 24-h maximum rainfall for the correspond-ing return period is

R24B = 29.294 ln�TB� + 55.5553 �10�

where R24B=24-h maximum rainfall in Wadi Bih. The observedvalues are within the 95% confidence interval of the fitted valueswith a coefficient of determination of about 0.96.

As the standard deviations are invariably more than the meanmonthly rainfall, the occurrence of rainfall is highly nonuniform,empirical equations and probability estimators may not satisfac-torily forecast future rainfall occurrence. Nevertheless, empiricalequations that are based on statistical analysis of rainfall in aridregions can be used for long-term predictions, e.g., maximumrainfall in 100 years, which are needed for design of storage damsand mitigation of floods.

Intensity–Duration–Frequency Analysis

Rainfall intensity is one of the most important factors affectingthe generation of surface water runoff especially from mountain-ous watersheds. The average intensity, peak intensity, and dura-tion are mainly dependent on the interevent time consideredduring the isolation of storm events.

The rainfall events are isolated with minimum interevent timeof 1 h. The durations of rainfall events and rainfall intensitieshave been calculated. Ranges of storm durations and intensities inthe three selected wadis are presented in Table 9. It is evidentfrom the available data that shorter duration rainfalls have higherintensity. Intensity duration curves have been developed as pre-

Table 9. Range of Rainfall Intensity and Duration

Wadi Data periodRange of duration

�h�Range of intensity

�mm/h�

Ham 1980–2005 0.25–26.0 0.6–58

Tawiyean 1982–2005 0.25–39.0 0.3–46

Bih 1983–2005 0.25–45 0.5–34

uration curve, Wadi Ham
sity d

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

sented in Figs. 6–8. However, it should be noted that the charac-teristic of intensity–duration curve is averaged over the specifiedduration. The curve represents the smoothed results of differentstorms.

The range of event duration varies from 0.25 to 26 h in WadiHam, from 0.25 to 39 h in Wadi Tawiyean, and from 0.25 to 45 hin Wadi Bih. The intensity of rainfall varies from 0.6 to 58 mm /hin Wadi Ham. In Wadi Tawiyean it varies from 0.3 to 48 mm /hand in the case of Wadi Bih it varies between 0.5 and 34 mm /h.Wadi Ham has lower durations of storm events and higher rainfallintensities as compared to Wadi Tawiyeana and Wadi Bih. To thecontrary, Wadi Bih is distinguished by its longer durations andsmaller intensities of rainfall events. In Wadi Ham, 96% of rain-fall events during the last 25 years had a duration of 6 h or less,Fig. 6. In Wadi Tawiyean and Wadi Bih rainfall events are gen-erally of longer durations, Figs. 7 and 8. Typical storm durations

Fig. 7. Smoothened intensi

Fig. 8. Smoothened inte



and the corresponding intensity ranges for the three selectedwadis are presented in Table 10.

The intensity–duration–frequency curves were developed andare presented in Figs. 9–11 for a frequency ranging from2 to 250 years. In Wadi Ham, the maximum expected intensityfor a 2-year return period is about 17 mm /h, whereas for the250-year return period the maximum expected intensity is about100 mm /h. In Wadi Tawiyean, the predicted maximum intensityis about 10 mm /h for the 2-year return period and 85 mm /h forthe 250-year return period. In the case of Wadi Bih, the maximumintensity for a 2-year return period is around 10 and 77 mm /h forthe 250-year return period. The variation in the storm intensity fora duration between 0.25 and 0.50 h is relatively small for thecases of Wadi Tawiyean and Wadi Bih as compared to Wadi Ham,Table 10. When the duration of the storm is 2.5 h or more therainfall intensity for all return periods in the three wadis fall

tion curve, Wadi Tawiyean

uration curve, Wadi Bih

ty dura

nsity d

14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

within 20 mm /h. Long rainfall storms are of small intensities inthe three wadis.

Drought Analysis

Drought is a normal and recurrent feature of the climate. It occursin virtually all climatic zones and its characteristics vary signifi-cantly among regions. Drought differs from aridity in that droughtis temporary; aridity is a permanent characteristic of regions withlow rainfall. Drought produces a complex web of impacts whichspans many sectors. The intensity, duration, and frequency ofdroughts have a major influence on societies.

The relations among drought intensity, duration, and frequencycan be analyzed using the conceptual model developed by Ponceet al. �2000�. The conceptual approach is applicable to subtropicaland midlatitudinal regions and is limited to meteorologicaldroughts lasting at least 1 year.

For any year for which P is the annual precipitation, droughtintensity is defined as the ratio of the deficit �Pma− P� to the mean�Pma�. For any particular year, an intensity of ��Pma− P� / Pma�=0.25 is classified as moderate, 0.5 is classified as severe, and0.75 is classified as extreme. For drought events lasting more than1 year, intensity is the summation of the individual annual inten-sities. Therefore, extreme drought intensities are generally asso-ciated with droughts of long duration.

Drought characterizations in Wadi Ham, Wadi Tawyiyean,and Wadi Bih have been performed using the conceptual modelof Ponce et al. �2000�. Drought intensity, duration, and frequen-

Table 10. Typical Storm Duration and its Intensity Range

Wadi

Intensity range �mm/h�

0.25 h 0.50 h

Ham 6–58 2–30

Tawiyean 0.5–48 0.5–38

Bih 2–34 2–30

Table 11. Drought Analysis of the Three Selected Wadis

WadiNumber of

yearsMean annualrainfall �mm�

Numberevents

Ham 25 151 6

Tawiyean 23 129 5

Bih 22 119 5

Fig. 9. Intensity–duration–frequency curve, Wadi Ham

JOU


cy for these three wadis were estimated and are presented inTable 11.

Drought intensity �per event� in Wadi Ham varies from 0.30to 4.52, with an average value of 1.41. Drought duration rangesfrom 1 to 7 years, with an average value of 2.35 years. The aver-age frequency of the drought is 4.17 years, Table 11. In WadiTawiyean, the drought intensity �per event� varies from 0.07 to5.03, with an average value of 1.67. Drought duration rangesfrom 1 to 7 years with an average value of 2.8 years. The averagefrequency of the drought in Wadi Tawiyean is 4.6 years. Onthe other hand, the drought intensity �per event� in Wadi Bihvaries from 0.30 to 5.28 with an average value of 1.56. Droughtduration ranges from 1 to 7 years, with an average value of2.6 years. The average frequency of the drought in Wadi Bih is4.4 years. The longest mean drought duration �2.8 years� oc-curred at Wadi Tawiyean and the shortest �2.35 years� occurred atWadi Bih. The average drought duration in the three wadis isabout 2.6 years.

Conclusions

The annual mean rainfall of the three selected wadis in the north-ern part of UAE ranges between 119 and 151 mm with standarddeviations varying between 109 and 126 mm. However, recentrainfall records indicate a remarkable decline in rainfall fre-quency, intensity, and duration. The probability of occurrence of75 and 50% of the average annual rainfall varies between 51–57% and 64–72%, respectively. More than 50% of the annual

h 2.00 h 3.00 h 5.00 h

1–9 0.5–12 1–6

0 1–12 0.5–4 0.5–6

1–6 0.5–10 1–3

Averageintensity

Duration�year�

Averageduration�year�

Averagefrequency

�year�

1.41 1–7 2.35 4.17

1.67 1–7 2.8 4.6

1.56 1–7 2.6 4.4

Fig. 10. Intensity–duration–frequency curve, Wadi Tawiyean

1.00

1–22

0.5–2

1–12

of


14(6): 535-544

Dow

nloa

ded

from

asc

elib

rary

.org

by

Uni

ted

Ara

b E

mir

ates

Uni

vers

ity o

n 10

/22/

17. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

rainfall occurs in February and March. The values of mean 1-dayannual maximum rainfall vary between 33 and 50 mm with stan-dard deviations in the range of 24–34 mm. The probability ofoccurrence of 50% of mean 1-day annual maximum rainfall in thethree wadis varies between 68 and 79%.

Rainfall events with durations of 2.5 h or more have intensitiesof 20 mm /h or less for all return periods. The developed intensitypatterns of the three wadis can be used in the assessment of sur-face water runoff from various rainfall events of different fre-quencies. This will be addressed in the writers’ next paper on theassessment and perdition of surface water runoff in the threewadis.

Weibull distribution provided good agreement with the avail-able rainfall records with a 95% confidence level of the estimatedquantiles in three wadis. The high values of standard deviationsand coefficients of asymmetry are attributed to scattereness andrandomness of the rainfall events in the area. The rainfall charac-teristics in Wadi Tawiyean and Wadi Bih are relatively similar dueto the proximity of the geographical locations of the two wadis.

Drought characterization has been performed for the threewadis using the conceptual model of Ponce et al. �2000�. Thethree wadis are classified as “arid” with average drought durationsof 2.35, 2.8, and 2.6 years in Wadi Ham, Wadi Tawiyean, andWadi Bih, respectively. The average drought duration in the threewadis is 2.6 years.

Acknowledgments

The research and results presented in this paper have been com-pleted within the activities of a project entitled “Assessment ofthe Effectiveness of Al Bih, Al Tawyean and Ham Dams inGroundwater Recharge using Numerical Models.” The projectwas support by the Ministry of Environment and Water �MEW�,

Fig. 11. Intensity–duration–frequency curve, Wadi Bih



Dubai, UAE. Several individuals from the MEW and the UAEUniversity have contributed in the field activities and data collec-tions related to this paper.

References

Al Hammadi, M. K. �2003�. “Assessment of groundwater resources usingremote sensing and GIS.” MSc thesis, Water Resources Master Pro-gram, UAE Univ., United Arab Emirates.

Al-Rashed, M. F., and Sherif, M. M. �2000�. “Water resources in the GCCcountries: An overview.” Water Resour. Manage., 14�1�, 59–75.

Baghdady, A. R. �1998�. “Petrography, mineralogy and environmentalimpacts of the sand dune fields of the greater Al Ain area, United ArabEmirates.” Ph.D. thesis, Faculty of Science, Ain Shams Univ., AinShams, Egypt.

Chow, V. T., Maidment, D. R., and Mays, L. W. �1988�. Applied hydrol-ogy, McGraw-Hill, New York.

Eaglson, P. S. �1970�. Dynamic hydrology, McGraw-Hill, New York.Halcrow International Partnership. �1984�. “Dams and recharge facilities

site investigation Wadi Tawiyean.” Geophysical survey, Vol. 3, Pt. A,Ministry of Agriculture and Fisheries �currently Ministry of Environ-ment and Water�, Dubai, United Arab Emirates.

Hann, G. J., and Shapiro, S. �1967�. Statistical models in engineering,Wiley, New York.

Helsel, D. R., and Hirsch, R. M. �2000�. Statistical methods in waterresources, 4th Ed., Elsevier Science, Amsterdam, The Netherlands.

Khazaei, E., Spink, A. E. F., and Warner, J. W. �2003�. “A catchmentwater balance model for estimating groundwater recharge in arid andsemiarid regions of southeast Iran.” Hydrogeol. J., 11�3�, 333–342.

Ministry of Environment and Water �MAF�. �2001�. Climatological data,Vol. 4, 1992–93 to 1999–2000, United Arab Emirates.

Ponce, V. M., Pandey, R. P., and Ercan, S. �2000�. “Characterization ofdrought across the climatic spectrum.” J. Hydrol. Eng., 5�2�, 222–224.

Sherif, M. M., et al. �2005�. “Assessment of the effectiveness of Al Bih,Al Tawiyean and Ham Dams in groundwater recharge using numeri-cal models.” Final Rep., Vol. 1: Main report, Ministry of Environmentand Water, Dubai, United Arab Emirates.

Singh, V. P. �1992�. Elementary hydrology, Prentice-Hall, EnglewoodCliffs, N.J.

Veneziano, D., and Furcolo, P. �2002�. “Multifractality of rainfall andintensity-duration-frequency curves.” Water Resour. Res., 38�12�,1306.

Veneziano, D., Langousis, A., and Furcolo, P. �2006�. “Multifractality andrainfall extremes: A review.” Water Resour. Res., 42, W06D15.

Viessman, W., and Lewis, G. L. �2003�. Introduction to hydrology, 5thEd., Prentice-Hall, Pearson Education, Inc., Englewood Cliffs, N.J.

Willems, P. �2000�. “Compound intensity/duration/frequency relation-ships of extreme precipitation for two seasons and two storm types.”J. Hydrol., 233�1–4�, 189–205.

14(6): 535-544

rainfall analysis for the northern wadis of united arab ... · coastal zones, and drainage basins....

Documents