CC BY
39
ГИДРАВЛИКА.ГЕОТЕХНИКА. ГИДРОТЕХНИЧЕСКОЕ СТРОИТЕЛЬСТВО
УДК 551.581 DOI: 10.22227/1997-0935.2020.1.85-99
Statistical and spatial variability of climate data in the Mareb-Gash river basin in Eritrea
Anghesom A. Ghebrehiwot1, Dmitry V. Kozlov2
1 Hamelmalo Agricultural College, National Higher Education and Research Institute; Keren, Eritrea; 2 Moscow State University of Civil Engineering (National Research University) (MGSU);
Moscow, Russian Federation
ABSTRACT
Introduction. Global reanalysis products are extensively used for hydrologic applications in sparse data regions. The establishment of inputs for hydrologic modelling from such global reanalysis requires prior checking and analyses. Materials and methods. The present study attempts to utilize Climate Forecast System Reanalysis (CFSR) datasets for the Mereb-Gash river basin in Eritrea, to prepare the input data for forthcoming hydrological modelling studies. The activities include statistical analyses, computation of PET, and drought indices using different methods so as to understand basin characteristics through the use of geospatial and geostatistical tools.
Results. The results of statistical analyses indicated that there was predominantly a significant monotonic trend in the majority of the data. Precipitation (P) and relative humidity tend to decrease, whereas temperature (T) and potential evapotranspiration (PET) tend to increase. Among the PET estimation methods, the Thornthwaite method gave inconsistent results as compared to Hargreaves and Penman-Monteith methods, the former being highly dependent on the elevation of the station. In most cases, it was found that Penman-Monteith produced the highest PET values.
Conclusions. Besides, Standardized Precipitation and Evapotranspiration Index (SPEI) analyses in the basin indicate persistent dry conditions over the period 2000-2013 and predominantly humid conditions over the period 1979-2000. The study concluded that the presence of a significant trend in most of the climatic variables and persistent drought conditions in recent ^ ^ years were found to be congruent with global and regional climatic studies that are highly likely linked to human and climate & T influence on the environment. 3. 5
V'
KEYWORDS: global reanalysis, potential evapotranspiration, standardized precipitation and evapotranspiration index, д Mereb-Gash river basin
АННОТАЦИЯ
Введение. Продукты глобального реанализа имеют широкое применение в гидрологии при недостаточности исход-
С
О
FOR CITATION: Ghebrehiwot A.A., Kozlov D.V. Statistical and spatial variability of climate data in the Mareb-Gash river 2 —
o со
basin in Eritrea. Vestnik MGSU [Monthly Journal on Construction and Architecture]. 2020; 15(1):85-99. DOI: 10.22227/1997- t w 0935.2020.1.85-99 (rus.). У 1
uI
^ I
a 9
Статистическая и пространственная изменчивость климатических О 8
r
О 5'
данных в бассейне реки Мареб-Гаш в Эритрее o
АА. Гебрехивот1, Д.В. Козлов2 Ц м
1 Аграрный колледж Хамелмало, Национальный институт высшего образования и исследований; ° м
Кэрэн, Государство Эритрея; п о
2 Национальный исследовательский Московский государственный строительный университет ^ 6
(НИУ МГСУ); г. Москва, Россия
>KJJ t (
CD CD
ных данных. Формирование исходной информации для гидрологического моделирования на основе такого глобаль- < ^
ного реанализа требует предварительной проверки и анализа. -и О
Материалы и методы. Предпринята попытка использовать массивы данных реанализа системы климатических ^ 1
прогнозов (CFSR) для условий бассейна реки Мереб-Гаш в Эритрее с целью подготовки исходных данных для пред- ® <л
стоящих исследований в области гидрологического моделирования. Это исследование включило в себя статистиче- 5 п
ский анализ, расчет индексов потенциальной эвапотранспирации (ПЭТ) и засухи на основе применения различных 1 £
методов с целью изучения характеристик речного бассейна с использованием геопространственных и геостатисти- $ у
ческих инструментов. ф к
Результаты. Результаты статистического анализа показали, что в большинстве массивов данных в значительной сте- , ,
пени преобладают монотонные тенденции: осадки (Р) и относительная влажность имеют тенденцию к уменьшению, 2 2
тогда как температура воздуха (Т) и ПЭТ — тенденцию к увеличению. Среди методов оценки ПЭТ метод Торнтуэйта 2 2
о о
© Anghesom A. Ghebrehiwot, Dmitry V. Kozlov, 2020
Распространяется на основании Creative Commons Attribution Non-Commercial (CC BY-NC)
дал неоднозначные результаты по сравнению с методами Харгривза и Пенмана-Монтейта, причем использование первого метода сильно зависело от высотного положения метеостанции. В большинстве случаев было обнаружено, что по методу Пенмана-Монтейта получались самые большие значения ПЭТ.
Выводы. Анализ стандартизированного индекса осадков и эвапотранспирации (SPEI) по методу Пенмана-Мон-тейта в исследуемом речном бассейне указал на устойчивые засушливые условия в период с 2000 по 2013 г. и преимущественно влажные условия в период с 1979 по 2000 г. Сделан вывод о том, что имеющиеся тенденции для большинства климатических характеристик и устойчивого развития условий засухи в последние годы хорошо согласуются с результатами глобальных и региональных климатических исследований, и, по всей вероятности, связаны с влиянием на окружающую среду воздействия человека и изменения климата.
КЛЮЧЕВЫЕ СЛОВА: глобальный реанализ, потенциальная эвапотранспирация, стандартизированный индекс осадков и эвапотранспирации, бассейн реки Мереб-Гаш
ДЛЯ ЦИТИРОВАНИЯ: Гебрехивот A.A., Козлов Д.В. Статистическая и пространственная изменчивость климатических данных в бассейне реки Мареб-Гаш в Эритрее // Вестник МГСУ. 2020. Т. 15. Вып. 1. С. 85-99. DOI: 10.22227/1997-0935.2020.1.85-99
О о
N N О О N N
¡É ф U 3
> (Л
с и to in
И
ф <и
о %
---' "t^
о
о и со <т
8 «
<л ю
о О
ю со о о
I
О) О)
(Л
ю
Г ES
О (0 ф ф
ta >
INTRODUCTION
In any hydrological modelling process that intends to make predictions from ungauged catchments, three types of data are needed: runoff data, climatic data, and basin characteristics. The most widely used models that make use of such data include statistical, process-based or a combination, of both [1]. Hence, runoff Predictions for Ungauged Basins (PUB) (e.g., [2, 3]) recommended the assessment of available data and the information that can be obtained from such data to be the starting point in any hydrological studies. Different data sources for runoff PUB are available: global, local, and regional data sources followed by field observations [4]. If the resources are available, the most accurate runoff predictions can be obtained by utilising very local data describing the specific characteristics and behaviour of the system. While global or regional data can be useful for annual runoff prediction, more intensive local data can be needed for hydrograph prediction [1].
So far, limited hydrological studies have been carried out in the Mereb-Gash river basin in Eritrea. These limited studies (e.g.1, [5-8]) have sought to understand the hydrology of the area. Others (e.g., [9, 10]) have assessed the environmental repercussions due to flooding and possible mitigation mechanisms at the outlet of the basin. These modelling efforts have ranged from simple conceptual models to complex, physically-based distributed hydrological models. One of the main challenges highlighted in all these studies is the acute shortage of historical hydrometeorological data, suggestions for improved data collection and management system to increase the reliability of hydrological modelling efforts. Global reanalysis products are extensively used
1 Euroconsult. Sector study on national water resources and irrigation potential: component 1 report surface water resources. Arnhem, Netherlands, 1998.
for hydrologic applications in sparse data regions (e.g., [11-15]). For example, [14] evaluated ERA-5, CFSR, ERA-Interim, Modern-Era Retrospective analysis for Research and Applications (MERRA), and Japanese 55-year Reanalysis (JRA-55) for stream flow and annual water budget for two basins located in the diverse climatic settings in India. The results showed that these reanalysis products yielded satisfactory performance. Likewise, [15] evaluated three widely used global highresolution precipitation products, namely Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PER-SIANN-CDR), Tropical Rainfall Measuring Mission 3B42 Version 7 (TRMM 3B42V7), and CFSR against gauge observations and investigated their reliability as inputs in Soil and Water Assessment Tool (SWAT) to simulate streamflows. Results showed the potentiality of PERSIANN-CDR and 3B42V7 for streamflow prediction on at daily and monthly scale, respectively, over the two humid regions. At the same time, the performance of NCEP-CFSR for hydrological applications varies from basin to basin [11, 13, 16] used CFSR data for hydrological modelling in tropical and semi-tropical basins. The spatially interpolated CFSR data performed as well or better than single ground observations made further than 20-30 km, and sometimes better than individual weather stations less than 10 km from the basin centroid. This study further revealed the usefulness of CFSR data as a means to supplement ground records and consistently determine a baseline for hydrologic model performance. Fuka and others [13] used the CFSR to obtain historical weather data. They demonstrated its application to modelling five watersheds representing different hydroclimatic regimes in Ethiopia and the United States of America. Moreover, [16] assessed the applicability of CFSR climate data in model-
ling the hydrology of the Upper Blue Nile, Ethiopia. Results reveal that utilizing the CFSR precipitation and temperature data to force a watershed model provides stream discharge simulations that are as good as or better than models forced using conventional weather stations.
The satellite-based global CFSR climatic dataset for the period 1979-2013 on a daily basis might be used for hydrologic modelling to predict streamflow from the poorly gauged Mereb-Gash river basin. However, prior to usage of the global climatic datasets as input to hydrological modelling, various statistical processes should be undertaken the fact that this is the first attempt to utilize global climatic data for the aforementioned purpose in the study area. Such activities are essential to understand the human and climate variability impact on the environment under the influence of natural fluctuations in temperature, precipitation, and other factors. In this regard, the non-parametric Mann-Kendall (MK) and Sen's Slope (SS) estimator are used to estimate the magnitude of the trend for the reference time series of 32 global satellite-based stations. Moreover, the determination of PET and other variables and parameters are inevitable procedures, because they are not only inputs to hydrological modelling but also help to understand the basin characteristics. Therefore, the present study focuses on the statistical analyses, computation of PET and drought indices using different methods. In addition, analyses of spatio-temporal variability and distribution of various climatic parameters and
variables would be carried out, aiming at understanding basin characteristics. The authors strongly believe that this study will contribute to future hydrologic modelling studies in the area.
METHODS AND MATERIALS
Study area
Water resources of Eritrea are divided into five major river basin systems: Mereb-Gash, Barka-Anseba, Danakil, Setit, and the Red Sea. The Mereb-Gash river basin, originating from the Eritrean highlands and the Ethiopian plateau, is a transboundary river basin that crosses two political borders; Ethiopia and Sudan. It is known by Mereb and Gash in the upstream and downstream areas, respectively; hence the name Mereb-Gash that represents the whole basin is commonly used. Figure 1 shows the geographical location of the Mereb-Gash river basin. Its outlet is at Kassalatown (Sudan), which is located at 36°23'20" E longitude and 15°26'50 N latitude, having an area and main channel length of 22,900 km2 and 500 km, respectively. According to Euroconsult1, out of the total area 17,400 km2 lies in the Eritrean territory. It is a long, narrow basin, having a basin shape factor and average basin slope of 11 and 7 %, respectively. The flow of the Mereb-Gash does not usually reach the Nile River; rather, it is lost in the sands of the eastern plains of Sudan. Even though the river basin has limited streamflow data, it is believed to be one of the basins with the highest potential for
< n
<d e t o
i H G I
CO CO
J to
uI
^ I
n °
S 3
o S
=! (
o n
c w
n M n 0
S 6
r ® t (
S ) i
<D
01
Fig. 1. Geographical location of the study area
W DO
■ T
s □
s y c o <D *
10 10 o o 10 10 o o
o o
tv N
o o
N N
* 0
U 3 > in
E M
to in
¡1 <D <u
O £ —■ "t^ O
O «J CD <f i-l S = 5
iD
o O
LO CO CD O i
CD cd
<f)
I
ïl 0 (0 ® a ta >
modern technology-based irrigation that can provide sustainable agricultural development in the region.
The river flow has high variation during the wet season. The mean annual flow is estimated to be 680 million m3 [9] and maximum flow 1,000 m3s-1 [10]. It is dry for much of the year (November-April) but subjected to flash floods, transporting 40 million m3 of sediment [10] during the rainy season. The rainy season usually starts in June and ends in September with a series of flood flows. The flash floods are mainly due to mountainous and dissected topography of the upper-third and partly the middle-third of the basin, poor vegetation cover, unfavourable geological formations which are associated with poor soil formation and low infiltration rates. The downstream plains in the lower-third of the basin have been experiencing recurrent flooding; for example [9, 10] reported that the recurrence interval of severe floods for Kassala town is 1 to 5 years. Likewise, the western plain areas, e.g., the town of Tesseney, had been repeatedly exposed to flooding [17]. The long-term annual rainfall in the area is highly variable, usually ranging from over 900 mm to less than 300 mm [18]. The temperature in the area is also highly variable that depends on the season of the year; it is the highest in April and May and lowers in December and January. Typically minimum and maximum temperatures range from less than 10 °C in the highlands to more than 45 °C in the lowlands. Data sources
Climatic data, which are available globally at 38 km resolution, were obtained from the CFSR over a 35 years period (1979-2013) on a daily basis. These datasets include precipitation (P), minimum and maximum temperatures, wind speed, relative humidity, and solar radiation. Moreover, a high-resolution digital elevation dataset for the study area was acquired from The Shuttle Radar Topography Mission (SRTM) at 1 arc-second global resolution. The spatio-temporal analyses were done with the help of the Quantum Geographic Information System (QGIS). Data analyses
Non-parametric monotonic trendtest To test the CFSR dataset for monotonic trend, MK and SS are employed in this study. Non-parametric MK test is commonly employed to detect monotonic trends in a series of environmental, climatic, or hydrological data. The MK trend test assumes that under the null hypothesis of no trend, the time series is independent and identically distributed. The null hypothesis H0 is that the data come from a population with independent realizations and are identically distributed. The alternative hypothesis HA is that the data follow a monotonic trend. The univariate MK for a monotone trend in a time series (X,, j = 1, 2, 3, ..., n) of data is based on the test statistic S defined as follows:
S = J J sgn (X, - X, ),
(1)
k=l j=k+1
where X are the sequential data values, n the length of the data set, and
sgn ( x ) =
1 if x > 0, 0 if x = 0, -1 if x < 0.
(2)
Mann [18] and Kendall [19] have documented that when n > 8, S is approximately normally distributed with mean, E (S) = 0 and variance c2 given by
CT2 = |n(n -1)(2n + 5) -¿t, (t, -1)(2tj + 5)j/18, (3)
where p is the number of the ties of the extent i and t. is the number of data points in the jth tied group. The statistic S is approximately normally distributed, provided that the following Z-transformation is employed:
Z =
S -1 .
if S > 0,
CT
0 if S = 0,
(4)
S +1 .
if S < 0.
S is closely related to Kendall's t as given by:
where
D =
= s
T= D '
n -1) -1 ¿0(0 -1)
2 2 j=1
—n(n -1) 2
(5)
(6)
The magnitude of the slope of trend proposed by Sen [19] is estimated by:
b = Median
C xj xl ^
V/ < 0,
(7)
where b is the estimate of the slope of trend and xj is the lth observation. The slope determined by equation (7) is a robust estimate of the magnitude of the monotonic trend. Negative values indicate a decreasing trend, 0 values no trend, and positive values indicate an increasing trend.
Potentia/ évapotranspiration (PET)
There are several temperatures and radiation-based formulae for calculating PET, such as Blaney-Criddle, Turc, Penman-Monteith, Thornthwaite, Hargreaves, and so on. Three of these methods, which are built-in in the "SPEI" package in ^-programming language that uses user-selected input data, have been selected for the determination of PET. The first method computes the monthly PET according to the Thornthwaite [20] equation. It is the simplest of the three methods and can be used when only temperature data are available. But, variables that can affect PET, such as wind speed, surface humidity, and solar radiation, are not account-
CT
ed for. In cases where more data are available, a more advanced method to calculate PET is often preferred to make a more complete accounting of drought variability. However, these additional variables can have significant uncertainties. The second method uses the Hargreaves [21] equation that computes the monthly reference evapotranspiration ET0 of a grass crop. This method requires data on the mean external radiation. If such data are not available, it can be estimated from the latitude and the month of the year. The third method calculates the monthly reference evapotranspiration ET0 of a hypothetical reference crop according to the FAO-56 Penman-Monteith equation described in [22]. The original parameterization of Allen et al. [22] is used, corresponding to a short reference crop of 0.12 m height. The method requires data on the incoming solar radiation or sunshine hours or cloud cover, relative humidity, minimum and maximum temperature, wind speed, and geographical information. Drought characterization
A drought is a natural phenomenon that occurs over a variable period as a result of prolonged shortages in the water supply2. It can have a substantial impact on many economic sectors and people. Drought indicators or indices are often used to help monitor droughts. According to World Meteorological Organization (WMO)2, there are numerous types of drought indicators and indices; Aridity Anomaly Index (AAI), Aridity Index (AI), Palmer Drought Severity Index (PDSI), Standardized Precipitation Index (SPI), Rainfall Anomaly Index (RAI), Standardized Precipitation Evapotranspiration Index (SPEI) are some among others. A list of such methods, along with their strengths and limitations, is presentedin2. To assess the degree of droughts in the study area, SPEI based on 32 global satellite-based stations, within and around the Mereb-Gash basin, have been considered. The spatial and temporal comparisons of droughts are prepared by analysing the historical droughts by the characteristics of duration, severity, and magnitude.
The SPEI, a climate drought index which was proposed by [23] and modified by [24], was designed to take into account both precipitation and potential evapotranspiration (PET) in determining drought on a range of timescales. SPEI uses the basis of SPI [25] but includes a temperature component, allowing the index to account for the effect of temperature on drought development through a basic water balance calculation. Vicente-Serrano and others [23] showed that SPEI has the advantage of combining multi-scalar character with the capacity to include the effects of temperature
2 WMO and GWP, Integrated Drought Management Programme Handbook of Drought Indicators and Indices. 2016; 1173.
variability on drought assessment. Ever since its development, SPEI has been used in an increasing number of climatologic and hydrologic studies [24] based on 30-50 years climatic database.
SPEI uses the monthly (or weekly) difference between precipitation and PET. This represents a simple climatic water balance, which is calculated at different time scales to obtain the SPEI. For given values of P. and PET,, the climatic water balance D. that provides a simple measure of the water surplus or deficit for the analysed month is calculated as:
D =P- PET,
(8)
The calculated D values could be aggregated at different time scales using the same procedure as that of SPI [23]. The probability density function of a three-parameter log-logistic distributed variable is given by:
f (x)=pÍ x-i
a V a
p-1
1+ Í X-
(9)
where a, p, and y are scale, shape and origin parameters, respectively, for D values in the range (y > D < ro).
Parameters of the log-logistic distribution can be obtained following different procedures. The L-mo-ment procedure is the most robust and easy approach [26]. When L-moments are calculated, the parameters of the log-logistic distribution are givenby [27] as follows:
P =
2w1 - w0
a =
6w1 - w0 - 6w2 (wo - 2w1 )p
r(1+1/p)r(1-1/p)'
1
Y = w0-ar|1 + -lr|1 --
1
p
(10)
(11)
(12)
where r(P) is the gamma function of p.
In Vicente - Serrano and others [23], when the log-logistic a, p, and y distribution parameters were calculated, the probability-weighted moments (PWMs) method is used, based on the plotting-position approach [28], where the PWMs of orders are calculated as:
1 N
= -1(1 - F )' D,,
s N tT '' '
(13a)
where F. is a frequency estimator following the approach of Hosking [28]:
F=
,- 0.35
N '
(13b)
< DO
<D O t O
i H
G 1 cC
0 </> § </>
1 c
y 1
J to
^ I
n °
CD 3 o
zs ( o §
E w & N
§ 2 n 0
C 6
r ® t (
CD ) fi
<D
01
where i is the range of observations arranged in increasing order and N is the number of data, and D. is the difference between precipitation and potential evapo-
№ DO
■ T
s □
s y c o <D *
10 10 o o 10 10 o o
o o
tv N
o o
N N
¡É 0
U 3 > in
E M
to in
il <D <u
O ë
o
o _
lg<
o
co
™ O
o
ro
o
o O
LO co CD O i
CD cd
ÍD
Ï* iE 35
o iñ o a ta >
transpiration for the month i. The unbiased PWMs are obtained according to:
Table 1. SPEI classification of dry and wet conditions
1 N
=1E
N t?
N - i s
N -1 s
(14)
The probability distribution function of D according to the log-logistic distribution is then given by:
F ( x) = J f ( x) dx =
1 +
a
x - y
(15)
Category SPEI Probability, %
Extreme wet SPEI > 2.0 2.3
Severe wet 1.5 < SPEI < 2.0 4.4
Moderate wet 1.0 < SPEI < 1.5 9.2
Normal -1.0 < SPEI < 1.0 68.2
Moderate dry -1.5 < SPEI < -1.0 9.2
Severe dry -2.0 < SPEI < -1.5 4.4
Extreme dry SPEI < -2.0 2.3
With F(x) the SPEI can easily be obtained as the standardized values of F(x) as given in Vicente - Serrano [23]:
c0 + cw+c2w2
SPEI = W —
1 + djW + d2W + d3W
where
W = .y/-2ln(p) for p < 0.5
(16)
(17)
p is the probability of exceeding a determined D value, p = 1 - F(x). If p > 0.5, p is replaced by 1 -p and the sign of the resultant SPEI is reversed, C0 = 2.525517, C = 0.802853, C2 = 0.010328, dl = 0.432788, d2 = = 0.189269 and d3 = 0.0011308. SPEI has an intensity scale in which both positive and negative values are calculated, identifying wet and dry events (Table 1). The average value of the SPEI is 0, and the standard deviation is 1. SPEI of 0 indicates a value corresponding to 50 % of the cumulative probability of D, according to a log-logistic distribution.
RESULTS AND DISCUSSION
Precipitation data of 35 years on a daily basis from CFSR has been processed to find out the monthly, seasonally, and annual variability by using non-parametric statistical analyses. Fig. 2 shows the spatial variability of the long-term annual rainfall for the Mereb-Gash basin. It can be observed that rainfall is highly variable that comply with3- 4, ranging from over 1,100 mm in the upper-third to less than 500-700 mm in the lower-third and middle-third of the basin. A small area within the territory of the basin, which is located at the remotest part in the Ethiopian-Eritrean highlands, receives the highest rainfall. Visual comparison of the annual rain-
3 INC, Eritrea: Initial national communication, Asmara, Eritrea, 2001.
4 IWRM, Action Plan for Integrated Water Resources Management (IWRM) in Eritrea, Asmara, Eritrea, 2008.
Fig. 2. Spatial variability of long-term rainfall based on CFSR dataset for Mereb-Gash
w
s
fall from CFSR data with the one presented in various documents (e.g.3, [30]) shows that the former gives higher precipitation values. Substantial areas in the upper-third and lower-third receive rainfall, ranging from 330-715 mm. Fig. 3, a shows the seasonal cycle of the mean monthly rainfall at selected global station near the Gherset dam. The seasonal cycle of monthly rainfall of all the stations is about the same except for variability in magnitude; rainfall is peak during the rainy (summer) season, in the month of either July or August. Fig. 3, b, c, and d) shows linear regression trend analyses of long-term annual relative humidity, precipitation, and maximum temperature, respectively, for the selected station near the Gherset dam. The relative humidity and precipitation regression analyses indicate a decreasing trend, whereas maximum temperature shows an increasing
trend. Similar trends were observed in almost all the other global satellite-based stations.
MK and Sen's slope (SS) monotonic trend analysis methods are applied to long-term annual precipitation (P), temperature (T), and PET to P ratio, as shown in Table 2. MK's p-values of P, T and PET/P are higher than the significance level (a = 0.05) at 95 % confidence interval for 34.4, 37.5 and 31.3 % of the stations, respectively. Hence the null hypothesis of no significant monotonic trend in the series is accepted. On the contrary, the p-value of P, T and PET/P are less than the significance level (a = 0.05) at 95 % confidence interval for 65.6, 62.5 and 68.7 % of the stations, respectively; hence, the alternative hypothesis of the presence of significant monotonic trend is true. Moreover, SS results of P of all the stations are less than 0 at a 95 % confidence
y = -0.1807* + 3.2523 R =0.2897
5 6 7 8 Month
a
10 11 12
1979
1985 1991
1997 Year b
2003 2009
800 600 400
I 200 1 0 I 200
I -400 600 800
24,325* + 48552 R, = 0,5075
il|lll Mil
ÏÏ—1J
'U
u-
11
inr
1979 1984
1989
1994 Year
c
1999 2004 2009
u
!= 3
T3
1.5 -l
0.5 H
0
-0.5 -1 -1.5
y = 0.0464a- + 92.69 R= 0.2608
1979 1985 1991
1997 Year d
2003 2009
< DO
<D O t O
i H
G I cC
0 </> § </>
1 C
y 1
J CD
^ I
n °
c 3 o
=! (
o §
& N § 2
n g
r 6 t (
CD ) fi
(D
01
Fig. 3. Regression analyses on long-term climatic data of a selected station near Gherset dam, Mereb-Gash: a — mean monthly rainfall; b — mean annual relative humidity; c — precipitation; d — maximum temperature
W DO
■ T
s □
s y c o (D *
10 10 o o 10 10 o o
Table 2. MK p-values and SS for various climatic variables
Station code Elevation, m P T PET PET/P
MK SS MK SS mmy-1 MK SS
142381 1596 0.13 -13.02 0.26 0.02 2159.8 0.19 0.02
142384 1816 0.18 -6.60 0.16 0.02 2105.7 0.15 0.03
142388 1762 0.35 -3.00 0.10 0.03 1970.1 0.32 0.09
142391 2021 0.00 -13.00 0.01 0.03 1895.9 0.00 0.08
142394 2797 0.07 -11.73 0.00 0.04 1483.1 0.02 0.01
145378 1057 0.00 -27.30 0.08 0.03 2178.0 0.00 0.04
145381 1238 0.13 -6.00 0.06 0.03 2446.3 0.14 0.20
145384 1708 0.05 -8.10 0.02 0.03 2087.1 0.05 0.05
145384 1163 0.00 -20.83 0.00 0.05 1994.2 0.00 0.23
145388 1413 0.00 -6.60 0.00 0.03 2108.5 0.00 0.18
145391 1670 0.00 -16.03 0.00 0.05 1906.4 0.00 0.08
145394 2331 0.06 -11.40 0.00 0.04 1593.7 0.03 0.02
148369** 835 0.00 -24.10 0.03 0.03 2330.8 0.00 0.10
148372 1155 0.00 -34.10 0.05 0.03 2266.8 0.00 0.08
148375 843 0.00 -33.80 0.01 0.03 2257.1 0.00 0.09
148378 1163 0.01 -17.34 0.05 0.03 2221.7 0.01 0.04
148381 1270 0.01 -13.90 0.10 0.03 2129.2 0.03 0.04
148384 1634 0.00 -11.60 0.05 0.03 1912.5 0.00 0.05
148388 1762 0.00 -15.40 0.01 0.03 1970.1 0.32 0.10
148391 1545 0.00 -14.90 0.00 0.05 1854.1 0.00 0.07
148394 2324 0.35 -5.10 0.01 0.03 1779.6 0.27 0.03
151366 581 0.00 -20.80 0.00 0.04 2405.7 0.00 0.24
151369t 635 0.00 -27.40 0.00 0.04 2322.4 0.00 0.15
151372 878 0.00 -25.80 0.00 0.03 2324.8 0.00 0.10
151375" 917 0.01 -11.34 0.03 0.03 2409.2 0.01 0.09
151378 972 0.13 -6.00 0.06 0.03 2394.6 0.13 0.19
151381 995 0.01 -8.34 0.06 0.03 2293.4 0.01 0.18
151388* 2430 0.35 -8.63 0.05 0.03 1758.5 0.38 0.01
151391 1644 0.48 -4.20 0.02 0.03 1824.2 0.57 0.02
155366 555 0.00 -18.00 0.00 0.03 2381.4 0.00 0.31
155369 833 0.00 -20.36 0.00 0.03 2435.2 0.00 0.20
155372 833 0.00 -20.36 0.00 0.03 2350.5 0.00 0.20
o o
N N
o o
N N
* 0
U 3 > in
E M
to in in 0
¡1 <D <u
O g
---' "t^
O
o cj
CD <f
8 « Z ■ ^
w «
ot E — ■
c
E o ¿T O
Ln O
S 1
o E
CD ^
T- ^
£
CO o
■S
I
i!
0 (0 ® a
CO >
NB: Station codes with asterisk represent stations that are used hereafter for illustration purposes; *nearAdi-Halo dam, **near Ghersetdam,^nearFanco dam, and "near Haikota town.
level, indicating a significant decreasing trend. But, the statistical analyses of whole reference time series data of T and PET/P highlight that the trends look predominantly increasing at a 95 % confidence level.
Other credible studies in the area could not verify these results, but regional studies give more or less similar indications. For example, a regional study on precipitation and temperature variables in the Horn of
Africa [31] over the period 1930-2014 showed that statistically, the trend of precipitation decreased insignificantly; the trend of temperature was observed to drop very significantly between 1930 and 1969, but it was dramatically elevated from 1970 to 2014. These overall trends are in line with the global predictions that revealed increased land surface air temperature and evapotranspiration and decreased precipitation amount,
in interaction with climate variability and human activities, leading to desertification in Sub-Saharan Africa (where the study area is located), parts of East and Central Asia, and Australia5.
Using the Thornthwaite, Hargreaves, and Penman-Monteith methods from the "SPEI" package in R programming language, monthly and annual PET were calculated for all the 32 stations. Fig. 4 shows the comparison of PET estimation methods at different times of the year at two stations in the upper-third near the Adi-Halo dam and lower-third near the Gherset dam. The stations near Adi-Halo and Gherset dams are located at different elevations; 2,430 and 835 m above mean sea level, respectively. The Thornthwaite method was found to be highly dependent on the elevation of the station; it produced much lower PET values in loca-
5 IPCC "Climate Change and Land: IPCC Special Report on Climate Change, Desertification, Land Degradation, Sustainable Land Management, Food Security, and Greenhouse gas fluxes in Terrestrial Ecosystems-Summary for Policymakers." 2019.
tions at higher altitudes as compared to the other two methods. Fig. 4, a shows PET of the three methods at a station near the Adi-Halo dam with the Thornthwaite and Penman-Monteith method providing the lowest and highest values, respectively, whereas Hargreaves-based PET for the most part standing between the two methods. For example, in January, it can be seen that the Thornthwaite method is giving PET of 49.6 and 61.8 % lower than Hargreaves, and Penman-Monteith, respectively and the same is more or less true for the remaining months. Hargreaves and Penman-Monteith methods do not show such noticeable dependence on the elevation of a station and produced consistent PET, as shown in Fig. 4, a and b. In almost all cases, Penman-Monteith happens to be with the highest PET values.
Regression analyses were carried out to establish a relationship between the two methods with respect to Penman-Monteith. The results show a polynomial line of 2nd degree, having a coefficient of determination R2 equal to 0.877 and 0.781 for Hargreaves vs. Pen-man-Monteith and Thornthwaite vs Penman-Monteith,
250
200 ■
S 150 E t-'
ft- 100 -
50 -
0 4-1
Q Hargreaves □ Penman
! Thornwhaite
6 7 Month
a
10
12
400
: 2oo
6 7 Month
b
10 11
< DO
<D O t O
i H G 1
cC
y
o n
I C
y 1
J to
El n
CD 3 o
=¡ ( n
E № & N
§ 2 n 0
D £
r 6 t (
D ) Í
<D
01
Fig. 4. Comparison of PET estimation methods in Mereb-Gash basin: a — at the upper third of the basin near Adi-Halo dam; b — at the lower third of the basin near the Gherset dam
№ DO ■
s □
s y c o <D X
10 10 o o 10 10 o o
o o
tv N
o o
N N
* 0
U 3 > in
E M
to in
¡1 <D <u
o £
o
o _
g<
o
co
™ O
o
ro
o
o O
LO CO CD O i
CD cd
iD
■S
I
0 (0 ® a m >
y = 0.00053x: - 1 .843.*; + 3268 R2 = 0,877
1800 1900 2000 2100 2200 2300 2400 2500 PET-Penman, mm/year
a
-j
40
20
-20
-40
1150
k 1100
%
I 1050
I .1 till I if,, 1 ?
J! i 1 i
1900 2000 2100 2200 PET-Pcnman, mm/vcar b
2300
2400
850 1800
1900 2000 2100 2200 2300 PET-Pcnman, mm/ycar
c
2400 2500
60 40 20 1 0 CJ d 20 40 -60 - -
T * , J J, r
ll 11 ' I
1 ■
1900 2000 2100 2200 PET-Penman, mm/vear d
2300
2400
Fig. 5. Example on the relationship between PET estimation methods at a station near Adi-Halo dam: a — Hargreaves vs. Penman-Monteith; b — residual plot of Hargreaves vs. Penman-Monteith; c — Thornthwaite vs. Penman-Monteith; d — residual plot of Thornthwaite vs. Penman-Monteith
respectively (Fig. 5, a and c). Similarly, the residual plots shown in Fig. 5, b and d demonstrate randomly dispersed points around the horizontal axis, indicating a linear regression model could be appropriate for the data. It is obvious that such relationships could not be established for areas with different climatic and topographic conditions. For example [32], obtained various relations for different climatic conditions. Fig. 6 shows
the spatial variability of long-term annual PET of the Mereb-Gash basin for the values presented in Table 2 The upper-third of the area has the lowest PET, which is logical and continues to increase as we move towards the middle-third and lower-third portions of the basin.
Fig. 7 provides information about drought conditions at shorter and longer time scales (1, 3, 6, and 12 months) SPEI at a station near the Fanco dam in
Fig. 7. The SPEI at lower-third of the Mereb-Gash basin near the Fanco dam at different time scales: a — 1 month; b —
& '(0(0 3 months; c — 6 months; d — 12 months O
Mereb-Gash. The period until the late 1990's show quasi-continuous humid conditions, with some minor drought periods. The persistent drought conditions during 20002013 are also clearly identified by the SPEI, independent of the time scale. Remarkably enough, the SPEI values of all the stations show similar trends. The spatial variability of the annual SPEI for the driest year (2003) for the whole basin is presented in Fig. 8. Unfortunately, the whole area had been exposed to severe and extreme drought conditions the fact that SPEI values are < -1.0.
The same procedure was employed to the wettest year (1998), providing positive SPEI values (Fig. 9). As per the SPEI classification (Table 1), the basin was predominantly under normal and moderate wet conditions.
These findings are in congruence with global, regional, and local studies. For example, intergovernmental panel on climate change, IPCC5 showed that over the period 1961-2013, the annual area of drylands has increased with sizeable inter-annual variability. Also, the frequency and intensity of droughts have increased in
o o
N N
o o
N N
* 0
U 3 > in
E M
to in in 0
¡1
Fig. 8. Spatial variability of the SPEI for the driest year (2003)
<U <u
O g —■ "t^ o
O u
CD <f
3 *
cm 5
iD
in
o O
LO CO CD O i
cd cd
£
in O
■S
I
iE 35
o (0 ® a
CO >
Fig. 9. Spatial variability of the SPEI for the wettest year (1998)
some regions, including the Mediterranean, West Asia, many parts of South America, much of Africa, and north-eastern Asia. A regional study on drought in the Horn of Africa [31] over the period 1930-2014 using the SPEI showed that the region experienced from mild to moderate drought throughout the study period with severe to extreme drought in some regions, particularly in 1943, 1984, 1991, and 2009. Thus, it would be necessary to consider the IPCC5 recommendation on actions that should be taken in the near-term, based on existing knowledge, to address desertification, land degradation, and food security while supporting longer-term responses that enable adaptation and mitigation to climate change. These include actions to build individual and institutional capacity, accelerate knowledge transfer, enhance technology transfer, and deployment, enable financial mechanisms, implement early warning systems and undertake risk management.
CONCLUSIONS
Conclusions inferred from this study could be summarized in the following statements:
The statistical analyses of the satellite-based global CFSR climatic data show that there is predominantly a significant monotonic trend. MK's ^-values of precipitation, temperature, and potential evapotranspiration are less than the significance level (a = 0.05) at a 95 % confidence interval for the majority of the stations. Hence, the null hypothesis of no significant monotonic trend is
rejected. Moreover, SS analyzes son precipitation from all the stations are less than 0 at a 95 % confidence level, indicating a decreasing trend. But, temperature and potential evapotranspiration trends were found to be predominantly increasing at a 95 % confidence level.
Among the potential evapotranspiration estimation methods, the Thornthwaite method was highly dependent on the elevation of the station. It produced much lower potential evapotranspiration values in locations at higher altitudes than at lower altitudes as compared to Hargreaves and Penman-Monteith. On the other hand, latter methods did not show such noticeable dependence on the elevation of a station and produced consistent estimation. In most parts of the cases, it was found that Penman-Monteith was providing the highest potential evapotranspiration estimations among the three methods.
The SPEI drought index of all stations shows sustained dry conditions during 2000-2013 and predominantly wet conditions during 1979-2000.
Finally, the reasons for the presence of significant trend in most of the climatic variables and persistent drought conditions in recent years in the study area in particular and in the region, in general, are highly likely to be linked to human activities and climate variability. That resulted from the land-use change, the land-use intensification and climate change contributing to desertification, and land degradation. Hence, concerted efforts are badly needed to reverse these human and climate-driven environmental repercussions.
< n
<D O t O
i H
G 1
cC
cn cn
REFERENCES
1. McGlynn B.L., Bloschl G., Borga M., Bormann H., Hurkmans R., Komma J. et al. A data acquisition framework for runoffprediction in ungauged basins. Runoff Prediction in Ungauged Basins, 2013; 29-52. DOI: 10.1017/CBO9781139235761.006
2. Hrachowitz M., Savenije H.H.G., Bloschl G., McDonnell J.J., Sivapalan M., Pomeroyet J.W. et al. A decade of Predictions in Ungauged Basins (PUB) — a review. Hydrological Sciences Journal. 2013; 58(6):1198-1255. DOI: 10.1080/02626667.2013.803183
3. Sivapalan M., Takeuchi K., Franks S.W., Gupta V.K., Karambiri H., Lakshmi V. et al. IAHS Decade on Predictions in Ungauged Basins (PUB), 2003-2012: Shaping an exciting future for the hydrological sciences. Hydrological Sciences Journal. 2003; 48(6):857-880. DOI: 10.1623/hysj.48.6.857.51421
4. Beven K.J. Rainfall-Runoff modelling: The primer. Chichester, UK, John Wiley & Sons, Ltd, 2012. DOI: 10.1002/9781119951001
5. Gehbrehiwot A., Kozlov D. GIUH-Nash based runoff prediction for Debarwa catchment in Eritrea. E3S
Web of Conferences. 2019; 97:05001. DOI: 10.1051/ e3sconf/20199705001
6. Kozlov D.V., Ghebrehiwot A.A. Efficacy of digital elevation and Nash models in runoff forecast. Magazine of Civil Engineering. 2019; 87(3):103-122. DOI: 10.18720/MCE.87.9
7. Alemngus A., Mathur B.S. Geomorphologic instantaneous unit hydrographs for rivers in Eritrea (East Africa). Journal of Indian Water Resources Society. 2014; 34(1):1-14.
8. Ghebrehiwot A.A., Kozlov D.V. Hydrological modelling for ungauged basins of arid and semi-arid regions: review. Vestnik MGSU [Monthly Journal on Construction and Architecture]. 2019; 14(8):1023-1036. DOI: 10.22227/1997- 0935.2019.8.1023-1036
9. Bashar K.E. Gash river flash floods challenges to Kassala town: Mitigation and risk management. 2011.
10. Elhassan E.S.E., Ibrahim A.M., Abdalla A. Flood modeling water appraisal and land reclamation: a case study of gash river. SUST Journal of Engineering and Computer Science (JECS). 2015; 16(3):37-45.
y -»■ J to
^ I
n CD
C 3
0 CC
=! (
01
o n
c w
n M n 0
C £
r 6 t ( 1°
cc ) Í
<D
01
W DO
■ T s □
s y c o <D *
10 10 o o 10 10 o o
11. Auerbach D.A., Easton Z.M., Walter M.T., Flecker A.S., Fuka D.R. Evaluating weather observations and the Climate Forecast System Reanalysis as inputs for hydrologic modelling in the tropics. Hydrologi-calProcesses. 2016; 30(19):3466-3477. DOI: 10.1002/ hyp.10860
12. Dile Y.T., Srinivasan R. Evaluation of CFSR climate data for hydrologic prediction in data-scarce watersheds: An application in the Blue Nile river basin. JAWRA Journal of the American Water Resources Association. 2014; 50(5):1226-1241. DOI: 10.1111/ jawr.12182
13. Fuka D.R., Walter M.T., MacAlister C., De-gaetano A.T., Steenhuis T.S., Easton Z.M. Using the climate forecast system reanalysis as weather input data for watershed models. Hydrological Processes. 2014; 28(22):5613-5623. DOI: 10.1002/hyp.10073
14. Mahto S.S., Mishra V. Does ERA-5 outperform other reanalysis products for hydrologic applications in India? Journal of Geophysical Research: Atmospheres. 2019; 124(16):9423-9441. DOI: 10.1029/2019jd031155
15. Zhu Q., Xuan W., Liu L., Xu Y.-P. Evaluation and hydrological application of precipitation estimates
o o derived from PERSIANN-CDR, TRMM 3B42V7, and o o NCEP-CFSR over humid regions in China. Hydrologi-* " calProcesses. 2016; 30(17):3061-3083. DOI: 10.1002/ ^ a hyp.10846
£ ¡a 16. Dile Y.T., Srinivasan R. Evaluation of CFSR
| — climate data for hydrologic prediction in data-scarce ® ¡¡2 watersheds: an application in the Blue Nile River Ba-Jj* 0) sin. JAWRA Journal of the American Water Resources s |= Association. 2014; 50(5):1226-1241. DOI: 10.1111/ |2| jawr.12182
JL. . 17. Mann H.B. Nonparametric tests against trend.
J § Econometrica. 1945; 13(3):245. DOI: 10.2307/1907187
O <u 18. Kendall M.G. Rank correlation methods.
o ^ 4th ed. London, Charles Griffin, 1975. o y
< 19. Sen P.K. Estimates of the regression coef-c ficient based on Kendall's Tau. Journal of the Ameri-og c can Statistical Association. 1968; 63:1379-1389. DOI: z | 10.1080/01621459.1968.10480934 22.g 20. Thornthwaite C.W. An approach toward a rati § tional classification of Climate. Geographical Review. £ ^ 1948; 38(1):55. DOI: 10.2307/210739 g ° 21. Hargreaves G.H., Samani Z.A. Reference o E crop evapotranspiration from temperature. Applied Ei ° Engineering in Agriculture. 1985; 1(2):96-99. DOI:
^^ 10.13031/2013.26773 2: £ £
22 J
3 Received September 22, 2019.
^ jj Adopted in a revised form on November 23, 2019.
v ® Approved for publication December 29, 2019. s *
iE 35
o in ® a
CO >
22. Allen R.G., Smith M., Pereira L.S., Perrier A. An update for the calculation of reference evapotranspiration. ICID Bulletin. 1994; 43(2):35-92.
23. Vicente-Serrano S.M., Beguería S., López-Moreno J.I. A multiscalar drought index sensitive to global warming: The standardized precipitation evapotranspiration index. Journal of Climate. 2010; 23(7):1696-1718. DOI: 10.1175/2009jcli2909.1
24. Beguería S., Vicente-Serrano S.M., Reig F., Latorre B. Standardized precipitation evapotranspiration index (SPEI) revisited: Parameter fitting, evapotranspiration models, tools, datasets and drought monitoring. International Journal of Climatology. 2014; 34(10):3001-3023. DOI: 10.1002/joc.3887
25. Mckee T.B., Doesken N.J., Kleist J. The relationship of drought frequency and duration to time scales. Proceedings of the Eighth Conference on Applied Climatology. 1993; 179-184.
26. Ahmad M.I., Sinclair C.D., Werritty A. Log-logistic flood frequency analysis. Journal of Hydrology. 1988; 98(3-4):205-224. DOI: 10.1016/0022-1694(88)90015-7
27. Singh V.P., Guo H., Yu F.X. Parameter estimation for 3-parameter log-logistic distribution (LLD3) by Pome. Stochastic Hydrology and Hydraulics. 1993; 7(3):163-177. DOI: 10.1007/bf01585596
28. Hosking J.R.M. L-Moments: analysis and estimation of distributions using linear combinations of order statistics. Journal of the Royal Statistical Society: Series B (Methodological). 1990; 52(1):105-124. DOI: 10.1111/j.2517-6161.1990.tb01775.x
29. Van Rooy M.P. A rainfall anomaly index independent of time and space. Notos. 1965; 14:43-48.
30. Alemngus A., Amlesom S., Bovas J.J.L. An overview of Eritrea's water resources. International Journal of Engineering Research and Development. 2017; 13(3):74-84.
31. Ghebrezgabher M.G., Yang T., Yang X. Long-term trend of climate change and drought assessment in the Horn of Africa. Advances in Meteorology. 2016; 2016:1-12. DOI: 10.1155/2016/8057641
32. Gotardo J.T., Rodrigues L.N., Gomes B.M. Comparison of methods for estimating reference evapotranspiration: An approach to the management of water resources within an experimental basin in the Brazilian cerrado. Engenharia Agrícola. 2016; 36(6):1016-1026. DOI: 10.1590/1809-4430-eng.agric. v36n6p1016-1026/2016
Bionotes: Anghesom A. Ghebrehiwot — Lecturer, Department of Agricultural Engineering; Hamelmalo Agricultural College, National Higher Education and Research Institute; Keren-397, Keren, Eritrea; bahghi2012@ gmail.com;
Dmitriy V. Kozlov — Doctor of Technical Sciences, Professor, Head of the Department of Hydraulics and Hydraulic Engineering; Moscow State University of Civil Engineering (National Research University) (MGSU);
26 Yaroslavskoe shosse, Moscow, 129337, Russian Federation; SPIN-code: 5878-6674; Scopus: 36787104800, ResearcherlD: B-4808-2016, ORCID: 0000-0002-9440-0341; [email protected].
Поступила в редакцию 22 сентября 2019 г. Принята в доработанном виде 23 ноября 2019 г. Одобрена для публикации 29 декабря 2019 г.
Об авторах: Ангхесом Алемнгус Гебрехивот — преподаватель кафедры агроинженерии; Аграрный колледж Хамелмало, Национальный институт высшего образования и исследований; Кэрэн-397, Кэрэн, Государство Эритрея; [email protected];
Дмитрий Вячеславович Козлов — доктор технических наук, профессор, заведующий кафедрой гидравлики и гидротехнического строительства; Национальный исследовательский Московский государственный строительный университет (НИУ МГСУ); 129337, г Москва, Ярославское шоссе, д. 26; SPIN-код: 5878-6674, Scopus: 36787104800, ResearcherlD: B-4808-2016, ORCID: 0000-0002-9440-0341; [email protected].
< П
о е t с
iH
G Г сС
У
o со
§ CO
У z
У 1
J со
^ I
n °
с 3 o
=! ( o&
о §
E w & N
§ 2 n 0
r 6 t (
CD ) ¡1
®
01
« DO
■ T
s у с о Ф *
10 10 о о 10 10 о о
