See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/295074267

The Climate of Ethiopia

Chapter · October 2015

DOI: 10.1007/978-94-017-8026-1_3

CITATIONS READS

24 25,525

3 authors:

Massimiliano Fazzini Carlo Bisci

University of Ferrara University of Camerino
37 PUBLICATIONS   123 CITATIONS    30 PUBLICATIONS   136 CITATIONS   

SEE PROFILE SEE PROFILE

Paolo Billi
University of Ferrara
99 PUBLICATIONS   1,361 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

RItMare View project

Development of Next-Generation Sustainable Land Management (SLM) Framework to Combat Desertification View project

All content following this page was uploaded by Carlo Bisci on 19 February 2016.

The user has requested enhancement of the downloaded file.

 The Climate of Ethiopia
3
Massimiliano Fazzini, Carlo Bisci, and Paolo Billi

Abstract
Several papers have been published on different issues regarding the climate of Ethiopia or of
some specific region. This presentation attempts to revise the knowledge of the climate of
Ethiopia by means of updated, longer time series and including a larger number of meteo-
stations than previous studies. Basic climatic parameters such as temperature, rainfall, relative
humidity, wind, evapotranspiration, and aridity are considered and their spatial distribution is
analyzed. The main results of such elaborations have been regionalized to obtain climatic
maps by means of geostatistical interpolation, also taking into account topogeographic
variables. These parameters were also used to update the Köppen classification of the
Ethiopian territory. Climate change is a very important issue with worrying repercussions on
agriculture and hence the social and economic development of the country. Trends of
temperature and annual, spring, and summer rains were interpolated for the last 3–5 decades.
Temperature shows a markedly increasing trend especially as regards the minimum values,
whereas annual rainfalls tend to decrease with the spring, ‘small rains’ decreasing at faster rate.

Keywords
Climate Ethiopia Temperature Precipitation Aridity index

3.1 Introduction large spatial variations in temperature and precipitation (see

Sect. 3.3). The climate of Ethiopia is therefore mainly con-
Ethiopia is among the largest countries of Africa and it is trolled by the seasonal migration of the Intertropical Con-
characterized by a wide variety of landscapes, with marked vergence Zone (ITCZ) and associated atmospheric
contrasts in relief and altitudes ranging from about 155 m circulations as well as by the complex topography of the
below sea level of Assale Lake, in the Danakil depression, to country. Landscapes with contrasting characteristics in terms
about 4,533 m a.s.l. at Ras Dejen (EMA 1988). For these of physiography and elevation, such as the highlands and the
reasons and given its geographic position close to the lowlands, experience a variety of climates from desert cli-
equator and the Indian Ocean, the country is subjected to mate to that typical of equatorial mountains.
Climate, in turn, has many obvious implications on
landforms and morphodynamic evolution of natural land-
M. Fazzini
P. Billi (&)
scapes as much as on the living conditions of local people, in
Physics and Earth Sciences, a country whose economy is heavily dependent on rain-fed
University of Ferrara, Ferrara, Italy agriculture (Hadgu et al. 2013).
e-mail: [email protected] The relevance of climate and its variability on natural
M. Fazzini resources of Ethiopia and their potential exploitation has been
e-mail: [email protected] recognized since long by the first Western explorers visiting
C. Bisci the country. The establishment of a few weather stations with
School of Science and Technology, regular observations since the beginning of the twentieth
University of Camerino, Camerino, Italy

P. Billi (ed.), Landscapes and Landforms of Ethiopia, World Geomorphological Landscapes, 65

DOI 10.1007/978-94-017-8026-1_3, © Springer Science+Business Media Dordrecht 2015
66 M. Fazzini et al.

century led Ethiopia to assume a prominent position in the (Lovett and Wood 1976), nonetheless, they are of great
Horn of Africa as far as the availability of reference meteo- interest for a preliminary description of the general climate
rological data is concerned, also within the operational of Ethiopia and surrounding regions at the nineteenth–
framework of international climate organizations. twentieth-century transition. Given their spatial distribution,
Several studies on the climate of Ethiopia have been these data can be considered suitable only for a climatic
published throughout the past two centuries (see the next point characterization and as a baseline, though some sta-
sections), but only very few of them benefited from the tions may have changed location and instrumentation, to
inherent, ongoing expansion of the time series in refining the study climate variations, if any, but do not allow to define
definition of the different climate types and their spatial the climate of Ethiopia with sufficient detail.
variability. Instead, they rather focused on specific topics,
with climate change and its implications becoming the most
investigated theme in the last decade. The aim of this chapter 3.2.2 More Recent Data and Studies
is therefore to update the knowledge of the main character-
istics of the climate of Ethiopia and to analyze the occur- In Ethiopia, most of the longest rainfall and temperature
rence of climate change trends on the base of time series records began in the 1950s and 1960s. Many other records
longer than those used in previous studies. started during the 1980s, following the impulse from the
Ethiopian Government to expand the network, partly in
response to the mid-1980s drought. In 1951, a small mete-
3.2 Previous Studies orological unit was formally established within the Civil
Aviation Department of Ethiopia, to provide meteorological
3.2.1 Historical Data information solely for flying purposes. As the need for
meteo-climatic data increased, a Meteorological Department
In sub-Saharan Africa, century long time series of meteo- was established in 1964, under the Civil Aviation Authority.
climatic data are scarce and sparse over large portions of Later, this department became the modern National Meteo-
territory (Houghton et al. 2001; Conway et al. 2004). Ethi- rological Agency (NMA), after designation as an autono-
opia is a large country, with very few well-documented long mous organization on December 31, 1980 (ENRAEMED
series of climate observations. The longest and uninterrupted 2003).
climate record for the Horn of Africa is the rainfall and Since 1984, FAO (1984) has been publishing mean
thermometric (but also wind, pressure, relative humidity, monthly agroclimatic data, including decadal rainfall data
and sunshine) data record from Asmara which began in 1890 for Ethiopia. FAO obtained these data from the National
and from Addis Ababa since 1898. These data derived from Meteorological Agency and Chernet (1982) used them for a
four intermittent sources, covering different time intervals: concise description of the Main Ethiopian Rift Valley cli-
Russian (1898–1903), Italian (1903–1904), British (1901– mate, whereas Gamachu (1988) based on them his study on
1904), and French (1898). climatic elements of mountainous regions of Ethiopia.
Some preliminary data on temperature, rainfall, relative In 1994, NOAA–NCDC put together a huge database at
humidity, and wind for Addis Ababa and Addis Alem, plus global scale, the ‘Global Historical Climate Network
sparse climatic information for a few other sites in Ethiopia, (GHCN),’ which includes daily data of several meteo-sta-
were reported by Eredia and De Castro (1914), but the tions located in Ethiopia, the source of which, however, is
quality of these data is questionable for both the short always the NMA.
duration and the methodology of measurements.
The study by Fantoli (1940) considered the meteo-cli-
matic data of three thermometric stations in Ethiopia—Addis 3.2.3 The National Atlas
Ababa (1902–1930), Gambela (1914–1932), and Harar
(1902–1918)—and other six thermometric stations, includ- In 1988, the Ethiopian Mapping Authority published the first
ing Asmara (1890–1930), located in the bordering territories, edition of the National Atlas of Ethiopia. The Atlas is likely
and 18 rain gauges, some of which were located beyond the first publication with a comprehensive description of the
modern Ethiopia. For a few of these meteorological stations, climate of Ethiopia. In fact, it includes several maps
additional meteo-climatic parameters such as atmospheric depicting the spatial distribution of a few climatic parameters
pressure, relative humidity, wind speed and direction, and such as rainfall, rainfall patterns, hail, frost, temperature,
insolation were reported. sunshine hours, climate classification, wind velocity and
These data series were used by Hurst and Black (1937), direction and the location of the meteorological stations as
Fantoli (1940, 1965), WWR (1959), and Conway et al. well. The data to prepare these climate sections of the Atlas
(2004). Though some stations have incomplete records were provided by the Ethiopian Meteorological Agency as
3 The Climate of Ethiopia 67

far as 1977, and the data analysis and summary are partly evapotranspiration, whereas the spatial and temporal distri-
based on the studies of Gamachu (1977). Though the spatial bution of temperatures have not been widely and satisfac-
distributions of climatic parameters for the three decades torily analyzed. The majority of modern studies deal with
before the mid-1970s are reported, this publication still limited areas and, commonly, are constrained by the avail-
represents a reliable reference and a milestone in the history ability of homogeneous time series of good quality. For
of climatological studies in Ethiopia. This paper draws on these reasons, modern investigations on the general charac-
the inspiring climatic maps and chapters of the National teristics of the climate of Ethiopia are missing, with the
Atlas of Ethiopia and attempts to update the analysis of exception of the National Meteorological Agency publica-
climatic parameters and climate classification considering tions (e.g., Tadege 2001, 2007) that provide a reliable
also the data of the following three decades as far as 2010. description of the climate across the whole country, though
limited to the end of the second millennium and early
beginning of the third. In the last decade, the NMA has also
3.2.4 Modern Studies expanded the number of recording stations to a remarkable
quantity and presently the entire territory of Ethiopia is
In the scientific literature, there are a number of papers on densely, though heterogeneously, covered (Fig. 3.1).
precipitation in Ethiopia and related problems, such as In the 2007 NMA report, temperature data are considered,
extreme rains, droughts, trend variations, and and for the 1951–2006 interval, increasing rates of 0.13 °C/

Fig. 3.1 Mean annual temperatures

68 M. Fazzini et al.

decade and 0.37 °C/decade for mean annual minimum and analyzed the temperature and rainfall trends for the 1948–
maximum temperature, respectively, are presented. For the 2006 interval and performed a simulation projecting these
1960–2006 interval, the report by McSweeney et al. (2010) trends to 2050 under the assumption of a gradual doubling of
indicates a mean annual temperature increase of 0.28 °C per greenhouses gases. For the time series bounded to 2006,
decade and a significant increasing trend in the frequency of these authors find that air temperatures increase at the rate of
hot days and a decrease of cold days. These authors also about 0.03 °C per year across most of Ethiopia, with the
found that no statistically significant trend is recognizable exceptions of the lowlands and the northern Rift Valley,
for mean precipitation in any season. whereas rainfall trends indicate a weak increase in the arid
In a more recent study on the upper Blue Nile basin, lowlands of southeastern Ethiopia.
based on statistical and geostatistical techniques and gridded Droughts have been studied by Shanko and Camberlin
data, reconstructed from NMA weather stations and meteo- (1998)—in connection with the development of tropical
rological satellite records, Mengistu et al. (2013) found an cyclones—and by Viste et al. (2013). These latter authors,
insignificant change in temperature across the 1981–2010 analyzing rainfall data for the 1971–2011 interval, found that
interval. the drought patterns are largely influenced by the variation in
Mekasha et al. (2013) have analyzed temperature and the seasonal precipitation cycle among different areas of
rainfall extremes for 11 stations and found a general ten- Ethiopia.
dencies of increasing warm and decreasing cold extremes, Scarce information is available on other climatic parameters
whereas trends in precipitation extremes were much more such as velocity and direction of winds, air humidity, barometric
variable and increasingly inconsistent among neighboring pressure, and evapotranspiration (Enku and Melesse 2013).
stations. This study, however, does not cover the wide Though the climate of Ethiopia received a lot of attention
diversity of topography and landscapes of Ethiopia; there- from scientists and many papers have been published on
fore, its results cannot be considered as representative for the different climatic topics, given its complexity and variability
whole country. and the availability of longer time series, an updated sum-
Yet, the study of Selashi and Zanke (2004) on rainfall mary and description of the main parameters is considered
amount and rainy days changes across the country is based here of great help for those interested in the subject.
on the data from 11 meteo-stations spanning the 1965–2002
interval. These authors found no trend in the annual, kiremt
(the main, monsoon-type rainy season in the local language, 3.3 Summary Description
see the next sections) and belg (the small, spring time of the Climate of Ethiopia
rainfall in the local language) rainfall totals and rainy days
over the largest part of Ethiopia. A study on extreme rainfall The climate of Ethiopia is mainly controlled by the seasonal
and dry spell events across the same 1965–2002 interval, migration of the ITCZ and associated atmospheric circulations
based upon the same time series, has been carried out by (Beltrando and Camberlin 1993), the complex physiography
Selashi and Camberlin (2006) who found contrasting, and the marked contrast in elevation among large different
though weak, trends in extreme precipitation for both the areas of the country. The transition between lowlands and
kiremt and the belg in different geographic areas, showing a highlands is commonly very sharp, resulting in a variety of
high variability for both rainfall seasons. These results climates, from very arid to very humid typical of equatorial
contrast with a previous work by Easterling et al. (2000), mountains, with further differentiation at local scale. More-
whose findings indicate that extreme intensity is decreasing over, precipitation varies with latitude, decreasing from south
over the Ethiopian highlands. to north, whereas the meteorological framework is deeply
Other papers report climatic information on specific areas affected also by elevation and physiography, especially as
or regions of Ethiopia (e.g., Conway 2000a, b; Bewket and regards temperature distribution and anemometric character-
Conway 2007; Taye and Zewdu 2012). The paper by istics. Finally, the proximity of the Asiatic continent has to be
Bewket and Conway (2007) reports about the temporal and considered as well. In winter, in fact, the contrast between the
spatial variability of rainfall for a relatively small portion of thermal anticyclone of western Asia and Egypt and the
the country, the Amhara Region, and found no consistent equatorial low pressures determines the presence of trade
emergent pattern or trend in daily rainfall. These authors, winds blowing from northeast to southwest. These winds,
however, pointed out that the time span selected may have a relatively cool but rather dry, control the dry period (bega in
relevant influence on the results of trend analysis. Cheung the local language). In spring, the influence of southwestern
et al. (2008) arrived at similar conclusions, verifying that no winds, coming from the Congo basin, determines the season of
significant changes or trend in annual rainfall is evident at ‘little rains’ (belg in the local language) that can bring rela-
the national or watershed level in Ethiopia. Finally, Jury and tively abundant precipitation in the southern part of the
Funk (2013), using gridded data from different sources, country. In summer, the Guinean monsoon, consisting of
3 The Climate of Ethiopia 69

equatorial warm and humid winds, results in bountiful rains intervals and of the meteo-stations to be considered in this
(kiremt in the local language) which are also substantially study.
influenced by the orographic diversity mentioned above. The notable physiographic variability of the country in
Such a complex meteorological framework is reflected by many cases did not allow to fill the gaps by means of
the distribution of annual precipitation. In the Danakil ordinary correlation analysis with other stations in the
depression, it is constantly less than 250 mm but can be as vicinity, resulting in insignificant correlation coefficients
low as 50 mm. By contrast, on the highlands, 2,000 mm can (Hadgu et al. 2013). Such constraint is particularly relevant
be locally exceeded. Similar values of annual precipitation, for time series and trend analysis, especially for rainfall.
however, are recorded also in the southwestern lowlands, Notwithstanding such limitations and nonuniformity, some
likely due to a larger contribution of the spring rains. indications on long-term trends can be given anyway.
Altitude patently determines marked annual thermal Jury and Funk (2013) report about spatial variability of
gradients with sharp transitions from zones with a desert rainfall on the basis of gridded data and interpolation. Given
climate—among the hottest of the planet—to tropical, high the low density of meteorological stations and the rugged
mountain climates with minimum temperatures below zero topography of Ethiopia, this procedure does not seem to
in each month and modest annual thermal amplitude. provide a realistic pattern of rainfall distribution. Addition-
ally, the data set used by these authors includes a five sta-
tions correlation to fill data gaps, and a correlation as low as
3.4 Data Sources 0.2 has been accepted as significant. In the worst case, the
mean values of the time series are included. This procedure
As briefly introduced in the previous section, the high vari- seems, at least, questionable; in fact, it does not avert the
ability of physiography and orography of Ethiopia is reflected data limitations and introduces arbitrary conditions that are
in the complex spatial distribution of precipitation and tem- not necessarily making the data analysis stronger or more
perature as briefly introduced in the previous section. Having reliable than the use of raw data.
in mind such a complexity, large sets of data and long time Spatial distribution of rainfall has been investigated also
series for the main climatic parameters are required to produce by Viste et al. (2013) by using the raw data of the NMA, but
an updated description of the climate of Ethiopia and, possi- their study focused mainly on monthly precipitation. Other
bly, of its variations through time. For this purpose, a large authors such as Seleshi and Cumberlin (2006) investigated
number of monthly temperature and, whenever possible, daily the spatial and seasonal (bega, belg and kiremt) rainfall
rain gauge data were collected from different sources extremes, but their paper is focused on 5 days consecutive
including the Ethiopian National Meteorological Agency rains rather than daily rainfall.
(NMA), which is the main provider of field data measure- Notwithstanding the awareness of data limitations, the
ments, and other international organizations such as the approach of this paper attempts to complement the analyses
Global Historical Climatology Network (GHCN) of the of seasonal and spatial variability reported in previous
National Oceanic and Atmospheric Administration (NOAA) papers (e.g., Segele and Lamb 2005; Seleshi and Cumberlin
in the USA and the United Nations Food and Agricultural 2006; Korecha and Barnston 2007; Viste et al. 2013).
Organization (FAO), the data sets of which are substantially As an alternative to the manipulation of data beyond any
based on NMA data. The NMA produces daily rainfall data reasonable confidence and notwithstanding the existing gaps
for all the many stations that are presently operative; in 22 of and limitation and the need for the time series as longest as
them also hourly precipitation is measured, whereas FAO possible, in order to attain a certain degree of coherence with
reports rainfall data at decadal (10 days) intervals. the actual physical phenomena, the following procedure was
The temperature data of the NMA, and those of the used. Statistical tests were carried out in order to ascertain
GHCN as well, are reported as daily data for the whole the quality of data and the uniformity of the temperature time
operational time interval of first level meteo-stations and as series. The standard normal homogeneity test (Alexanders-
monthly data for lower level gauges, whereas FAO provides son 1986) and the Kolmogorov–Smirnov test were used to
only decadal data. determine the homogeneity of records, whereas the Mann-
Kendall test was used to verify the statistical significance of
trends at a 0.05 level of significance.
3.5 Data Validation and Processing Both the quality tests indicate that rainfall time series are
generally homogeneous, whereas only four out of 18 tem-
Unfortunately, the above data sets are commonly affected by perature time series proved inhomogeneous. Furthermore,
long and reiterated gaps and may include entire years of the variability of annual rainfall and temperatures over the
measurements missing. This posed several problems in the study period was analyzed by calculating the coefficient of
validation of data and in the selections of uniform time series variability (CV).
70 M. Fazzini et al.

In order to provide a general picture of temperatures and Ethiopian borders and thus to improve as much as possible
precipitation variability across the whole territory of Ethio- the reliability and quality of the analyses (Fig. 3.1).
pia in the last decades and in consideration of the above- The elevation of the stations considered varies between
mentioned quality limitation of the available time series, a 10 m a.s.l. at Massawa and 2,760 m a.s.l. at Fitche. Though
simple procedure, based on the normalization of data, was marked differences are observed for the average annual and
used. For each station, the annual precipitation and the mean monthly values in the highlands—typical of equatorial high
minimum and maximum temperature of each year have been mountains—and in the northeastern lowlands, Tigray and
normalized and the Zi score calculated as follows: the Omo valley, the climate of Ethiopia is generally very
pleasant and among the most equilibrate of the African
Zi ¼ ðDi  Dm Þ=sd continent since the temperatures are mitigated by the high
average elevation. Only in the Danakil Desert areas the
in which Di is the parameter annual value for the ith year, Dm temperatures reach extreme values.
is the mean value of the time series, and sd its standard The lowest mean annual temperatures are recorded at ele-
deviation. vations over 2,300–2,600 m a.s.l. (Fig. 3.1 and Table 3.1),
For each year of the time span considered, a representa- irrespective of their geographic position. At Fitche, the highest
tive Z score has been calculated by averaging the Zi scores of and coldest station, the mean annual temperature is 14.2 °C,
all the stations. In this way, a new time series of Z scores is whereas at the elevation of 2,000 m it is around 18.5 °C.
obtained. By this procedure, though not orthodox, it is In the Rift Valley, the temperatures gradually increase
possible to minimize the relative effects of gaps that are from south to north. Around the elevation of 1,500 m, they
present and almost randomly distributed within the whole are slightly over 20 °C and the 25° isotherm stands around
data set and to depict the time distribution of higher and 1,000 m a.s.l. At lower elevations, the temperatures notice-
lower than average values. Any attempt to fill the gaps ably increase and peak to 30 °C in the steppe area of Gore
through correlations with neighboring stations as suggested and to higher values in the Danakil Desert.
by Jury and Funk (2013) resulted to be unsatisfactory. On The mean maximum temperatures are particularly pleas-
the other hand, the use of normalized data and averaged ant at elevations above 1,500 m. At Fitche, they are 20,
Z scores has been considered appropriate to reduce the effect 23.5 °C at Addis Ababa and about 25 °C around the 2,000 m
of local outliers and to delineate trends that depict a general elevation. At 1,000 m, the mean values are close to 30 °C but
tendency for the entire country. This issue, however, would can be higher than 35 °C in the southwestern lowlands and,
need further detailed and more sound statistical analyses that mainly, in the steppe areas and the Danakil Desert (Fig. 3.2).
are beyond the scope of this paper. The mean minimum temperatures (Fig. 3.3) are rather
For the main climatic parameters, the data elaboration is low all across the highlands, where a few days of frost are
portrayed in colors overlying a shaded relief map. The commonly recorded every year. At Robe Bale, the station
regionalization of the data has been obtained through geo- with the lowest minimum temperatures, the mean annual
statistical interpolations considering the local elevation, lat- value is 8 °C, whereas at Addis Ababa it is 10 °C. Higher,
itude, distance from the sea and the largest lakes, and slope typically tropical values of 20–25 °C are recorded at ele-
aspect for the whole area. For temperatures, the theoretical vations below 500 m.
monthly solar radiation was considered as well. All the maps The lowest temperature ever recorded in the meteo-sta-
were interpolated and drawn using advanced GIS tools tions considered is −7 °C on the Bale Mountains (highest
(ESRI ArcGIS–ArcInfo). peak 4,377 m a.s.l.), whereas in Addis Ababa frost condi-
tions only sporadically occur.
In Ethiopia, one of the hottest areas of the planet is found:
3.6 Temperatures the Danakil depression (Vinassa de Regny 1931; Fantoli
1940). Unfortunately, recent data are not available to con-
To analyze the spatial distribution of temperature, the firm this evidence. At Dallol (130 m b.s.l.), which is con-
monthly data from 30 meteo-stations, with continuous series sidered the hottest, inhabited place of the world, Pedgley
or with negligible gaps throughout the World Meteorologi- (1967) measured a mean annual temperature of 34.7 °C,
cal Organisation (WMO) 1981–2010 reference interval, mean maximum temperatures of 41.2 °C, and a peak of
were selected (Table 3.1). Additional 17 meteo-stations, 45.7 °C for the 1960–1966 interval. According to Billi
located in the adjoining territories of Djibouti, Eritrea, (personal communication), in November 2007, temperatures
Somalia, and Sudan, were considered as well (Table 3.1) in of 42–44 °C were commonly recorded in the Samoti plain in
order to account also for the conditions right beyond the the Eritrean Danakil.
3 The Climate of Ethiopia 71

Table 3.1 Main temperature data

Meteo-stations H Tmean Tmax Tmin Ea Ed Mh Mc
(m a.s.l.) (°C) (°C) (°C) (°C) (°C)
Addis Ababa 2,354 17.0 23.5 10.4 3.1 13.1 May Dec
Al Damazin 470 27.5 35.1 19.8 7.2 15.3 Apr Sep
Alamata 1,520 22.5 30.0 14.9 6.3 15.1 Jun Jan
Alemaya 2,047 16.5 23.7 9.3 5.7 14.4 Jun Dec
Arba Minch 1,285 23.8 30.2 17.3 2.9 12.9 Mar Jul
Asab 14 30.4 35.4 25.4 9.5 10.0 Aug Feb
Asela 2,430 15.4 21.9 8.9 2.7 13.0 May Dec
Asmara 2,325 15.2 22.9 7.4 5.0 15.5 Jun Dec
Awasa 1,750 20.0 27.3 12.6 1.8 14.7 Apr Nov
Bahir Dar 1,770 19.1 26.9 11.2 4.6 15.7 May Jan
Belet Uen 198 28.6 34.8 22.4 2.6 12.4 Apr Aug
Berbera 1 30.5 35.2 25.8 13.1 9.4 Jul Jan
Borama 1,454 21.2 27.3 15.0 4.9 12.3 Mar Jul
Burao 970 22.8 29.3 16.3 6.2 13.0 May Jan
Debre Marcos 2,515 16.4 22.5 10.3 3.7 12.2 Apr Aug
Dire Dawa 1,260 25.5 31.8 19.1 7.0 12.7 Jun Jan
Djibouti 21 29.8 33.8 25.7 11.3 8.1 Jul Jan
Fitche 2,750 14.2 20.3 8.1 3.1 12.2 May Nov
Galcayo 247 30.4 34.1 26.7 9.8 7.4 Jul Feb
Gambela 484 27.6 34.8 20.4 5.3 14.4 Mar Jun
Garowe 473 30.3 33.9 26.7 9.9 7.2 Jul Dec
Gewane 627 21.5 24.9 18.1 8.0 13.8 Jun Dec
Gode 295 28.9 34.7 23.1 2.8 11.6 Mar Nov
Gonder 1,967 20.0 26.7 13.3 4.6 13.4 Apr Aug
Gore 2,002 18.9 23.9 13.9 3.7 10.0 Mar Jul
Hargeysa 1,347 22.3 27.4 17.1 5.0 10.3 May Jan
Hayk 2,030 18.3 25.8 10.7 5.8 15.1 Jun Dec
Jijiga 1,644 19.4 26.5 12.3 4.5 14.2 Aug Jan
Jimma 1,725 19.7 27.6 11.8 3.0 15.8 Apr Dec
Kassala 502 29.5 37.1 21.8 8.3 15.3 May Jan
Kebry Dehar 549 26.6 33.1 20.1 2.2 13.0 Apr Aug
Kobo 1,610 22.1 29.3 14.9 7.0 14.4 Jun Dec
Kombolcha 1,903 19.3 26.5 12.1 5.9 14.4 Jun Dec
Mandera 231 29.5 34.9 24.1 3.7 10.8 Mar Jul
Massawa 10 29.4 34.5 24.3 10.1 10.2 Jul Jan
Mekele 2,070 17.6 23.8 11.3 4.6 12.5 Jun Dec
Metahara 947 25.4 33.3 17.4 6.3 15.9 Jun Dec
Mogadishu 7 26.7 30.2 23.1 3.1 7.1 Apr Aug
Moyale 1,097 22.4 27.3 17.5 5.2 9.8 Feb Jul
Negelle 1,544 20.7 26.3 15.0 4.7 11.3 Mar Jul
Nekemte 2,080 18.3 24.0 12.6 4.1 11.4 Mar Jul
Robe Bale 2,480 14.8 21.6 8.0 2.2 13.6 Jun Dec
Sirinka 2,081 18.8 24.1 13.5 6.1 10.6 May Dec
Ziway 1,640 20.6 27.1 14 3.3 13.1 Apr Dec
H elevation, Tmean mean annual temperature, Tmax mean annual maximum temperature, Tmin mean annual minimum temperature, Ea mean annual
temperature excursion, Ed mean daily temperature excursion, Mh hottest month, and Mc coldest month
72 M. Fazzini et al.

Fig. 3.2 Mean maximum annual

temperatures

Fig. 3.3 Mean minimum annual

temperatures
3 The Climate of Ethiopia 73

40
y = -0.0051x + 35.344
35 R² = 0.83
30 y = -0.0058x + 30.102
R² = 0.92
°C 25

20 Tmax
15 Tmean

10
Tmin
y = -0.0065x + 24.861
5 R² = 0.92
0
0 500 1000 1500 2000 2500 3000
H (m a.s.l.)

Fig. 3.4 Lapse rate for maximum mean and minimum mean annual temperatures

From the above considerations and Fig. 3.4, it is evident

that elevation is a main parameter influencing temperature in 3.7 Precipitation
Ethiopia and adjoining areas. The lapse rate is 5.8 °C every
1,000 m, and the mean annual 10 and 5 °C isotherms were To describe the rainfall characteristics of Ethiopia, the
estimated to be around 3,450 and 4,300 m a.s.l., respec- monthly data of 49 meteo-stations in Ethiopia and 9 in the
tively. The maximum and minimum temperature lapse rates adjoining territories of Eritrea, Djibouti, Somalia, and
are slightly different with 5.1 °C/1,000 m and 6.5 °C/ Sudan, and the daily data of 27 Ethiopian stations were used
1,000 m, respectively. (Table 3.2). The longest time series is that of Moyale cov-
Considerable differences characterize the diurnal and ering 73 years of almost uninterrupted daily data from 1935
annual temperature ranges (Fig. 3.5). The former is rather to 2008, whereas the average length of the Ethiopian time
large, especially in the central-western steppe and desert area series considered is 33 years (ranging between 20 and
(Fig. 3.6), varying between 10 and 15 °C; the latter is much 68 years), with only four stations covering a shorter interval
narrower, commonly less than 5 °C in the highlands areas (10–20 years). These latter were included as well in order to
below 2,000 m in elevation and in the tropical areas attain the larger and denser distribution of data as possible
(Fig. 3.6). In fact, sunny days prevail during most of the year for the spatial interpolations used in producing the annual
in almost the whole country, including the areas with a rainfall distribution map (Fig. 3.7).
monsoon climate; the alternation of cloudy and clear sky The annual precipitation, averaged all across the country,
conditions results in ample thermal differences (Fig. 3.5 and is 817 mm but, given the complex physiography and the
Table 3.1). By contrast, the annual thermal amplitude is different seasonal and spatial influences of the prevailing air
limited and seems to be controlled by the intertropical cir- masses and winds, a large diversity is observed among various
culation and the monsoon pattern rather than elevation. In regions. The highest annual rainfall is recorded at Gore, in the
the central-western highlands and in the Rift Valley, the western highlands at an altitude of 2,002 m a.s.l., with
hottest month is in spring and anticipates the summer, big 2,101 mm, whereas the lowest value in Ethiopia is found at
monsoon rains (kiremt), whereas in the northern and eastern Elidar with 153 mm. This rain gauge is located in the Afar
quadrants it occurs at the beginning of summer. triangle at an elevation of 423 m s.l.m., but the driest site is
In the highlands, the coldest month is in the middle of Asab, in Eritrea, with only 45 mm. Gore and Elidar are not the
winter, whereas in the southern and eastern regions, most and the least elevated stations, respectively, considered
including also high elevated areas, subjected to perturbed in this study; nevertheless, they share the extreme values of
conditions brought about by the monsoon and characterized annual rainfall. This is reflected by the poor correlation
by heavy downpours and cloudy skies, the coldest month (R2 = 0.42) between precipitation and elevation. Figure 3.7
occurs in summer. shows that higher precipitation occurs in the western part of
74 M. Fazzini et al.

Fig. 3.5 Annual temperature range

Ethiopia and is partly irrespective of the station altitude. In precipitation (Fig. 3.11) may be higher than the kiremt,
Fig. 3.7 also, a westward gradient is rather evident too. Fol- contributing about 50 % of the annual rainfall such as at a
lowing these considerations, a physiographic factor (F) was few stations in the south and in Ogaden. Figure 3.9 shows
developed. It is the ratio of the meteo-station elevation (H) to also that the belg rains have little influence on the annual
L, i.e., the difference between its longitude and the 34th precipitation as they explain only 31 % of its variability.
meridian, taken as a western reference. Figure 3.8 shows that The spring rains are also the most unpredictable for their
85 % of annual precipitation variability is explained by the variation coefficient is 47 %, whereas the summer rains
physiographic parameter F = H/L. (CV = 33 %) result the most dependable for the farmers. Only
Since in two-thirds of the meteo-stations, the summer on the coast, namely at Djibouti and Massawa, the winter
rains (kiremt) contribute more than 50 % of the annual rains account for most of the annual precipitation. Along the
precipitation, with peaks of 80–85 % in the north, and 87 % Eritrean coast and in Ogaden, the largest variability of annual
of the annual precipitation variability is explained by the precipitation is found (Aseb CV = 127 %), whereas the least
kiremt (Fig. 3.9); the monsoon and its seasonal and geo- variability is observed in the central highlands (Table 3.2).
graphic pattern appear to be the most relevant factors in This observation is confirmed by the diagram of Fig. 3.12 in
controlling the spatial distribution and amount of rainfall in which the variation coefficient is well explained (R2 = 0.75)
Ethiopia (Fig. 3.10), though in some places the belg by the physiographic parameter F.
3 The Climate of Ethiopia 75

(a) Dire Dawa (Fig. 3.13). Hourly data for Mekele and Ziway show that
35 maximum intensities around 20 mm h−1, with peaks of
30 40–50 mm h−1, are rather common in these two localities.
Soil erosion is a major environmental and economic
25
problem in Ethiopia. In order to highlight the regions
°C

20 Tmin affected by severe hydro-meteoric erosion conditions, the

15
Tmax Modified Fournier Index (MFI) (Arnoldus 1980) was cal-
culated (Table 3.2) and the spatial distribution of this
10
parameter is mapped (Fig. 3.14). At two-thirds of the
5 weather stations considered, rainfall erosivity is from mod-
J F M A M J J A S O N D
erate to high and at 43 % of them it is high. The higher
(b) 35 Bahir Dar values are found in the central and part of the northern
highlands and, subordinately, in the central portion of the
30 Somali plateau margin. These areas are classified as sub-
25 jected to high and very high runoff by Berhanu et al. (2013)
20 and high erosion rates would be expected, but the occurrence
°C

Tmin
15 of soils with low erodibility, as reported by the USLE’s
Tmax
10 factor K map reported by these authors, partly mitigates this
threat.
5
0
J F M A M J J A S O N D
3.8 Wind
Fig. 3.6 Mean maximum and minimum monthly temperatures at Dire
Dawa and Bahir Dar In Ethiopia, the wind regime is influenced by barometric
synoptical variations associated with the position of the
ITCZ. Winds are not strong because the Horn of Africa is
Rainfall intensity was investigated using daily data not a cyclone-genetic area and, as pointed out in Sect. 3.6,
recorded at 26 stations. The daily interval was selected temperatures are rather uniform throughout the whole year.
because, unfortunately, long time series of hourly data are not As a consequence, winds, though seldom absent, have low
available. Certainly, this is a strong constraint to sort out the velocities, and only during deep convective processes, short
regions with aggressive rains since commonly, also during but intense gusts and whirls are observed. Windiness is
the monsoon season, rainstorms are short (1–3 h) and a daily characterized by a large spatial variability due to the phys-
datum may be not sufficiently accurate to describe the rainfall iographic complexity which affects the prevailing seasonal
intensity parameter, especially in relation to the activation of air masses, especially along the Rift Valley and its margins.
soil erosion processes. Nevertheless, also using daily data, From the analysis of daily and, occasionally, 6 h data
some general results can be drawn. The mean maximum from a few meteo-stations located in the highlands, in Ti-
rainfall intensity (Imax) at each meteo-station was calculated gray, along the Rift Valley, and in Ogaden for the 2000–
by averaging the maximum values recorded in each year, 2006 interval (Fig. 3.15), it is evident that in the central-
whereas Ip is the highest value ever recorded by each rain northern part of the country, in a few sites of the Eritrean
gauge. Gambela is subjected to the highest rainfall intensities coast and the Sudan lowlands, winds from the northern
with an average of 89.4 mm in 24 h and a maximum value quadrant dominate, whereas in the southern areas, winds
ever recorded of 181.4 mm/24 h. By contrast, the lowest from the second and the third quadrants prevail throughout
values are observed in Elidar (Imax = 27.8 mm/24 h) and the whole year. On the margin of the Somali Plateau, wes-
Gewane (Ip = 31.6 mm/24 h), i.e., two rain gauges located in tern winds prevail, whereas in the far southwest of Ethiopia
the Afar depression characterized by desert and semidesert southwestern winds are more common. On the highlands,
climate conditions. Neither physiographic factor, such as the annual frequency of days with calms and light breezes
elevation or the geographic position, nor any pluviometric amounts to 12–18 % of the total.
parameter, such as annual rainfall, is able to explain the At seasonal level and particularly during the kiremt,
variability of rainfall intensity, which is likely more influ- noticeable differences from the above framework can be
enced by local orographic or morphologic factors. traced:
High intensity rains may occur in any month, but the 1. In winter, the situation described above is confirmed with
highest frequencies are recorded in April and August, i.e., in winds from the first quadrant and from northwest pre-
the middle of the spring and summer rainy seasons vailing on the central highlands and on the Somali
76 M. Fazzini et al.

Table 3.2 Characteristic data of the rain gauges considered in this study
Meteo-stations H P (mm) CV (%) P (mm) P (mm) P (mm) k/a Imax Ip Tr 100 mm/ MFI
(m a.s.l.) annual annual belg kiremt bega % mm/24 h mm/24 h 24 h (years)
Adami Tulu 1,636 756.8 27 186.2 447.6 123.0 59.1 47.4 81.9 68.6 93.9
Addis Ababa 2,354 1,204.8 18 238.4 863.5 102.9 71.7 186.1
Adigudum 2,095 490.5 49 52.5 428.5 9.5 87.4 127.1
Al Damazin 470 691.0 56.0 596.0 39.0 86.3 137.3
Alamata 1,520 750.4 24 219.9 387.1 143.4 51.6 52.8 96.0 32.4 106.2
Alemaya 2,047 802.9 26 248.4 431.0 123.5 53.7 56.4 118.0 15.3 95.0
Arba Minch 1,285 931.9 17 375.9 236.6 319.4 25.4 99.7
Aseb 14 45.4 127 6.1 20.6 18.7 45.4 7.5
Asela 2,430 1,225.5 17 321.6 762.8 141.1 62.2 150.2
Asmara 2,325 512.3 34 87.8 375.7 48.8 73.3 109.7
Awasa 1,750 951.5 16 292.9 472.6 186.0 49.7 99.3
Azezo 1,966 1,153.0 146.0 914.0 93.0 79.3 220.7
Bahir Dar 1,770 1,413.2 17 105.0 1,189.8 118.4 84.2 293.7
Belet Uen 198 259.0 138.0 21.0 100.0 8.1 53.7
Bilate 1,500 744.2 23 240.3 326.9 176.9 43.9 78.9
Borciota 1,062 505.0 122.0 309.0 74.0 61.2 76.4
Burao 970 195.0 102.0 62.0 31.0 31.8 33.0
Butajira 2,000 1,118.1 19 377.7 560.6 179.8 50.1 121.1
Comar 764 920.0 90 785 45.0 85.3 180.5
Debre Marcos 2,515 1,325.5 11 208.9 970.1 146.5 73.2 204.4
Dengeco 2,111 750.5 22 249.6 429.0 71.9 57.2 53.8 98.8 21.0 103.6
Derbiga el Agemsa 1,644 678.0 239.0 340.0 99.0 50.1 80.4
Desse 2,250 1,207.1 16 244.6 789.5 173.0 65.4 54.9 94.0 47.0 189.4
Dire Dawa 1,260 638.7 27 220.2 313.6 104.9 49.1 54.8 122.3 18.1 78.3
Djibouti 21 188.0 71 41.59 17.52 128.9 9.3 24.8
Elidar 423 153.1 40 29.0 99.1 25.0 64.7 27.8 55.0 26.5
Fitche 2,750 1,125.2 12 183.0 826.0 116.2 73.4 51.4 90.9 42.8 226.2
Gambela 484 1,114.1 22 187.3 697.6 229.2 62.6 89.4 181.4 139.7
Gedaref 599 579.7 16 15.8 525.2 38.7 90.6 146.6
Gewane 627 521.9 21 135.3 268.7 117.9 51.5 29.1 31.6 71.8
Gina Ager 3,160 1,693.2 20 415.3 977.0 300.9 57.7 74.3 170.8 6.4 216.9
Goba 2,743 943.7 20 292.4 412.2 239.1 43.7 98.9
Gode 295 236.9 61 135.0 8.4 93.5 3.6 58.4 174.0 7.0 50.7
Gonder 1,967 1,131.9 18 139.4 878.4 114.1 77.6 52.4 99.1 40.9 206.7
Gore 2,002 2,101.3 23 439.7 1,261.9 399.7 60.1 62.1 107.7 24.9 247.4
Hargeysa 1,347 350.0 120.0 210.0 20.0 60.0 50.6
Hayk 2,030 1,174.0 17 280.9 717.1 176.0 61.1 62.1 132.8 16.5 176.4
Jijiga 1,644 721.3 39 249.9 329.2 142.3 45.6 75.6
Jimma 1,725 1,500.5 13 401.1 821.0 278.4 54.7 53.0 105.7 39.5 164.4
Kassala 502 278.6 43 13.0 229.0 36.6 82.2 49.0 104.9 56.2
Kebri Dehar 549 325.3 49 170.4 20.6 134.3 6.3 59.4 128.0 11.8 77.7
Kobo 1,610 725.8 22 151.5 411.8 162.5 56.7 53.2 101.5 18.3 97.7
Koka Dam 1,595 698.8 48 135.6 496.0 67.3 71.0 112.3
Kombolcha 1,903 1,043.6 15 221.3 663.1 159.2 63.5 160.9
Konso 1,053 805.6 14 358.2 153.0 294.4 19.0 55.2 96.9 28.4 86.2
(continued)
3 The Climate of Ethiopia 77

Table 3.2 (continued)

Meteo-stations H P (mm) CV (%) P (mm) P (mm) P (mm) k/a Imax Ip Tr 100 mm/ MFI
(m a.s.l.) annual annual belg kiremt bega % mm/24 h mm/24 h 24 h (years)
Kulubi 2,410 1,022.3 18 308.7 565.6 148.0 55.3 60.2 100.5 14.4 125.6
Mandera 231 286.7 56 159.4 4.5 122.8 1.6 59.6
Massawa 10 175.8 58 22.5 19.6 133.7 11.1 24.0
Mekele 2,070 598.7 27 91.3 487.7 19.7 81.5 49.4 95.5 59.6 162.6
Meki 1,660 771.2 21 185.6 482.1 103.5 62.5 103.6
Metahara 947 549.2 23 140.2 322.4 86.6 58.7 52.3 96.0 31.5 78.0
Moyale 1,097 685.0 40 352.5 63.4 269.1 9.3 68.7 167.6 6.3 104.1
Negelle 1,544 726.6 27 414.5 52.7 259.4 7.3 66.1 137.0 8.2 128.6
Nekemte 2,080 2,037.7 12 369.3 1,429.0 239.4 70.1 75.9 137.5 7.0 294.9
Robe Bale 2,480 876.3 16 281.0 380.1 215.2 43.4 43.3 112.3 8.7 92.1
Sirinka 2,081 1,128.9 14 206.1 361.3 561.5 32.0 161.8
Ziway 1,640 742.5 24 201.5 442.1 98.9 59.5 89.3
Zuquala 2,980 1,124.8 22 251.8 763.3 109.7 67.9 68.0 132.6 8.7 178.3
H elevation, P precipitation, CV variation coefficient, k/a contribution of the kiremt rainfall to the annual precipitation, Imax mean maximum rainfall intensity in 24 h, Ip
highest rainfall intensity in 24 h ever recorded, Tr return time for a rainfall intensity of 100 mm in 24 h, and MFI Modified Fournier Index Arnoldus 1980)

Fig. 3.7 Distribution of annual precipitation

78 M. Fazzini et al.

2500 plateau margin, respectively. Elsewhere, southeastern

and southwestern winds prevail.
2000
2. In spring, only small changes occur. In fact, the western
P (mmyr -1 )

1500 winds rise in latitude along the Sudanese border as far as

Tana Lake, whereas in the southernmost areas they are
1000 y = 37.642x 0.5458 replaced by winds from the second quadrant. On the
R² = 0.85 Somali plateau margin and in Ogaden northeastern winds
500
predominate.
0
3. During summer, wind circulation turns eastward on the
0 200 400 600 800 1000 1200 1400 central-northern highlands, whereas such orientation is
F
not so evident in the area of Addis Ababa and in the Rift,
Fig. 3.8 Correlation between annual precipitation (P) and the and finally to northwest in Eritrea (e.g., Asmara).
physiography factor (F—see text for explanation) Southwestern winds predominate on the Sudanese low-
lands and as far as Gambela where winds from east and
southeast prevail. These winds are present also in the
2500
y = 3.7944x + 272.61 y = 2.5191x + 346.28 southern part of the country as far as the Somali plateau
R² = 0.84 R² = 0.31 margin (e.g., Harar).
2000
4. In autumn, the part of Ethiopia to the west of an ideal
Annual P (mm)

1500
Gonder-Nekemte line and the southern sectors are sub-
Kiremt
jected to western air masses, whereas in other parts of the
Bega
1000 country eastern air masses prevail with the exception of
Belg
the Harar region where northwestern winds are more
y = 1.3332x + 248.38
500
R² = 0.87
common.
On the highlands, mean wind speed is typically low
(12 km h−1) but increases in the Great Lakes Region, in
0
0 200 400 600 800 1000 1200 1400
Season P (mm) the sub-desert and bush lands of Ogaden and around the
border with Sudan, especially during the kiremt (EMA
Fig. 3.9 Correlation between annual precipitation (P) and the
cumulative precipitation of kiremt and belg main rainy seasons and
1988).
bega, the dry spell

Fig. 3.10 Distribution of the

summer, monsoon-type rains
(kiremt in the local language)
3 The Climate of Ethiopia 79

Fig. 3.11 Distribution of the little, spring rains (belg in the local language)

0.7 25
0.6
20
0.5
Frequency (%)

y =-0.128ln(x) + 0.9833
0.4
PCV

R² = 0.75 15
0.3 Imax
0.2 10 Ip
0.1
5
0
0 200 400 600 800 1000 1200 1400
F 0
J F M A M J J A S O N D
Fig. 3.12 Correlation between the annual rainfall variation coefficient
(PCV) and the physiographic factor (F—see text fro explanation) Fig. 3.13 Monthly distribution of the frequency of occurrence of the
annual maximum rainfall intensity in 24 h (Imax) and of the highest
value (Ip) recorded throughout the period of recording for all the meteo-
stations considered in this study
80 M. Fazzini et al.

Fig. 3.14 Spatial distribution of

rainfall erosivity expressed by the
MFI (Arnoldus 1980)

The most part of Ethiopia—and in particular the north—

45
is subjected to the influence of trade winds commonly rich in
40
water vapor. Obviously, the spatial distribution of humidity
35
is tightly correlated with precipitation. In the southwest of
30 AWASA
Ethiopia, areas with very high humidity, among the most
kmh -1

25 GORE
humid in the African continent (e.g., the great equatorial
20 GODE
lakes region), are found.
15 ADDIS ABABA
By contrast, during winter, high values are recorded in
10 GONDER
Ogaden and along the Red Sea, whereas low humidity
5
characterizes the Sudanese border belt with mean values
0 around 23 %.
J F M A M J J A S O N D
The data recorded in 13 meteo-stations during the last
Fig. 3.15 Monthly variation of average wind speed for a few meteo- decade indicate that average annual relative humidity is
stations about 60 % on the highlands, in the Rift and along the Kenya
border (Fig. 3.16), but even higher values are presumed for
very high mountain areas (EMA 1988). Lower values
between 40 and 50 % are instead recorded in the northern-
3.9 Relative Humidity most portion of the Rift (e.g., Dire Dawa—Fig. 3.16).

Only inadequate data of relative humidity are available since

few stations have been measuring this parameter in the last 3.10 Cloudiness and Sunshine
decades. Notwithstanding this constraint, combining the
available data with the maps of the National Atlas (1988), Ethiopia is the country with the most extensive cloudiness in
some considerations on the spatial distribution of this Africa. Unfortunately, only very few data of Fantoli (1940)
parameter can be drawn. are available and refer to the first decades of the twentieth
3 The Climate of Ethiopia 81

90 3.11 Evapotranspiration and Aridity

80
70 Evapotranspiration is limited by soil moisture supply, sur-
60 face and atmospheric temperature and the dew point
ADDIS ABABA
threshold (Law et al. 2002). The coupling of evapotranspi-
Hr (%)

50
DIRE DAWA
40 ration, temperature, and precipitation is particularly pro-
BAHIR DAR
30 nounced in moisture-limited subtropical regions at the
DAMAZINE
20 interface between wet (monsoonal) and dry climate regimes.
10
Potential evapotranspiration (PET) was calculated fol-
0
lowing the Thorntwaite’s method, using the data from 29
J F M A M J J A S O N D meteo-stations for the 1981–2010 time range (Fig. 3.18). The
same data were used also to reckon the aridity index (Ai)
Fig. 3.16 Monthly distribution of mean relative humidity for a few
representative meteo-stations
(Fig. 3.19), known also as the desertification index (UNEP
1992), as Ai = P/PET in which P is annual precipitation.
Almost all the areas over 2,000 m a.s.l. are characterized
12 by annual PET values less than 1,000 mm, whereas pre-
ADDIS ABABA
cipitation is typically around 1,500 mm or higher. Therefore,
10 BAHIR DAR
the highlands can be classified as humid and hyper-humid in
AWASA
8 the Gore–Jimma–Tepi triangle (Fig. 3.18). As elevation
S h (hours)

MEKELE
decreases, PET commonly increases as it is observed in the
6 KEBRY DEHAR
Rift Valley. Extreme evapotranspiration conditions are
GORE
4 found in the Danakil depression where values of as much as
NEGHELLE
4,000 mm can be estimated for Dallol. This figure, combined
2 MOYALE
with inappreciable precipitation and extremely hot temper-
DAMAZINE atures, makes this area as one of the most arid on the planet.
0
ASMARA Ai values less than 0.65 are typical of drylands (Thomas
J F M A M J J A S O N D
and Middleton 1994) and UNEP (1992) takes 0.65 as the
Fig. 3.17 Monthly distribution of mean sunshine hours for a few threshold value to identify desertification prone areas. Large
representative meteo-stations
portions of Ethiopia fall in this category: the whole Ogaden,
the Afar triangle, the Danakil depression, the northern por-
century. According to this author, average cloudiness is 2/10 tion of the Rift Valley, and the belt across the border with
in January and 6/10 in July, whereas clear sky days are a Sudan (Fig. 3.19).
little less frequent than mixed and cloudy ones (170 vs 195).
The maximum and minimum cloudiness occurs during the
rainy season and the boreal winter, respectively. In Ogaden 3.12 Climate Classification
and mainly in the Danakil depression, cloudiness is almost
an exceptional phenomenon. Along the Eritrean coast and As highlighted in Sect. 3.3, the climate of Ethiopia is mainly
for a few kilometers inland, night advection fogs may controlled by the seasonal migration of the ITCZ and by the
commonly form and turn into morning drizzle, integrating complex topography of the country. This combination
the scarce precipitation of this area. results in a variety of climate types and climate conditions
Sunshine is obviously complementary to cloudiness. that may vary also within a short distance (Fig. 3.20).
Monthly data of sunshine hours (Sh) measured by nine sta- A tight relation is evident between elevation and climate
tions in the last decade (Fig. 3.17) show that most of the type, especially along the plateau margins. East of this
highlands and the lowlands close to the Sudan border receive margin and in the southeastern portion of the country, a
more than 7 h of mean daily sunshine, whereas from Tigray predominantly arid and very hot climate (tropical A, Köppen
to the southeast of the country the respective value increases classification) fades, throughout a belt of climatic transition
to 10 h. During autumn and winter, 300–350 h per month are at higher elevations, into a temperate C climate, character-
commonly recorded in Tigray and Ogaden. In summer, vice ized by mild temperatures, higher than 18 °C in every
versa, in the highlands between Gonder and Addis Ababa, month, and by rainfalls with a maximum in summer (Tigray
the sunshine commonly does not exceed 50 h per month. and Somali plateau margin), in spring and a secondary
82 M. Fazzini et al.

Fig. 3.18 Spatial distribution of

potential evapotranspiration

Fig. 3.19 Spatial distribution of

the aridity index (UNEP 1992)
3 The Climate of Ethiopia 83

Fig. 3.20 Map of the Köppen

climate classification

maximum in autumn (southern portion of the country) rather Over 2,600 m a.s.l. and in a few areas of the Semen and
than constant precipitation throughout the four seasons (type Bale Mountains, the climate becomes relatively cold with
AM of the southwestern areas). The thermal transition from mean monthly temperatures less than 10 °C. At elevations
Af, Aw, or BSh to the temperate C climate occurs at an higher than 3,300 m a.s.l., the climate is characterized by very
elevation ranging between 1,800 and 2,000 m a.s.l., though low temperatures and it is classified as type H which includes
in the northernmost part of the highlands the climate the peaks of the intertropical massifs. Here, vegetation con-
becomes cooler and remains arid also at high elevations (e.g. sists mainly of alpine prairie and arboreal species is absent.
Asmara, type BSh). In the areas with a temperate climate, the
temperature ranges are never large and the minimum mean
temperatures are recorded in summer, during the big, mon- 3.13 Climate Change
soon-type rains, whereas the milder months are in spring.
The prevailing climate is Cw, with a patent scarcity of pre- Climate change is one of the most investigated issues of
cipitation during the winter semester (bega). Ethiopian climate since its rain-fed agriculture is largely
In the tropical climate region with a dry season in winter dependent on the amount and regular onset of seasonal
(Aw), areas characterized by the maximum temperature in precipitation (Hadgu et al. 2013). Throughout the last cen-
spring—just before the beginning of the kiremt—and the tury, temperature and rainfall trend lines indicate a climatic
minimum temperature during the perturbed phase (July or anomaly that cannot be associated with ordinary cyclic
August)—climate sub-types Awg or Awt—are also found oscillations of these parameters.
(e.g., Jimma, Gambela, and Gonder). In the Rift Valley (e.g., In all the meteo-stations considered, an average temper-
Awasa, Ziway), these conditions are paired by a particularly ature increase of 1.1 °C is recorded for the 1981–2010
small annual range (less than 3 °C) and the climate is Awi. interval (0.04 °C per year). The same increasing rate is
In places of the southwestern portion of the country (e.g., observed for the mean minimum and maximum tempera-
Nekemte and Bonga), rainfalls are rather uniformly distrib- tures. The highest rates are recorded at Neghelle (2.7 °C in
uted in all months (cool variant of the tropical climate) and 30 years), Gonder and Robe Bale (2 °C in 30 years). Such
determine the Cfc climate. marked positive trends do not seem to be influenced by the
84 M. Fazzini et al.

(a) Precipitation trends are very difficult to analyze since the

1.5
time series considered are punctuated by gaps and, some-
1.0
times, the data of 3–5 years in a row are missing. As
0.5 discussed in Sect. 3.5, any attempt to fill the gaps with a
correlation procedure failed to give reliable results. More-
Z (Tmax)

0.0

-0.5 over, the available time series span different intervals which
-1.0
start and end in different years. Nevertheless, some general,
statistically nonsignificant tendency can be depicted on the
-1.5
basis of 20 of the longest (29–46 years) time series
-2.0
selected. 12 stations out of 20 show a negative trend;
-2.5
however, in order to reduce the weight of the gaps scat-
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
tered across the data set, the trend analysis has been
(b) restricted to the 1961–2009 interval and a representative
1.5
time series has been constructed by averaging the annual
1.0 precipitation at all the stations for each year. The rainfall
0.5 anomalies with respect to the long-term mean are plotted
and a negative trend of about 2.6 mm per year is observed
Z (Tmin)

0.0

-0.5
(Fig. 3.23). This diagram confirms that the 1980s and the
first decades of the twenty-first century were characterized
-1.0
by recurrent droughts that caused the spread of severe
-1.5
famines from 1983 to 1985, known as the Great Famine
-2.0 (1 mln fatalities), and another one in 2003 (Bewket and
-2.5 Conway 2007).
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007

The normalized data of the kiremt (Fig. 3.24 a) and belg

(Fig. 3.24 b) rains over the last five decades show that the
Fig. 3.21 Variation through time of normalized mean maximum spring rains are decreasing at a rate 1.5 times faster than the
(a) and minimum (b) temperatures. Dotted line is the 5 years moving
mean
monsoon summer rains. As discussed above, the belg rains
are less voluminous, less reliable, and poorly predictable,
and though agriculture in Ethiopia rely mainly on the sum-
local physiography nor by the expansion of the urban areas. mer rains, the decreasing trend of the spring rains may pose
In fact, at Addis Ababa, the temperature change is negligible severe constraints to the economic development of the
(0.15 °C/30 years) as are the modest decreasing rates mea- country.
sured at Mekele, Alamata, and Arba Minch (0.1–0.2 °C/ The 20 meteo-stations selected are too small in number to
30 years). sort out any significant regional pattern of change in pre-
Eight meteo-stations have longer time series starting in cipitation as neighboring stations may have contrasting
the 1950s (Fig. 3.21). The analysis of their normalized data trends (e.g., Gore and Jimma, the trend line angular coeffi-
shows a higher increasing rate as far as the 1980s, whereas cient of which is −16.2 and +1.4, respectively).
during the last 15–18 years the thermal signal becomes Very few stations have data suitable for an analysis of
stable at 0.6–0.8 °C above the semi-secular average. It is also maximum rainfall intensity variation with time. However, it
worth noticing that for these stations, the mean temperature seems that no significant change occurred during the 1961–
increase is mainly due to the contribution of the minimum 2009 interval.
temperatures that increase at a rate twice as much the Older data are available only for very few stations and
maximum temperatures. refer mainly to the study of Fantoli (1940) who worked on
Figure 3.22 shows the individual temperature changes data ranging approximately from 1890 to 1936, with dif-
recorded by the same eight meteo-stations for two periods: ferent lengths and time intervals, but for some stations the
1953–1980 and 1981–2010. At Addis Ababa, the mean beginning of the measurements is not indicated. Any com-
temperature reported by Fantoli (1940) for three decades at parison with this author’s data is therefore speculative since
the beginning of the twentieth century is the same (16.4 °C) neither information is provided about the exact location of
as that of the 1953–2010 interval, but the former have a the rain gauge nor and whether it was moved during the
much larger daily excursion, i.e., 15.2 versus 13.2 °C. operation period as discussed by Conway et al. (2004).
3 The Climate of Ethiopia 85

Mean temperature
1953-1980 1981-2010

24.7 25.5
20.5
18.1 18.8 18.6 19.5 18.8 19.3 19.1
16.0 16.9 16.1 16.8 15.6 16.6

ADDIS ABABA ALEMAYA DEBRE DIRE DAWA GORE JIMMA KOMBOLCHA NEGELLE
MARCOS

Fig. 3.22 Mean annual temperatures recorded at eight meteo-stations with long data records relative to two distinct intervals, 1953–1980 and
1981–2010

400 (a) Kiremt

1.5
300
1.0
200
Pa (mmyr-1)

0.5
100
Z (P)

0.0
0
-0.5
-100 y = -0.0057x + 0.106
-1.0
-200
y = -2.5602x + 63.964 -1.5
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
-300
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009

(b) Belg
1.5
Fig. 3.23 Rainfall anomalies obtained by averaging the annual
precipitation of all the meteo-stations considered (Pa). Dashed line is 1.0
the 5 years moving mean
0.5
Z (P)

However, just for a qualitative reference, the differences in 0.0

annual precipitation for the 1890–1936 and the 1960–2009 - 0.5
intervals recorded by ten meteo-stations are reported in
Table 3.3. In six out of ten stations, a moderate to marked - 1.0
y = -0.0083x + 0.1604
decrease in rainfall emerges, with an average value of - 1.5
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
−81.1 mm, and in two of them (Adami Tulu and Kassala), it
is very pronounced (around −176 mm). The average rainfall
increase of the other four stations is 58.4 mm, but at Gam- Fig. 3.24 Variation through time of normalized kiremt (a) and belg
bela, a maximum difference of +175 mm is observed. (b) precipitation
Also for the analysis of rainfall intensity change over a
long time, only the data of Fantoli (1940) were used. the reasons explained above, is highly questionable and is to
Unfortunately, only five stations can be used for a compar- be taken just as a mere indication. Though Fantoli (1940)
ison with the modern data of this study (Table 3.4). Two provides some data on other climate parameters such as
stations, Adami Tulu and Gore, show a modest decrease relative humidity, evaporation and wind speed, they refer to
between the two periods with −5.9 and −19.3 mm in 24 h, very few stations and any comparison with modern data is
respectively, and are paired by a small increase at Dese and totally insignificant.
Kassala, 7.0 and 15.5 mm/24 h, respectively. The most Finally, within this framework of climatic change, it is
relevant change is observed at Gambela with an increase of also worth noticing that a few areas, such as the Rift Valley,
about 100 mm in 24 h. The significance of these results, for characterized by aridity index values in the 0.65–0.8 range,
86 M. Fazzini et al.

Table 3.3 Mean annual precipitation difference between the old data Table 3.4 Maximum daily rainfall comparison between the old data of
of Fantoli (1940) and the present study Fantoli (1940) and the present study
Meteo-stations 1890–1936 1960–2009 Difference 1890–1936 1960–2009 Difference
Adami Tulu 756.8 581.1 −175.7 Adami Tulu 87.8 81.9 −5.9
Addis Ababa 1,204.8 1,251.7 46.9 Dese 87.0 94.0 7.0
Asmara 512.3 491.7 −20.6 Gambela 81.3 181.4 100.1
Belet Uen 259.0 260.1 1.1 Gore 127.0 107.7 −19.3
Djibouti 188.0 126.0 −62.0 Kassala 89.4 104.9 15.5
Gambela 1,114.1 1,289.5 175.4
Gore 2,101.3 2,059.3 −42.0
High rainfall intensities in 24 h may occur every month
Kassala 500.0 322.6 −177.4 but are more common in the middle of the two main rainy
Massawa 175.8 186.0 10.2 seasons. The highest average and absolute rainfall intensities
Moyale 685.0 675.9 −9.1 are recorded at Gambela (89.4 and 181.4 mm/24 h, respec-
tively), whereas the lowest values are measured at Elidar and
Gewane in the Afar triangle. The MFI shows that rainfalls
in the last two decades, frequently experienced desertifica- are very aggressive over large part of Ethiopia; the higher
tion conditions with Ai values of less than 0.5. values of rainfall erosivity are found in the central and part
of the northern highlands and, subordinately, in the central
portion of the Somali plateau margin.
3.14 Concluding Remarks The southwest portion of Ethiopia is characterized by very
high relative humidity values, among the highest in the whole
The climate of Ethiopia, though mainly controlled by the African continent, whereas the drier areas are in the northern
position of the ITCZ, is highly variable in space and time. part of the Rift. In the last three decades, the aridity index of
The large contrasts in elevation and morphology that char- this latter area, Ogaden, Afar triangle, Danakil depression and
acterize the country are very effective in controlling local the belt across the border with Sudan, commonly decreased
climate conditions to such an extent that even sites within a beyond the critical value of 0.65; hence, these regions have to
short distance from one another may show very different be considered as desertification prone areas.
climatic conditions. In fact, even though mean, maximum Among the climatic parameters considered in this study,
and minimum temperatures show an almost constant lapse long time series were available only for temperature and
rate, no significant correlation was found for rainfall at precipitation data. The change of both maximum and mini-
national level. Annual rainfall does not change significantly mum temperature through the last three decades is rather
with elevation and seems to be more influenced by the evident with an average increase of 1.1 °C. The data analysis
geographic position of the rain gauge with respect to the of the eight weather stations with longer time series (starting
Sudanese lowlands. The introduction of the F factor, which in the 1950s) shows a marked temperature increase as far as
combines elevation with a specifically developed longitude the 1980s, whereas its rate slowed down and has become
parameter, improved remarkably the statistical explanation almost stable in the last 15–18 years. In most of the stations
of rainfall variability in terms of geographic position. considered, it is the minimum temperature, rather than the
The large orographic variability results in a wide range of maximum temperature that contributes to the mean temper-
temperatures that may be as low as −7 °C in the Bale ature increase. The former increases at a rate twice as much
Mountains and as much as 50 °C in the Danakil Desert. higher than the latter. For the eight stations with longer time
The higher rainfalls are recorded in the western high- series, the comparison between 1953–1980 and 1981–2010
lands, with 2,101 mm year−1 at Gore, and the lower in the intervals indicates that in the most recent decades, mean
Afar depression with 145 mm year−1, though even lower temperatures show an average increase of about 0.9 °C. By
values are expected for the desert areas of the Danakil. The contrast, annual precipitation is characterized by a decreas-
summer, monsoon rains account for more than 50 % and as ing trend with an average decline of about 125 mm in
much as 85 % of the annual rainfall, whereas the spring rains 49 years, with the small spring rains decreasing at a rate 1.5
are more variable in amount and predictability. time higher than the monsoon summer rains.
3 The Climate of Ethiopia 87

References Hurst H, Black P (1937) The Nile basin vol VI. Monthly and annual
rainfall totals and the number of rainy days at stations in and near
the Nile basin for the period ending 1937. Ministry of Public Works
Alexandersson H (1986) A homogeneity test applied to precipitation Press, Cairo
data. J Clim 6(6):661–675 Jury MR, Funk CR (2013) Climatic trends over Ethiopia: regional
Arnoldus HM (1980) An approximation of the rainfall factor in the signals and drivers. Int J Climatol 33:1924–1935
universal soil loss equation. In: de Boodts M, Gabriels D (eds) Korecha D, Barnston AG (2007) Predictability of June–September
Assessments of erosion. Wiley, Chichester, pp 127–132 rainfall in Ethiopia. Mon Weather Rev 135(2):628–650
Beltrando G, Camberlin P (1993) Interannual variability of rainfall in Law B, Falge E, Baldocchi D, Bakwin P, Berbigier P, Davis K, Dolman
the eastern Horn of Africa and indicators of atmospheric circula- A, Falk M, Fuentes J, Goldstein A, Granier A, Grelle A, Hollinger
tions. Int J Climatol 13:533–546 D, Janssens I, Jarvis P, Jensen N, Katul G, Mahli Y, Matteucci G,
Berhanu B, Melesse AM, Seleshi Y (2013) GIS-based hydrological Monson R, Munger W, Oechel W, Olson R, Pilegaard K, Paw UK,
zones and soil geo-database of Ethiopia. Catena 104:21–31 Thorgeirsson H, Valentini R, Verma S, Vesala T, Wilson K, Wofsy
Bewket W, Conway D (2007) A note on the temporal and spatial S (2002) Environmental controls over carbon dioxide and water
variability of rainfall in the drought-prone Amhara region of vapor exchange of terrestrial vegetation. Agric For Meteorol
Ethiopia. Int J Climatol 27:1467–1477 113:97–120
Chernet T (1982) Hydrogeology of the Lakes Region, Ethiopia. Lovett R, Wood CA (1976) Rainfall reliability in Ethiopia. Weather
Memoir N.7, the Provisional Military Government of Socialist 31:417–424
Ethiopia, Addis Ababa McSweeney C, New M, Lizcano G (2010) UNDP climate change
Cheung WH, Senay GB, Singh A (2008) Trends and spatial distribution profile: Ethiopia. http://countryprofiles.geog.ox.ac.uk
of annual and seasonal rainfall in Ethiopia. Int J Climatol 28:1723– Mekasha A, Tesfaye K, Duncan AJ (2013) Trends in daily observed
1734 temperature and precipitation extremes over three Ethiopian eco-
Conway D (2000a) Some aspects of climate variability in the north east environments. Int J Climatol. doi:10.1002/joc.3816
Ethiopian Highlands—Wollo and Tigray. SINET—Ethiop J Sci Mengistu D, Bewket W, Lal R (2013) Recent spatiotemporal temper-
23:139–161 ature and rainfall variability and trends over the upper blue Nile
Conway D (2000b) Climate and hydrology of the upper blue Nile river. river basin. Int J Climatol, Ethiopia. doi:10.1002/joc.3837
Geogr J 166(1):49–62 Pedgley DE (1967) Air temperature at Dallol, Ethiopia. Meteorol Mag
Conway D, Mould C, Bewket W (2004) Over one century of rainfall 96(1967):265–271
and temperature observations in Addis Ababa, Ethiopia. Int J Segele ZT, Lamb PJ (2005) Characterization and variability of Kiremt
Climatol 24:77–91 rainy season over Ethiopia. Meteorol Atmos Phys 89(1):153–180
Easterling DR, Evans JL, Groisman PYa, Karl TR, Kunkel KE, Seleshi Y, Camberlin P (2006) Recent changes in dry spell and extreme
Ambenje P (2000) Observed variability and trends in extreme rainfall events in Ethiopia. Theoret Appl Climatol 83:181–191
climate events. Bull Am Meteor Soc 81:417–425 Seleshi Y, Zanke U (2004) Recent changes in rainfall and rainy days in
Enku T, Melesse AM (2013) A simple temperature method for the Ethiopia. Int J Climatol 24:973–983
estimation of evapotranspiration. Hydrol Process. doi:10.1002/hyp. Shanko D, Camberlin P (1998) The effects of the southwest Indian
9844 Ocean tropical cyclones on Ethiopian drought. Int J Climatol
ENRAEMED (2003) Ethiopian natural resources and environmental 18:1373–1388
meta-database. http://clearinghouse5.fgdc.gov/enraemed/NMSA. Tadege A (ed) (2001) Initial national communication of Ethiopia to the
php United Nations Framework Convention on Climate Change (UN-
Eredia F, DeCastro L (1914) Sula climatologia dell’Etiopia. Boll Real FCCC). National Meteorological Services Agency, Addis Ababa
Soc Geog Ser V III(8):843–884 Tadege A (ed) (2007) Climate change national adaptation programme
Ethiopian Mapping Authority (EMA) (1988) The National Atlas of of action (NAPA) of Ethiopia. National Meteorological Services
Ethiopia. EMA, Addis Ababa Agency, Addis Ababa
Fantoli A (1940) Elementi preliminari del clima dell’Etiopia. Sansoni, Taye M, Zewdu F (2012) Spatio-temporal Variability and Trend of
Firenze Rainfall and Temperature in Western Amhara, Ethiopia: a GIS
Fantoli A (1965) Contributo alla climatologia dell’Etiopia. O.P.I, Roma approach. Glob Adv Res J Geogr Reg Plann 1(4):65–82
FAO (1984) Agroclimatological data for Africa, vol 1: Countries North Thomas DSG, Middleton NJ (1994) Desertification: exploding the
of the Equator. FAO Plant Production and Protection Series No. 22, myth. Wiley, Chichester
FAO, Rome UNEP (1992) Status of desertification and implementation of the
Gamachu D (1977) Aspects of climate and water budget in Ethiopia. United Nations plan of action to combat desertification. Report of
Addis Ababa University Press, Addis Ababa the Executive Director, United Nations Environment Programme,
Gamachu D (1988) Some patterns of altitudinal variation of climatic Nairobi
elements in the mountainous regions of Ethiopia. Mount Res Dev 8 Vinassa de Regny P (1931) La geologia delle Alpi Dancale. Bollett Soc
(2/3):131–138 Geol Ital 50(1):1–24
Hadgu G, Tesfaye K, Mamo G, Kassa B (2013) Trend and variability of Viste E, Korecha D, Sorteberg A (2013) Recent drought and
rainfall in Tigray, Northern Ethiopia: analysis of meteorological precipitation tendencies in Ethiopia. Theoret Appl Climatol 112
data and farmers’ perception. Academia J Env Sci 1(8):159–171 (3–4):535–551
Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, WWR (1959) World weather records 1941–50. US Department of
Xiaosu D (eds) (2001) Climate change 2001: the scientific basis: Commerce, Weather Bureau, Washington, DC
contributions of working group I to the third assessment report of
the intergovernmental panel on climate change. Cambridge Uni-
versity Press, Cambridge

View publication stats