Research Journal of Chemical and Environmental Sciences

RJCES
Res J. Chem. Environ. Sci. Vol 7 [3] June 2019: 35-51
Online ISSN 2321-1040
OCODEN: RJCEA2 [USA]
©Academy for Environment and Life Sciences, INDIA
Website: www.aelsindia.com/rjces.htm
ORIGINAL ARTICLE
Climate variability and hydrogeological responses in the
mountainous area of Ennedi (Chad) using remote sensing
1MiradjHabib Djafar, 1Abderamane Hamit and 2Marie-louise vogt
1Department of Geology, Faculty of Exact and Applied Sciences, University of N’Djaména, Chad
2Centre d’Hydrogéologie et Géothermie, University of Neuchâtel, Neuchâtel, Switzerland

ABSTRACT
Through the study of climate variability and its response on the hydrogeology of the Ennedi area, this research aims to
improve knowledge on the functioning of the aquifer system of the Ennedi mountain area with a view to better
mobilization and sustainable management of water resources. Ennedi is characterized by an arid to hyperarid climate
where the only available water resource is groundwater, which has been greatly weakened by climate change in recent
decades. A study using remote sensing was carried out. The result of this study provided a good understanding of the
spatial and temporal variability of climate and the relationship between rainwater and groundwater. The rainfall maps
highlighted the heterogeneity of the spatial distribution of rainfall and made it possible to locate deficit and surplus
areas. Results obtained in this study confirm once again the possibilities offered by Landsat8 OLI/TIRS satellite data to
solve the energy balance equation by estimating evapotranspiration. The calculation of the water balance made it
possible to quantify the volume of water that will contribute to the recharge of the Paleozoic sandstone aquifer of the
Ennedi.
Key words: Ennedi, climate variability, hydrogeological responses, recharge, aquifer, remote sensing

Received 07.04.2019 Accepted 25.05.2019 © 2019 AELS, INDIA

INTRODUCTION
Located on the northern and eastern borders of Chad, Ennedi is a spacious geomorphological unit that,
together with Borku and Tibesti, forms the southern border of the great SAHARA. It contains a huge
mountainous massif (which is the mountainous area in question) with impressive natural beauty and
biodiversity systems. The Ennedi massif is a succession of sandstones ranging from Cambro-Ordovician to
Upper Devonian, resting in discordance on the Precambrian basement and underlying the marine
limestone of the Carboniferous age. It is highly fractured and eroded by wind and thermal cycles that have
significantly increased its storage capacity and transmissivity[10]. At an altitude of 1,450m, it records
increased precipitation due to these increases compared to the surrounding lowlands, thus giving it a
favourable climate[18]. Due to its remoteness and uncomfortable climatic conditions, the Ennedi region
has still relatively been unexplored. There is very little knowledge about water resources. Ennedi is
characterized by an arid to hyperarid climate where the only available water resource is groundwater,
which has been greatly weakened by climate change. Recent climate trends show that since 1970 the area
has been subject to a 50% decrease in annual rainfall [3]and a temperature increase of 0.8°C [6]compared
to the previous period. It is expected that future developments will continue to follow this same trend,
putting the resource under pressure with significant consequences. Evaporative events are extremely
high compared to rainfall [12]; [3], only part of the precipitation infiltrates, the importance of this
infiltration determines the possibilities of groundwater recharge. Quantifying the proportion of
precipitation that recharges groundwater and capturing its frequency would go a long way towards
improving our knowledge of storage capacity and aquifer dynamics for effective water resource
management.This study aims to evaluate the spatial and temporal variation of rainfall in the Ennedi, to
see the response of the aquifer system to these variations, and to make a water balance including an
estimate of the quantity of water that would contribute to modern groundwater recharge as well as the
frequency of this recharge. Test new methodologies to overcome the lack of field information in this
region using an innovative approach based on the use of the latest advances in space technology. In
particular, the estimation of evapotranspiration from Landsat8 images by the energy balance equation
using the S-SEBI approach.

RJCES Vol 7 [3] June 2019 35 | P a g e © 2019 AELS, INDIA

 Djafar et al

Geographical and Geological Background

The Ennedi region is located in the north-east of Chad between the 15th and 21st degrees of latitude North
and the 20th and 24th degrees of longitude East (Fig.1). It is characterized by a Saharan climate, arid in its
southern part and hyperarid in its northern part. This character is linked to the annual influence of the
monsoon which brings modest rainfall, especially in the southern part of the Ennedi. The region contains
several morphological units: the Mourdi depression in the north, the Djourab on the western edge of the
massif, the Mortcha plain and the Ennedi massif.The Ennedi massif takes the form of a rectangular
triangle with a total estimated surface area of 51,210 km2[14]. It is bounded to the south by the
Precambrian basement of the Ouaddaï and to the north by the Mourdi depression which separates it from
the Erdis relief line and the Jef-Jef plateaus constituting the boundary between the Chadian and Libyan
basins [18]. To the east, the reliefs of the Erdébé plateau, the eastern end of the massif, are gradually
invaded and disappear under the sands of the Sudano-Libyan desert. To the west, the relief of the Ennedi
massif fades into the Bir Kora area, while the Precambrian basement and Cambrian-Ordovician
sandstones gradually sink deeper.

Figure1: Location of the study area

After the Pan-African orogeny (750 MA to 600 MA), an important basin was created at the level of the
Ennedi and which was filled from the beginning of the Cambrian by clayey-gravel deposits of continental
origin. These sandstone complexes lie in unconformity on the Precambrian basement and overlook the
carboniferous formations. The formations often have good permeability related to cracking and
fracturing due to tectonic actions [17]. Figure 2 illustrates the different geological units of the area
according to Wolf [19].
Data used and methods
The lack of field data has led us to use remote sensing data. Today, with the development of technology,
satellites are producing information as quickly as possible. They offer good regularity over time and
homogeneity in spatial coverage. The use of satellite data has increased significantly in recent years in
multiple research areas: estimation of daily precipitation, estimation of evapotranspiration,
determination of NDVI or NDWI.The database consists of three data sets. A first set was created using the
RFE2/FEWS-NET images to extract rains at monthly and ten-year time steps for the time series 2000-
2016. The second one contains the observed daily evapotranspiration values for each pixel of the
LandSat8 OLI/TIRS images over the period 2013 to 2015. And the last one concerns the variation in
storage obtained by synthesizing the GRACE data at monthly time steps.

RJCES Vol 7 [3] June 2019 36 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure2: Geological map of the study area (Source: Geological map of the Republic of Chad by Wolf
[19])
Method for analyzing rainfall variability
Since RFE2 provides quarterly, bimonthly, monthly and decadal data, we have chosen the latter two for
this study. The ten-year data were used to calculate the water balance because of their temporal
resolution (10 days). The monthly data were used to study rainfall variability. As the rainy season in the
Ennedi massif is limited only to July, August and September, the data set used concerns only these three
months (from the first decade of July to the last decade of September).
The analysis of rainfall variability by anomaly (relative deviation) made it possible to characterize the
rainfall between the inter-annual mean of the series studied and the annual mean totals. The anomaly
calculation makes it possible to highlight the surplus and deficit periods within a time series.

Where
E: Relative deviation or anomaly (%)
Pi: Average annual precipitation recorded during the year i (mm)
Pmoy = Mean inter-annual precipitation of the series studied.

Method for estimating evapotranspiration

To estimate evapotranspiration, Landsat8 OLI/TIRS images are used. Indeed, they make it possible to
detect changes and variations in the presence of water at a scale adapted to the size of the hydrographic
network that conserves water for a sufficient time to be captured by the satellite. To cover the study area,
three Landsat scenes were required: 81/48, 180/48 and 180/49. From these scenes, we chose a series of
images without clouds (0% clouds) but with the presence of surface water accumulations that are used to
calculate evapotranspiration. The lack of water in the images results in a cessation of evapotranspiration
phenomena. However, the exchange of energy between the ground and the atmosphere continues. On this
basis, we used 42 Landsat8 OLI/TIRS satellite images: 14 for 2013, 18 for 2014 and 10 for 2015. Details of
all these images are recorded in Table 1.

RJCES Vol 7 [3] June 2019 37 | P a g e © 2019 AELS, INDIA

 Djafar et al

Tableau1: Description of the images used to calculate evapotranspiration

Path/raw 2013 2014 2015
image ID Acquisition date ID Acquisition date ID image Acquisition date
image image
181/48 234 August 22nd 237 August 25 th 240 August 28 th
181/48 250 September 7th 285 October 12 th 272 September 29 th
181/48 282 October 9 th 301 October 28 th 288 October 15 th
181/48 298 October 25 th 317 November 13 th 304 October 31 st
181/48 333 November 29 th
180/48 227 August 15 th 214 August 2nd 217 August 5 th
180/48 243 August 31st 230 August 18 th 249 September 6 th
180/48 259 September 16 th 246 September 3 rd 281 October 8 th
180/48 275 Octobrer 2nd 262 September 19 th
180/48 307 November 3rd 278 October 5 th
180/48 294 October 21 st
180/48 310 November 6 th
180/48 326 November 22nd
180/49 243 August 31 st 230 August 18 th 217 August 5 th
180/49 307 November 3 rd 246 September 3 rd 249 September 6 th
180/49 323 November19th 278 October 5 th 281 October 8 th
180/49 339 December5th 310 November 6 th
180/49 355 December 21st 326 November 22nd

For the determination of evapotranspiration by remote sensing, the latent heat flux (λE which is the
equivalent of evapotranspiration) is generally estimated through the resolution of the energy balance
equation [11]:
λE = Rn – G – H
λE: the flow of latent heat released by evaporation (W.m-2) ;(W.m-2);
H: the sensible heat flux released into the atmosphere (W.m-2);
Rn: the net radiation at the surface (W.m-2;
G: the heat flux in the ground (W.m-2).
The method used to estimate evapotranspiration is based on the S-SEBI (Simplified Surface Energy
Balance Index) approach. This approach uses the correlated relationship between latent heat and sensible
heat, which is highly dependent on the water content of the system, to estimate the evaporation
fraction[15]. The general assumption is that wet surfaces have an evaporation fraction of 100% and
completely dry surfaces have an evaporation fraction of 0%. The evaporation fraction, considered stable
at a one-day scale [2], is used to allow the passage from an instantaneous magnitude of latent heat flux
measured by the satellite to a daily magnitude that corresponds to the actual evapotranspiration [15].
The following methodological flowchart (Fig. 3) illustrates the handling procedures.

RJCES Vol 7 [3] June 2019 38 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure3: Method of calculating evapotranspiration from Landsat8 OLI/TIRS images

The preliminary processing consists in transforming the bands into spectral luminances. An
atmospheric correction follows this step to remove shadows that can be confused with water
accumulation areas. After this correction, the spectral luminances are converted into radiances and
reflectance's and are then used to calculate the surrounding albedo( ) and surface temperature.
 Surrounding albedo ( ) characterizes the fraction of solar radiation falling at the bottom of the
atmosphere that is reflected from the surface [1]. The procedure used to calculate the albedo is the one
proposed by Da Silva and his collaborators[4]:

r: Surface albedo
rtoa: Albedo without atmospheric correction (without unit)
ratm: Atmospheric albedo = 0.03
τoc2: Atmospheric transmittance in the solar radiation domain = 0.5
 The surface temperature was calculated using the method of Du and his
collaborators[5]according to the formula:
TC = Tk – 273.15
With

Where
TC : Surface temperature in degrees Celsius
Tk : Surface temperature in Kelvin
b10T and b11T : the temperatures (in Kelvin) of bands 10 and 11 respectively;
 The evaporation fraction (EF) is the energy partition between the different components of the
energy balance. It is calculated according to the S-SEBI [15]. It is expressed as a function of surface
temperature ( , albedo ( ) and variables a, b, c and d which are manually determined using the

RJCES Vol 7 [3] June 2019 39 | P a g e © 2019 AELS, INDIA

 Djafar et al

properties of wet/dry pixels and surface water conditions from the graph relating reflectance and surface
temperature (fig. 3).

Evapotranspiration can be defined as the amount of water that is dissipated into the atmosphere by the
vaporization process. Daily evapotranspiration is obtained by combining the daily net radiation and the
evaporation fraction. We thus move from an instantaneous quantity to a daily quantity. Parodi method
was used to express actual daily evapotranspiration (AET)[13].

With Rn the daily net radiation (W m-2)

Calculation of the water balance
We considered a water balance whose general assumption is that all waters that precipitate within the
watersheds converge towards the rivers. Some of this water evaporates and some infiltrates (considered
here as the amount of water that contributes to the recharge). The balance is characterized by the only
input, which is the contribution of precipitation, and the only output, which is evapotranspiration. The
difference between these two components is therefore the residual meaning the quantity of water
involved in the recharge. Underground flows and contributions from adjacent basins are excluded and
are not included in this exercise. The water balance is established according to the following general
equation:
P - ETR = +R+I
With:
P: Precipitation (in mm);
AET: Actual evapotranspiration (in mm).
R Run-off (mm)
I: Infiltration (mm)
Where R+I is the effective rain that contributes to recharging.

RESULT AND DISCUSSION

Annual rainfall variability from RFE2/FEWS-NET
The distribution of rainfall from RFE2/FEWS-NET is very irregular. This irregularity shows that the
region is subject to particular meteorological phenomena that do not follow a single trend. Rainfall is
generally less than 100 mm/year with a high inter annual variability, typical of an arid climate. The
wettest year is 2010 with a total of 113 mm and the driest year is 2004 with a total of 24 mm (Fig. 14). It
can be seen that at the series level, each wet year is followed by a less humid year than it is. And, when
two less humid years follow one another, the following year will be particularly humid (example of the
year 2010). This same observation was observed from 2014 onwards, when there was a decrease in the
annual totals for 2015 and 2016. If the evolution follows the trend explained, it is expected that 2017
would be wetter than 2015 and 2016. The inter annual average of the series is 61mm. Note that
RFE2/FEWS-NET overestimates precipitation estimates, however, it is important not to lose sight of the
errors that these values may have.
The coefficient of variation (defined as the ratio between the anomaly and the inter annual mean) is
slightly high (0.42), therefore the dispersion of the annual mean rainfall amounts around the inter annual
mean has varied considerably.

RJCES Vol 7 [3] June 2019 40 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure4:Histogram of average annual precipitation from 2000 to 2016 calculated from

RFE2/FEWS-NET

The analysis of rainfall variability by the anomaly test (relative deviation) made it possible to distinguish
two climatic periods (Fig. 5). A period of poor rainfall (2000 - 2009) characterized by a dry climate and a
surplus period (2010 - 2016) where the climate is humid.

Figure5: Histogram of the rainfall anomaly

The observation of the rainfall distribution histogram (Fig. 6), showing the rainfall amounts for the two
periods, shows a shift to the right of the graph for the rains from 2010 to 2016 compared to the decade
2000-2009. This shift is explained by an increase in precipitation for this period. Although there is a very
high variability that is difficult to identify, a rainfall deficit can be observed over the dry period. The
annual average for this period is 51 mm. From 2010 onwards, there is a recovery in good rainfall, annual
totals of over 100 mm (2010 and 2014) have been recorded and the annual average rainfall is 75 mm, an
increase of 24 mm on average.

RJCES Vol 7 [3] June 2019 41 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure6: Rainfall distribution histogram

Annual precipitation mapping

The annual precipitation mapping reflects the spatial evolution of rainfall variability while highlighting
deficit and surplus areas. There are four well-defined climate zones: The spatial distribution of
precipitation is heterogeneous throughout the study area. Rainfall variability, strongly influenced by
climate change in recent decades, reflects the great heterogeneity in the spatial and temporal distribution
of rainfall during this period. The analysis in Figure 7 shows a clear incurvation of isohyets marked by a
gradient that gradually and steadily decreases in a southwest to northeast direction: mean annual rainfall
totals range from over 100 mm in the extreme southwest to values below 20 mm in the Mourdi
depression. There are four well-defined climate zones:
 The extreme southwest at the level of the Mortcha plain and part of the southern edge of the massif
(Bachiquélé) which are the most watered receive more than 100 mm of rainfall;
 The southern edge of the massif, at the foothills of the Fada and Archei line, which lies between the
100 and 50 mm isohyet;
 The high plains, at Bao, Moutougouni and the northern edge of the massif where the annual rainfall is
between 20 and 50 mm;
 The northern part of the Mourdi slope north of the 20 mm isohyet corresponds to the hyperarid zone.

Figure7: Spatial variation of precipitation from 2000 to 2016

RJCES Vol 7 [3] June 2019 42 | P a g e © 2019 AELS, INDIA

 Djafar et al

Precipitation mapping for both periods (surplus and deficit)

By referring to the calculation of anomalies, a distinction is made between the deficit period and the
surplus period. The cartographic representation of precipitation for these two periods varies
significantly from one period to another. The interpretation of Figure 17 shows that:
 During the 2000 - 2009 period, the northern part of the area (from the Mourdi basin divide and the
Mortcha/Djourab divide) and the northeastern part of the Bao basin, which occupy half of the area,
received rainfall of less than 20 mm/year. A reduction in rainfall is also observed in Fada, Archei and
Amdjarass, which have rains of between 20 and 50 mm/year. In short, there is a reduction in the
rainfall field compared to the interannual average resulting in a sliding of isohyets. Isohyet 80 slides
southwest and gives way to isohyet 50;
 The 2010 - 2016 period is characterized by a recovery in rainfall, which generally waters the entire
area. This increase is marked by the migration of isohyets. The isohyet of 20, which covers half of the
area in the previous period, migrates to the Northeast and shrinks to a small part. On the southern
edge of the massif, there is an increase in rainfall of more than 100 mm/year, a fact not observed
during the previous period.

Figure8: Spatial variability of rainfall in both periods (deficit and surplus)

Contribution of remote sensing to evapotranspiration estimation

Given the complexity of the handling techniques and the enormous time required for processing, the
study focused on the years 2013 to 2015 to estimate evapotranspiration. The variables required to
estimate the evapotranspiration obtained are: surface temperature, albedo, net radiation and evaporation
fraction.
Surface temperature
The values obtained for the surface temperature are speckled with errors for some images. It was
necessary to correct the data before it was used. For all images, the surface temperature varies between
10.2ºC corresponding to the minimum value of day 355 and 58.8ºC corresponding to the maximum value
of day 259. This spatio-temporal variability of temperature highlights the particularity of the landscape
characterized by bare ground occupying most of the area, which thus causes an advection phenomenon
resulting in an increase in the evapotranspiration process. The higher temperature values generally
correspond to surfaces where bare soils are dominant, while the lower values are due to very recurrent
humidity in some areas of the wadis where there is water accumulation and/or dense vegetation cover all
around (Fig.9).

RJCES Vol 7 [3] June 2019 43 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure9: Corresponding surface temperature on day 230 of 2014

Surface albedo
Albedo expresses the ratio between the energy of electromagnetic waves reflected by a surface, by
reflection or scattering and the energy of incident electromagnetic waves. The values obtained range from
0.02 to 0.95. Albedo is very high in bare soils covered with very light sands that over-reflect radiation. It is
a little low for the dark body, as in the case of rock formations, as shown in Figure 10. The surface albedo
does not vary too much from one image to another.

Figure10: Variation of the surface albedo

Net radiation
The net radiation values obtained from the images range from 3.7 W/m2 to 225.7 W/m2. Wet river beds
and surface water accumulation areas are represented by high values. Bare soils are characterized by low
values (Fig. 11).

RJCES Vol 7 [3] June 2019 44 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure11: Distribution of net radiation

Actual Evapotranspiration (AET)

Actual evapotranspiration is a major component of the water balance. It expresses the mass and energy
exchanges between the soil-water system and the atmosphere. The latent heat flux E, which represents
the energy equivalent of evapotranspiration, has been estimated by remote sensing methods that are
essential tools for evaluating evapotranspiration both in space and time. Values range from 0 to 7.9 mm
evapotranspiration per day. They seem to have some consistency despite the uncertainties. High
evapotranspiration values are observed over areas of surface water accumulation and in wet beds of
wadis where albedo is low (Fig. 12). Bare soils covered with sands that have a very high albedo have low
evapotranspiration values.

Figure12: Example of the distribution of evapotranspiration

RJCES Vol 7 [3] June 2019 45 | P a g e © 2019 AELS, INDIA

 Djafar et al

Variation in groundwater storage

With GRACE, it is now possible to track the interannual variation of groundwater storage. The test done in
this study revealed the strong annual and interannual variations (Fig. 13). Since the variations in the
gravity field related to rock masses do not change annually, it is certain that these variations are due to
water storage which varies not only annually but also month after month. And as in the region, there is no
other source of water that feeds groundwater tables except rainwater, it can be said that rainwater is
responsible for these variations. Coupled with precipitation data, GRACE data show some correlation (Fig.
14). We can see that the change in storage is very sensitive to rainfall variations. As the rainy season
approaches, the variations are so small that they are close to zero, which supports our assumption that
the storage variation is considered to be zero. A slight increase was observed at the end of July in
response to the rains that fall to recharge the aquifer system. This increase continues until the end of the
rainy season before relapsing.

Figure13: Storage variation between 2013 and 2016

Figure14: Correlation between GRACE data and rainfall data

Water balances
The calculation of the water balance is the most commonly used method for estimating recharge [8]. It is
also an important tool for better understanding certain issues related to water in arid zones. It therefore
makes it possible to highlight the evolution of effective rainfall given by the difference between rainfall
inputs and evapotranspiration losses. If we consider a water balance at annual time steps, we will
obviously have a deficit balance because annual evapotranspiration is much higher than annual
precipitation in any given year. Hence the interest of carrying out a monthly study that only concerns the
rainy season (July, August and September) since it is in these months that the rains are recorded. We have

RJCES Vol 7 [3] June 2019 46 | P a g e © 2019 AELS, INDIA

 Djafar et al

therefore calculated water balances for three consecutive years (2013, 2014 and 2015) to quantify the
proportion of precipitation that escapes evapotranspiration and reaches aquifers to recharge
groundwater (Table 2).
Table2:Summary of water balance results
2013 2014 2015
Days P (mm) ETR (mm)R (mm) P (mm) ETR (mm)R (mm) P (mm) ETR (mm)R (mm)
20-31 Jul 23.1 14.9 8.2
1-10 Aug 36.1 14.1 30.2 21.28 15.14 6.14
10-20 Aug 23.4 14.75 8.65 34.6 20.2 44.6 26.7 13.02 19.82
20-31 Aug 42.4 18.58 32.47 19.5 16.6 47.5 30.83 14.24 36.41
1-10 Sep 0.26 12.89 19.84 5.9 11.3 42.1 4.1 19.02 21.49
10-20 Sep 0.07 7.77 12.14 3.9 8.5 37.5 5.29 9.33 17.45
20-30 Sep 0 5.62 6.52 0 5.2 32.3 0 3.41 14.04
1-10 Oct 0 4.37 2.15 0 3.4 28.9 0 2.77 11.27
10-20 Oct 0 0.81 1.34 0 2.9 26 0 4.06 7.21
20-31 Oct 0 0.32 1.02 0 2.5 23.5 0 1.98 5.23
1-10 Nov 0 0.28 0.74 0 1.4 22.1
10-20 Nov 0 0.3 0.44 0 1.1 21
20-30 Nov 0 0.29 0.15
Water balance for 2013
The histogram below shows the variation over time of the water balance for the year 2013 (Fig. 14).
Although the rainy season begins in July, this month is in deficit. Indeed, there is no effective rainfall. The
evaporative demand from the atmosphere is very high with a strong decrease or even drying out of the
soil's water supply. The first effective rains begin to occur from the first decade of August and reach their
maximum at the end of this month with a height of 32.5 mm, it is the wettest month. The curve drops
sharply in the first decade of September with a decrease of more than 10 mm in 10 days and then
continues until it is cancelled at the end of the year. Despite the fact that in August we have a good and
efficient water wave, the extremely arid climate of the area has not allowed us to increase the availability
of water for recharge. Low rainfall with low intensities occurs during this year. As a result, there is a
residual of 0.15 mm at the end of the season, i.e. a rate of 0.22% of annual rainfall. There is no longer
enough water to supply the underground tanks. The year 2013 is less favourable to recharging.

Figure15: Evolution of effective rainfall in 2013

Water balance for 2014
This year (2014) seems more favourable to recharging than the previous one (2013). The water balance
records surpluses of more than 30 mm concentrated over the last twenty days of August and the first

RJCES Vol 7 [3] June 2019 47 | P a g e © 2019 AELS, INDIA

 Djafar et al

twenty days of September (Fig. 16). After repeated rainfall events in August, residual rains increased
significantly with a sharp 47.6 mm peak at the end of the month. After the last rains in September,
evapotranspiration continues to reduce residual rainfall until all surface water disappears. The residual
at the end of the season is 21 mm, or 17% of the annual rainfall.

Figure16: Evolution of effective rainfall in 2014

Water balance for 2015

The water balance for 2015 is different from previous years. The effective water wave recorded at the
beginning of the season is 6.1 mm and then gradually the maximum accumulation reaches 36.4 mm. The
evaporation recovery causes the residual rains to drop, a 15 mm drop is recorded in less than ten days.
This means that in this interval, there were no heavy rains with high intensities. At the end of the rainy
season, after all surface water has disappeared, there is a residual of 5.23 mm (Fig. 26). The rate
equivalent to this residual is 6% of the annual rainfall.

Figure17: Evolution of effective rainfall in 2015

DISCUSSION
Rainfall estimates from RFE2/FEWS-NET seem to be close to reality. Nevertheless, there are too many
gaps in the field data to compare them with them. The analysis of rainfall variability shows a very high
irregularity in the distribution of rainfall typical of a hyperarid environment. The influence of climate
change in recent decades is believed to be one of the main causes [3]. The anomaly calculation indicates
an upward trend marked by the recovery of good rainfall from 2010. This slight increase in rainfall does
not reach the average of the years before 1970. This result is consistent with those found at the sub-
regional level ([9]; [20]). This increase is also felt at the level of groundwater recharge, as confirmed by
the results of GRACE and water balances.

RJCES Vol 7 [3] June 2019 48 | P a g e © 2019 AELS, INDIA

 Djafar et al

The spatial distribution of precipitation shows a strong heterogeneity. As everywhere else in the country,
there is a latitudinal evolution following a gradual and fairly regular decrease in precipitation [16]. The
characterization of the four climatic zones would be due to the massif which constitutes a real climatic
barrier to monsoon flows. This barrier curves the isohyets, placing the city of Fada, located at an altitude
of about 570 m, and the city of Amdjarass, located at an altitude of about 830 m, on the same 80 mm
isohyet. To the south of the massif, the area is arid and receives modest rains that rarely exceed 100
mm/year. Below this, the whole area is hyperarid and receives very low rainfall. The very high heat
associated with disturbances in air mass circulation and other atmospheric parameters could be
responsible for this distribution. And on the other hand it would be the latitudinal and topographical
effects.
The results of the evapotranspiration estimation prove the effectiveness of the model. It seems to have
worked well to estimate evapotranspiration from Landsat8 OLI/TIRS images for the Ennedi area and the
entire northern part of the country. Because these arid areas have similar physical characteristics. We
expected values less than 10 mm and this was the case. It should be noted that the model underestimates
these values. This underestimation can be explained by uncertainties about intermediate variables
entering the model such as albedo and surface temperature. Indeed, Landsat8 images are acquired in the
morning around 9am, at which time the air temperature is low, which causes a decrease in the values of
these two parameters. The accuracy of the model depends on the presence of water in the images, so the
more water you have, the more accurate the estimate. However, at the end of July and in August, when
almost all the rains fall, reasonable evapotranspiration values are observed. In the absence of water on
the images, especially from September onwards, the model underestimates evapotranspiration.
However, these estimated values should be considered with caution, as there is a lack of point field data
measured on the ground to compare and validate them with satellite images. Similar studies have also
shown the effectiveness of satellite images in estimating evapotranspiration [11];[7].
The method we used is at the scale of the catchment areas, but in these hyperarid areas the infiltration of
water is very local and depends on the spatial and temporal variability of rainfall. If these constraints are
taken into account, the effective rainfall could be higher than that calculated at the basin scale. The results
of this study suggest that below a rainfall of 60 mm/year, effective rainfall is almost zero. However, there
is a nuance that needs to be clarified, as recharging depends on the intensity and duration of the instant
rainfall. This implies that this possible modern recharge could be explained by the fact that heavy rains
with high intensities sometimes fall, leaving a sufficient quantity of water to escape evapotranspiration.
The moisture accumulated during the summer and the geological structure of the region, consisting
mainly of highly fractured primary sandstones and therefore having high porosity, facilitate water
infiltration. Although there is a resumption of evaporation from the unsaturated zone, deep infiltration is
not excluded because the groundwater level generally does not exceed 30 m at the southern edge of the
massif. GRACE's results confirm these results. Because during the period 2013 - 2015, there is an
increase in the potential for groundwater recharge.
As the variation in the piezometric level is at least partly dependent on the conditions of the recharge of
the groundwater by rainwater, we can already make a correlation between them. The seasonal
fluctuations in piezometry measured in 2013 and 2016 on the same structures argue in favour of direct
seasonal recharge during the rainy season. The comparison concerns 26 structures, including 21
structures that capture the Paleozoic sandstone aquifer, 4 structures that capture alterations and
fractures of the basement and only one structure that captures alluvial groundwater. The result shows
that the piezometric level increased in 77% of cases, or 21 out of 26 structures (Fig. 18). In the rest of the
structures, there was a decrease in the piezometric level (6 structures or 23%). The increase in water
level varies from 0.12 to 4.39 m, with an average of 1.7 m. In general, these variations in groundwater
level can be related to precipitation. Water level fluctuations, although not significant, prove that the
water table receives a supply that renews water stocks.

RJCES Vol 7 [3] June 2019 49 | P a g e © 2019 AELS, INDIA

 Djafar et al

Figure18: Piezometric level fluctuations of wells

CONCLUSION
The results obtained in this study confirm the potential of satellite data to study climate variability and
assess recharge in arid regions with low cloud cover. Despite the uncertainties associated with the
different sources that have different spatial and temporal resolutions and despite the lack of ground data,
the approach we used to estimate this uncertainty seems to be good for estimating a number of climate
parameters. The advantage of these data is that they offer a good spatial distribution that is more precise
and detailed than those in the field. The only disadvantage is that they have a very degraded temporal
repetitiveness (10 days for FEWS-NET, 16 days for Landsat and about 1 month for GRACE), which hinders
the quality of their monitoring.The study of climate variability shows that despite the exacerbated
climatic conditions, the region is experiencing increased rainfall under the influence of the monsoon.
However, the annual and monthly distribution is very heterogeneous. Deficit and surplus years appear
with sometimes more than 100 mm of rainfall per year, which is rare in hyperarid areas. The annual
average found in this series is 61 mm/year, well below that estimated before this period. The anomaly
study identified two periods: a deficit period from 2000 to 2009 and a surplus period from 2010 to 2016.
The cartographic representation of annual rainfall shows that the evolution of rainfall follows a
southwest - northeast gradient. Isohyets evolve in curvature putting areas of different altitudes on the
same level. The southern part of the massif records more rainfall at the expense of the northern slopes
where the massif is a barrier to monsoon flows. Mapping of the two deficit and surplus periods shows
that the last period is marked by a migration of isohyets towards the Northeast, which results in an
increase in precipitation. Coupled with rainfall data, GRACE data show that the aquifer system is very
sensitive to rainfall variations. Water storage varies according to rainfall. As the rainy season approaches,
there is almost no variation, which proves our water balance hypothesis. The calculation of the water
balance allowed us to understand that despite the intense evaporation activities, in a humid year, there
will be a quantity of water that escapes and reaches the aquifers to feed the aquifers, sometimes sufficient
to meet the needs of the population. This resource deserves to be well managed, because with the current
climate change if years of drought are repeated, they will greatly weaken this resource. The S-SEBI model
tested confirms the possibilities of solving the energy balance equation from LandSat8 OLI/TIRS images
to estimate evapotranspiration in the northern part of the country, unlike theMODIS model tested by the
ResEau project.

REFERENCES
1. Arino O. (1990). Surface albedo and short wavelength radiation balance: satellite contribution. Thesis Univ.
Toulouse, INPT.
2. Bastiaanssen W.G.M, Ahmed M.D., Chemin Y. (2002). Satellite surveillance of evaporative depletion across the
Indus Basin, Water Resource.Res. 38, pp. 1273-1282.
3. BURGEAP (2001). Pastoral hydraulics project in the B.E.T. Well sections and characteristics. End of project
report, Ministry of the Environment and Water, 163p.

RJCES Vol 7 [3] June 2019 50 | P a g e © 2019 AELS, INDIA

 Djafar et al

4. Da Silva B.B, Braga A.C, Braga C.C, De Oliveira L.M.M, Suzana M. G. L, Montenegro S.M.G.L, Junior B.B (2016).
Procedures for calculation of the albedo with OLI-Landsat 8 images: Application to the Brazilian semi-arid. Res.
Brasileira de Engenharia Agrícola e Ambiental, v.20, n ° 1, pp. 3-8.
5. Du C, Ren H, Qin Q, Meng J, Zhao S, (2015) .A practical split-window algorithm for estimating land Surface
Temperature from Landsat 8 data. Remote Sens., Pp. 647-665.
6. FEWS-NET (2012). A climate trend analysis of Chad.Fact sheet 2012-3070.
7. Hamimed A, Nehal L, Khaldi A, Azaz H (2014). Contribution to the spatialization of the evapotranspiration of a
semi-arid agro-system in Algeria through the use of remote sensing and the METRIC model. Physio-Geo, Volume
8.
8. Jérôme A.N, Yoboue K.B, N’daoule R (2015). Rainfall variability and its impact on water supply in rural areas in
the department of Dimbokro (central-eastern Ivory Coast). European Scientific Journal, vol.11, No.35.
9. Mahé G (2006). Rain-flow variability in West and Central Africa in the 20th century: hydro-climatic changes, land
use and hydrological modeling. Thesis Univ. Montpellier 2, 160p.
10. EH (2014). Hydrogeological map of the Republic of Chad at 1: 500,000. Sheet NE-34-SE Ennedi.
11. Olioso A, Jacob (2002). Estimation of evapotranspiration from remote sensing measurements. La Houille
Blanche, n ° 1, pp. 62-67.
12. ORSTOM (1962). Study of surface hydrology in sub-desert and desert regions of black Africa. Study of the Ennedi
massif and the northern region of Mortcha (Chad Basin). Land and water v. 38 pp. 46-73.
13. Parodi G.N, (2002). Algorithms and theory - Version 1.3, AHVRR Hydrological Analysis System. 77P.
14. Roche M (1960). Overview of the climate and hydrology of the Ennedi massif and the Mortcha plain. Saharan
liaison bulletin n ° 37 pp. 41-51.
15. Roering G.J, Su Z, Menenti M (2000) .S-SEBI: A simple remote sensing algorithm to estimate the surface energy
balance. Phy. Chem. Earth (B), vol. 25, n ° 2, pp. 147-157.
16. Schneider J.L (2001). Geological, archeology, hydrogeology. BRGM, vol. 1, 462p.
17. SDEA (2003). Water resources and the environment. HCNE-MEE-PNUD-DAES, 158p. Schneider J.L (2001).
Geological, archeology, hydrogeology. BRGM, vol. 1, 462p.
18. IUCN (2016). Ennedi Massif: natural and cultural landscape (Chad). Evaluation report, ID 1475, pp. 105-113.
19. Wolf J.P (1964). Geological map of the Republic of Chad at 1: 1,500,000. BRGM.
20. Yao A.B, Goula B. T. A, Kouadio Z. A, Kouakou K. E, KANE A, Sambou S (2012). Analysis of climatic variability and
quantification of water resources in the humid tropical zone: the case of the Lobo watershed in the west-west of
the Ivory Coast. Rev. Isee. Sci. Technol., 19 (2012) 136 - 157.

CITE THIS ARTICLE

Miradj Habib Djafar, Abderamane Hamit and Marie-louise vogt. Climate variability and hydrogeological responses in
the mountainous area of Ennedi (Chad) using remote sensing. Res. J. Chem. Env. Sci. Vol 7 [3] June 2019. 35-51

RJCES Vol 7 [3] June 2019 51 | P a g e © 2019 AELS, INDIA