Hydroclimatic variability and flood risk on Naglanou and Akissa forests areas in Mono River Delta ( West Africa )

This study aims to analyze hydroclimatic variation (meaning rainfall and flow decrease or increase, regime irregularity, ...) in Mono River basin and flood risk for ecological conservation of Naglanou (Benin) and Akissa (Togo) hydrosystems. Climate, hydrology and planimetry data were used for descriptive statistical and spatial interpolation to determine rain/flow relationship, climate balance (rainfall less potential evapotranspiration), flow coefficient (relationship between rainfall and flow), flood risk thresholds (base on Standardized Precipitation Index) return period. Naglanou and Akissa forest areas record heavy rainfall ranging 161 to 277 mm in June and 90 to 130 mm in October, representing respectively 16 and 13% of annual rainfall (1961-2015). This unequal spatiotemporal rainfall distribution determines surface flows and moisture of these forest sites during water level rise periods. Moreover, increase of flow rate by 20.38% over 1961-2015 and 14% over 1961-2000 linked to rainfall since 1990 and impoundment of Nangbeto dam since 1987. Flood hazard thresholds are limited (424.8 m 3 .s -1 ), moderate (609.3 m 3 .s -1 ), significant (709.1 m 3 .s -1 ), and critical (824.1 m 3 .s -1 ). Return periods correspond to 2, 10, 20 and 50 years.


INTRODUCTION
Transboundary Biosphere Reserves were areas of a sustainable natural resources management in West Africa (Davis, 1994;Doumenge et al., 2001;Fournier et al., 2007).But recent population growth and climate change harmful effects has led to deterioration of these ecosystems (Ngandjui and White, 2000;IPCC, 2001;Mengue-Medou, 2002).Indeed, the project "Transboundary Biosphere Reserve of Mono River Delta" led by GIZ under UNESCO-MAB program aims was to contribute in a sustainable development of the ecological periods,contributing to habitat and biota degradation (GIZ, 2016).Well, both of the two forest sites, on the Mono River flood plain, receive fluvial water.Thus they are characterized by hydromorphic soils and lithosoils on sandstone and types of aquatic vegetation and semi aquatic which constitute ecological habitat for a specific wild fauna.
Therefore, to face global environmental change, the project "Transboundary Biosphere Reserve on the Mono River Delta" aims to contribute to sustainable development of Naglanou (Benin) and Akissa (Togo) ecological sites."Zoning of flood-prone lands as ecological reserves or protected wetlands can often help to meet broader environmental or biodiversity goals..." (Dudley et al., 2015).To this end, it is important first to characterize climate and hydrological variation in the Mono River basin and secondary to determine flood risk for ecological conservation of the Naglanou (Benin) and Akissa (Togo) hydrosystems.

METHODOLOGY
Daily rainfall from databases of the National Meteorology Directorate (DMN) in Benin and Togo and Athiémé station flow data provided by the General Directorate of Water in Benin are used.These climate and hydrological data cover the period 1961-2015.Among the 28 hydrometeorological stations of the Mono Basin, 4 (Anèho, Grand-Popo Tabligbo and Athiémé) near the forest sites help for local level analysis.Also satellite image (Landsat TM scenes and OLI, respectively over a period of 1986 to 2015 years), topography (DTM of STRM30) and field work data were helpful for spatiotemporal analysis.

Hydroclimatic functioning analysis
Rainfall data distribution mapping by Kriging is made to better analyze spatial hydroclimatic variability on the forest sites.It involves direct estimation of rainfall as well as its variance from daily data of the 28 stations.
To better implement this type of Kriging interpolation, a variogram analysis (Hennequi, 2010) was made to explore data spatial structure and check their autocorrelation.The semivariogram which discloses semi-variance as a distance function between monitoring stations and allows spatially link data was defined as follows: is the regionalized variable studied is the vector of coordinates is the distance vector and is the geographic area studied (21,500 km²).
To characterize surface water resources of the two forest sites, a good knowledge of the hydroclimatic functioning of Mono River basin is necessary.
Rainfall/flow relationship was used to measure the link or dependence level between rains and runoff on the watershed Mono.
Estimation of the water retention capacity of the Naglanou and Akissa forest sites is by development of slope maps, river systems, determination of wet areas and their depths.The formula used to estimate the volume of mobilized water (Savané et al., 2001) is: , with: V = the volume of mobilized water (m 3 ); S = the wetted surface area (km 2 ) of the site and H = the water depth (m).
On the lowest areas where grasslands are developing, water level on both of the sites is determined from a comparison of interviews results with ecological markers of water level change (Photo 1).So, two water levels data are considered: 50 cm for less rainy years (Photo 1b) and 110 cm for exceptional rainfall years (Photo 1a).In almost normal year, seasonal variation in water levels within the limits of 50 cm is the basis of observed micro-reliefs (gilgaï) formation (Photo 1b) on Naglanou forest site.
The volume mobilized water on Naglanou and Akissa forest sites is evaluated from marshy grassland area and temporal variation in water levels on each of the sites.
These standardized indices (WMO, 2012) are calculated based on the daily and monthly rainfall for multiple time scales.According to Cancelliere and Bonaccorso (2009), the standardized index is based on an equiprobability transformation of precipitation values aggregated at k-months into standard normal values, with k generally fixed according to the purpose of the analysis (example: k = 1, 3, 6, 9, 12, 24, 36 months).
Originally, McKee et al. (1993) propose for the calculation of the SPI Gamma distribution.This index is used with no monthly or multi-monthly time.But for this work, the classic standardized index (denoted Z) was used with time scale change to daily scale.Using rainfall and flow data, it is calculated by the formula: is daily rain or flow, μ and σ are respectively the mean and standard deviation of the study series.
Detection of rainfall and hydrological thresholds of flood risks was  1).This allowed classifying daily rain and flowing data in terms of flood hazards, for different risk levels (limited, moderate, significant and critical) in the Mono River Basin.This hazards categorization has been possible due to a transposition of McKee et al. (1993) classification to use daily data.Thresholds were determined from a standard series of annual daily flows peak at Athiémé hydrometric station.Seasonal cumulative rainfall corresponded to the related years for these thresholds were calculated.Different risk thresholds, into four classes are categorized as shown in Table 2.

Return period of Gumbel method
Frequency analysis enables to determine recurrence of hydroclimatic events and to estimate their return time (Lubès and Masson, 1991;Hubert and Bendjoudi 1997;Goula et al., 2006;Bois et al., 2007).Frequency model of this prediction is an equation describing the statistical behavior of a process.The model describes occurrence probability of a hydroclimatic event.A frequency model often used to describe the statistical behavior of extreme values (Ghanmi, 2014) is the statistical Gumbel distribution (double exponential law or Gumbel).Gumbel distribution of law F (x) is expressed as follows: , where a and b are the Gumbel model parameters.The distribution is then written: This is for the purpose of estimating the probability of exceeding F (xi) to be attributed to each value xi of rainfall and hydrological series as did by Mimikou et al. (1994); Bonaccorso et al. (2003) and Totin et al. (2009) to determine return periods of normal, abnormal, exceptional (rare) or very exceptional (very rare) hydroclimatic events.
For this empirical frequency Hazen: where r is the rank in the series of classified data by increasing values, n the sample size and x [r] the rank r value.
The return period T of a rainy or hydrologic event is defined as the inverse of the probability of the event frequency occurrence.

Rainfall and hydrological variability in the catchment of Mono River at Athieme
Two types of years are distinguished in Mono Delta basin: a wet year with five or six months (May to October) with surplus rainfall ranging from 0.5 to 49.3 mm and a dry year with two wet months (May-June) with surplus water range from 9.95 to 93.57mm.Reduction of wet sequences affects the rain/flow relationship which is illustrated by the Figure 2 at Athiémé hydrometric station at the daily time scale.Peak flows are recorded with a shift in precipitation optimum.Years of heavy rains generate abundant flows and therefore heavy flooding in Naglanou and Akissa forest areas.In addition, in dry years, marshes of the Mono Delta are also submerged following the of water release from Nangbéto dam as told by 95% of the surveyed populations.Dam water releases improve flow and humidify hydromorphic soil of Naglanou and Akissa forest.Thus soil profiles in a state of permanent water saturation, the first rainfall events cause

Hydrological incidence of Nangbeto Dam
Flow coefficient reflects changes on rainfall, role of geological bedrock and impact of vegetation cover (Mounier et al., 1993;Mahé et al., 2000).Runoff coefficient average at Athieme station is 20.38% over 1961-2015.
Analysis of monthly variation of flow coefficient before and after Nangbéto dam functioning (Figure 3) confirms water increase in recent years on the lower valley, but especially influence of dam water releases.

Water reserves on Naglanou and Akissa forest sites
Figure 4 shows water accumulation levels and its distribution on Naglanou and Akissa forest areas.Assessment of water volume on Naglanou and Akissa forests need first assessment of the area of marshy grassland area for the years 1986 and 2015.Swampy formations areas have increased from 64.5 to 77.4% in Akissa and 35.9 to 66.9% in Naglanou.Table 3 illustrates water volume to be mobilized on the Naglanou and Akissa sites for the two different water levels.
In the dry periods, with substantially 50 cm water height, only the marshy grassland areas are flooded.On the other hand, in the very wet period (water level 110 cm), savannas located more on the exposed land is flooded following the grasslands.
Spatiotemporal variation rate of water volume changes by 0.03 to 86.76% on the forest sites.This variation is more marked at level of 50 cm water depth (from 20.48 to 86.76%) than that of 110 cm (from 0.03 to 0.49%).This low change rate in exceptional rainy year nevertheless shows that these extreme events still occur, but are rare in time.Thus, in recent years, as in the Mono Valley the sites recorded a slight recovery of hydroclimatic events that flood regularly savannas.But water inflows are greater at Naglanou than Akissa.

Flood risks in the forest sites
Quantitative and qualitative assessment of flood risks on Naglanou and Akissa areas helped in its categorization and deduction of its consequences.

Characterization of hydrometeorological hazards of flood risks
Characterized flood hazards on Naglanou and Akissa forest sites are extremes rainfall and flow which could lead to the hydro-climatic disaster.
The annual daily peak rates at Athieme from 1961 to 2015 (Figure 5) varies from 90 to 951 m 3 /s.These help to identify the various risks of floods thresholds in Naglanou and Akissa sites based on the classification of McKee et al. (1993), implemented in the daily time.Flood hazards showed that 78% of flow annual peak rates are recorded from September to October.This is consistent with the rainy months (May to October) which concentrate more of 75% of seasonal rainfall (Houndénou, 1999;Totin et al., 2016)  Levels of green and yellow vigilance correspond to hydroclimatic events without major influence on ecological functioning of the forest sites.Moreover, levels of orange and red vigilance refer to the rare frequencies events but catastrophic for biodiversity.In consequence, it is important to establish an early warning system for efficient management of hydroclimatic crises on Naglanou and Akissa forest sites by also focusing on hazards return periods.

Return periods of flood hazards in Naglanou and Akissa forest sites
Annual daily peak rates on the Mono Basin at Athieme and seasonal cumulative rainfall (May to October) are presented in Table 4.
The rare frequency rates (≥ 25 years or = 50 years) for wet and very wet recurrences calculated over different periods on the Mono Basin are respectively 1077.5 and 1203.8 m 3 .s-1 . For normal or abnormal events of frequency ≥ 2 years or 5 years, or 10 years or 20 years, flow rates were 568.9, 772.5, 907.3 and 1036.5 m 3 .s-1 , respectively.
On the Mono River Delta, high flow rates are due to total seasonal rainfall.Thus, rare frequency flows are generated by total seasonal rain: 1056.4 mm and 1152.4 mm for ≥ 25 years or = 50 years return period.On the other hand, recurrent rain events (≥ 2 years or 5 years) or of 10 or 20 years frequency are respectively 669.9 mm, On the Naglanou and Akissa forest sites scale, rare frequency total seasonal rainfall (≥ 25 years or = 50 years) which can lead to exceptional hydrological events are respectively 1072.8 to 1210.7 mm and 1167.9 to 1346.5 mm.Seasonal cumulative rainfall are 636.3 to 689.6 mm for 2 years return period; 812.0 to 893.6 mm for 5 years"; 928.4 to 1016.6 mm for 10 years" and from 1040.0 to 1160.4 mm for 20 years".
In comparison, the years 1999, 2010 hydro-climatic events, in the Mono Basin, are comparable to those of 1963, 1968, but less intense.This confirms to certain extent a certainty of exceptional hydrological events return periods on the basin and in particular the two forest sites.
These frequent events of hydro-climatic disasters could be due not only to slight increase in rainfall but certainly also to combined effect of water releases from Nangbéto dam and vegetation cover degradation of in the basin.

DISCUSSION
In the current context of hydroloclimatic variation, it is important to quantify the impact on the hydrosystems.Thus, this study investigated climate and hydrological variation in the Mono River basin and to determine flood risk for ecological conservation of the Naglanou (Benin) and Akissa (Togo) hydrosystems.Relationship between rainfall and flow is perfect (r = 0.64) on the Mono River basin of 1961-2010.This flow dependence on rainfall is also highlight in Niger River basin (0.81) by Vissin (2007) on 1955-1992and 0.65 by Babatolu and Akinnubi (2014) on 1955-2010. But, in the White Volta River basin at Pwalugu, this correlation coefficient is 0.78 over the period 2003-2009(Kasei et al., 2013) ) on due to largely dependence of surface runoff on rainfall of the basin.Thus, these peaks of hydrological flow succeed the rainfall peak with a two to three lag month corresponding to the basin response time and soil horizons saturation.This corroborates the results of Totin et al. (2016) on the Ouémé River basin at Bonou.Nevertheless, the correlation gap could be linked land use land cover (Mahé, 2006;Tramblay et al., 2014;Roudier et al., 2014).In recent decades West Africa has seen major changes in land use, with strong impacts on runoff (Wittig et al., 2007).
These rainfall variations have led to strong fluctuations in river discharge with a generally negative trend from 1960 to 2010 (Descroix et al., 2013), especially in Sudanian areas.In Guinean areas the decrease has been more moderate.Mahe et al. (2013) underlined the nonlinear effect of this rainfall drop over much of West Africa, with a −20% decrease in rainfall resulting in a decrease of −60% in runoff.
Runoff coefficient average at Athieme station is 20.38% over 1961-2015 against 14% over the period 1961-2000 (Amoussou et al., 2012).Increasing flow coefficient is the direct result of improved rainfall on the decades 1990 and 2000, degradation of vegetation cover (Mahe and Olivry, 1995;Dhakal et al., 2012;Tramblay et al., 2014) increase of evapotranspiration, rising of the subsurface aquifer and presence of sandstone covered by hydromorphic soils with low retention capacity.
Monthly flow coefficient which was 2.3% (1961 to 1987) has almost increased (4.4%) between the years 1988 and 2015 mainly confirming significant contributions of Nangbéto dam in dry periods as already reported by Blivi and Rossi (1995) and Klassou (1996).This is an advantage in terms of water balance for Naglanou and Akissa areas biodiversity.
Studies concerning climate variability in the Mono River watershed revealed a slight increase of precipitation in the period after 1990 (Amoussou et al., 2012;Houéssou et al., 2015).Cumulative seasonal rainfall (May to October) is the mean factors of flood risk on the areas of Naglanou and Akissa.
Floodplain connectivity indicates a broad range of ecological functions including the current and potential floodplain connectivity indicates a broad range of ecological functions including the current and potential capacity of floodplains to store and convey floods, transport sediment and wood, retain and transform water pollutants, support forest ecosystems, and provide habitats for aquatic species, ….In addition, flooding poses a risk to many facilities located on floodplain including those that provide vital services (Konrad, 2015).Thus, frequent floods are read preferable than eventual floods for a better conservation of the ecosystems.Then, the lower return period (2 to 5 years) is desirable than the maximum return period (10 to 20 years), due to its almost permanent moisture role in the biodiversity conservation on the Naglanou and Akissa forests.

Conclusion
It appears that water availability in Naglanou Akissa forest sites depends on temporal distribution of rainfall, Mono River flows and its tributaries during periods of high water especially floods and dam Nangbéto water releases.The latter contributed to 97% of runoff in time of base flow in the basin and therefore to soils moistening.
Flow coefficient increase from 20. 38% (1961-2015) to 14%  confirms rainfall increase since the end of the 1990s and Nangbéto dam water releases in the basin.Morphological, pedo-geological and hydro-climatic factors induced large water accumulation on Naglanou and Akissa forest areas.So water reserve varies from 0.03 to 97% depending on the seasons, but more pronounced in the wet" period (from 20.48 to 97%).Forest areas have a high water retention capacity.This is a great advantage for the conservation of aquatic and semi-aquatic biodiversity on these areas.
Characterization of flood hazards on Naglanou and Akissa areas is determined from the categorization of rainfall and hydrometric thresholds.).Significant and critical thresholds for orange and red warning levels associated with rare frequency events are sometimes catastrophic for a biodiversity.
Return periods of flood hazards are at least 20 to 50 years for rare and very rare events, but more disastrous than events of a range from 2 to 10 years frequency.These cases are the most desired events for the protection of ecological habitat and aquatic biota.
Facing the Naglanou and Akissa forest areas vulnerability to floods, the subject of the next step will focus on development of an early warning system for hydro-climatic crises efficient management and better ecological monitoring.

Figure 2 .
Figure 2. Monthly rainfall / flow relationship on the lower valley at Athiémé.

Figure 3 .
Figure 3. Monthly variation of flow coefficient in the Mono catchment at Athieme.

Figure 4 .
Figure 4. Level and degree of water accumulation on Naglanou and Akissa forest areas.

Figure 5 .
Figure 5.Rates of annual daily flow peak and flood risks thresholds in Naglanou and Akissa forest sites.
in the Sudanian area where rises the main River that feeds Naglanou and Akissa sites.Flows thresholds are 424.

Table 1 .
McKee et al. (1993)PI values, categories and flood risk levels.fromMcKeeetal.(1993)andWMO(2012)andcompleted.NB:In the case of work, the operation of Table1is made on the values ranging from 0 to 2 and more. Extract

Table 2 .
Correspondence between standardized indices values and flood risk levels.
from the categorization of SPI / SFI values (Table

Table 3 .
Volume of mobilized water (m 3 ) on Naglanou and Akissa forest sites.
Significant risk threshold (Yellow alert)Moderate risk threshold (Orange alert) Limited risk threshold (Green alert)

Table 4 .
Return periods of seasonal rainfall hazards (May to October) and annual daily flow peak of flood risks on the Mono Basin.