Abstract
Stable isotopes in precipitation and vapor are a powerful tool for tracing the origin of moisture and mixing processes. This paper discusses time and space variation of δ18O in precipitation and controlling features over upper Blue Nile Basin using data from GNIP, observed data in 2014 and simulated data by AGCM. IsoGSM simulation in precipitation was verified with observation. The δ18O variation shows clear seasonality with the lowest18O values in August and dry season, and enriched18O in spring, June and September. Spring sample is enriched compared to summer, and assumed to be related with moisture sources. More enriched isotopes in spring and lower d-excess could be related to the source of air masses in short travel path from North Indian Ocean, Mediterranean and Red sea while summer rain is depleted with larger d-excess could be related to longer travel path of moisture from south Indian Ocean with mixing of potential evaporated moisture from open surface and transpired moisture from Congo vegetation and also from Gulf of Guinea. The isotopic statistics of three stations shows maximum, minimum and average value of (8.23‰, -11.73‰, 0.04‰) in Addis Ababa, (5.26‰, -12.74‰, and -2.52‰) in Entoto Hill and (4.08‰, -9.65‰, 2.41‰) in Debremarkos respectively. The δ2H- δ18O relationships, monthly weighted d-excess variation in the Basin revealed the temporal variation of δ18O in precipitation is essentially shaped by the source of the moisture and spatial differences is due to Rayleigh rainout effect along the moisture trajectory. The source of moisture is primarily controlled by the north south movement of ITCZ within the Basin. The study recommends the use of model simulated δ18O as good alternative for hydrological and hydrologeological investigations when needed.
Keywords: IsoGSM; Deuterium Excess; Upper Blue Nile; δ2H, δ18O, moisture source
Introduction
The stable water isotopes (δ18O and δ2H) are important tracers of the source of moisture and mixing processes (Bonne et al., 2014; Dansgaard, 1964; Delattre et al., 2015; Kebede and Travi 2012; Yoshimura et al., 2004), largely used to reconstruct continental palaeoclimates (Joussaume et al., 1984; Dansgaard et al., 1969) and used as a natural tracer for hydrological cycles (Dansgaard, 1964; Jouzel, 2003). Oxygen isotopes (δ18O) in precipitation are strongly influenced by the water source and trajectory history and its variation could be used to reconstruct the past atmospheric circulation (Burnett et al., 2004; Okazaki et al., 2015). The knowledge of spatial and temporal distribution of the stable water isotopes (δ18O and δ2H) in a big river Basin is useful for investigating hydrological, hydrogeological and paleoclimate information.
Recently, numerous hydrological and hydrogelogical applications rely on water isotope tracers to find out the geospatial origin of water, the type of geological materials, the mixing of geological materials with water. However, mapping the isotopic distribution of rainwater and/or surface water large River Basins are difficult due to lack of finer resolution data in space and time. The monthly weighted isotopic data from some GNIP stations only is an accessible for different isotopic studies now. Addis Ababa station is the only IAEA/WMO station with monthly weighted isotopic data of precipitation near to the upper Blue Nile River Basin. Other stations have only short -term rainfall isotopes were occasionally measured. Identifying isotopes in meteoric water of large spatial coverage in the world and big river Basin like the upper Blue Nile River is difficult. A simple interpolation or statistical regression methods was used in some areas (e.g. Bowen, 2003) to estimate isotopic distributions from the available information to understand the spatial information.
Mapping and modeling actual isotopes in meteoric waters (and paleoclimate) is useful for hydrological studies due to the growing ability and potential of modeling isotopic GCMs and the increasing availability of data. Recently, there is a focus that reflects a growing appreciation of unique information that water isotope data contribute to the analysis and understanding of hydrological cycle which is difficult in other techniques. Simulation of isotopes in precipitation by isotopic GCMs was first emerged during 1980s by coarse version of models (Joussaume et al., 1984; Jouzel et al., 1987) and subsequent developments have included sensitivity studies and simulations of isotopes. A variety of current isotope incorporated GCMs (Hoffmann et al., 1998; Mathieu et al., 2002; Noone and Simmonds, 2002; Schmidt et al., 2005; Yoshimura et al., 2008; Tindall et al., 2009; Risi et al., 2010) incorporate the complex fractionation processes during water phase changes in large scale air masses transport and provide three dimensional isotopic distributions which could be a good alternative to understand the spatial and temporal distribution of isotopes in large river Basins. However, AGCMs (which are only forced by the observed sea surface temperature) may not reproduce short term isotopic variability and limited spatial resolution and the imperfect representation of atmospheric circulation (Hoffmann et al., 2000).
The deuterium excess in water, defined as δD-8*δ18O (Dansgaard 1964) is able to study and understand water movement in the atmosphere–biosphere–lithosphere interfaces (Gat et al., 2000; Gat and Airey, 2006;), diagnostic tool for recognizing the contribution of evaporated moisture into the atmospheric water (Cui et al., 2009; Gat and Matsui, 1991; Ingraham and Taylor, 1991, Gat et al., 1994) and used to identify the origin of vapor source in precipitation (Rozanski et al., 1992). It can fingerprint evaporation more reliably than the independent stable water isotopes (δD and δ18O) variation as the second may sometimes lead to wrong conclusion that no evaporation has occurred (Barnes and Turner, 2000). Accumulation of recycled moisture into atmospheric water through evapotranspiration adds a considerable role in the hydrological cycle (Matsui et al., 1983; Rindsberger et al., 1983). The mixing of water from open surface water evaporation and transpired moisture from plants in the atmospheric air masses characterizes with different isotopic nature. Transpiration returns the precipitated waters essentially unfractionated to the atmosphere that does not change in the d- excess value and resulting atmospheric water becomes enriched in heavy isotopes and evaporation from an open water surface fractionates the isotopes of hydrogen and oxygen will result in a higher d-excess value than 10‰ (Cui et al., 2009; Dansgaard 1964; Gat et al., 1994). The d-excess values for continental precipitation events issued from oceanic moisture are close to 10‰ (Craig, 1961; Dansgaard, 1964). Variation of isotopic composition of Meteoric water depends on fractionation during phase changes from vapor to liquid or ice and vice versa (Dansgaard, 1964). The continental loss of moisture content by condensation of vapor during advection of air mass as it moves away from its source is dominant factors of fractionation in hydrological cycle (Yurtsever, 1975). However, the influence of mixing of evaporated water from open water bodies and/or transpired moisture from plants (Gat and Matsui, 1991), and storm trajectory (Friedman et al., 1992; Lawrence et al., 1982) was demonstrated to change isotopic signature of precipitation events. The seasonal oscillation of δ18O and δ2H in precipitation reflects the evaporative effect in rain droplets enhanced by meteorological conditions, the rain out effect through the moisture trajectory or the mixing of different air masses in moisture trajectories. The variations of δ2H and δ18O of Meteoric water have been explained with the Global Meteoric Water Line since (Craig 1961).
Global distribution of mean annual δ18O interpolated from Global Network of Isotopes in precipitation (GNIP) stations (Bowen and Wilikinson, 2002) shows more enriched δ18O in tropics and particularly in east Africa and Ethiopia. The high enrichment of δ18O in rainwater and the larger d-excess value in Addis Ababa first recognized by (Dansgaard, 1964) and related it with partial recycling of rain droplets and this irregular nature of stable water isotopes in Addis Ababa precipitation was also acknowledged by (Sonntag et al., 1979; Joseph et al., 1992; Rozanski et al., 1993; Darling and Gizaw, 2002; Hailemichael et al., 2002; Levin et al., 2009). The reason of high enrichment of precipitation isotopes is due to the effect of contribution of continental moisture in the sources of rainfall trajectory and which could be from the Congo vegetation (Sonntag et al., 1979; Rozanski et al., 1993; Kebede 2004 and Levin et al., 2009). More recently, Kebede and Travi 2012 broadly disclosed that δ18O anomaly is a real anomaly which is attached to sources of moisture and local meteorological processes with widespread data sets of meteoric waters in Ethiopia.
This study describes the temporal variation of stable water isotopes (δ2H and δ18O) in precipitation, their spatial variability signal and controlling features of their variation with time over upper Blue Nile Basin of Ethiopia. The stable water isotopes (δ2H and δ18O) data from observed precipitation isotopes in the upper Blue Nile River Basin, Global monitoring data of water isotopes from Global Network for Isotopes in Precipitation (GNIP) in Addis Ababa stations and Isotope-incorporated Global Spectral Model (IsoGSM) (Yoshimura et al., 2008) simulations in observational stations was used.
Precipitation samples were collected in three stations (Addis Ababa, Entoto Hill and Debremarkos) and analyzed by Liquid Water Isotope Analyzer in the school of Earth Sciences laboratory of Addis Ababa University, Ethiopia and the GNIP data of IAEA/WMO station rainfall isotope record at Addis Ababa (2360 masl) since 1961 has been used. The IsoGSM was dynamically downscaled and interpolated using 0.125° spatial resolution and a forcing time step of 6 hours in the upper Blue Nile River Basin. The 6 hourly simulations of precipitation δ18O in the grid point of two observational stations (Addis Ababa and Debremarkos) in the year 2014 were converted to daily and monthly δ18O. The daily and monthly variability of δ18O, the scatter relationship between δ2H and δ18O relationships and Deuterium- Excess variations was used to realize the features controlling temporal and spatial variation of heavy isotope composition in precipitation. The daily and monthly reproducibility of precipitation isotopes by IsoGSM were verified by observational samples in two stations (Addis Ababa and Debremarkos) in 2014 and Entoto Hill observational samples were used only for assessing spatial variations with the other two stations and also observing an altitude effect with Addis Ababa. Addis Ababa station and Entoto are near each other except variation in altitude.
IsoGSM is an isotope incorporated AGCM with the spectral nudging techniques was applied toward the atmospheric analysis using the NCEP/DOE Reanalysis (Kanamitsu et al., 2002). Spectral nudging techniques were believed to better represent large circulation fields as the short term isotopic variation in precipitation is more controlled by large scale moisture transport (Yoshimura et al., 2003; Yoshimura et al., 2004). The IsoGSM was evaluated with satellite data (Yoshimura et al., 2011) and observational data (Okazaki et al., 2015) and showed good performance with little inconsistency which could be related to the assumption of no land surface fractionation. The synthesis was made based on combining observed rain isotopes, IsoGSM model simulations and GNIP data. The temporal and spatial variation of δ18O in the study area distinguishes the dominant atmospheric processes that affect the isotopic composition of precipitation locally.
Data and Methods
Study Area and General Weather Conditions
Precipitation events analyzed were collected at three sites (Addis Ababa: 9.010N, 38.730E, altitude: 2355m; Entoto Hill: 9.120N, 38.770E, altitude: 2634m and Debremarkos: 10.340N, 37.720E, 2446m) during rainy season of 2014. In total of 139 rainfall events were collected in Addis Ababa (Mar 13, 2014 - Oct 10, 2014), 82 events in Entoto Hill (Jul 12, 2014 - Oct 12, 2014) and 80 events in Debremarkos (Jul 12, 2014 - Oct 12, 2014) were collected in the upper Blue Nile River Basin. The location of sampling site is shown in Fig. 1. Precipitation events were collected within 24 hrs intervals using a standard rain collector which prevents evaporation once the rain passed through the tube. The volume of 24 hours composite precipitation sample was transferred into 1.5 mL glass bottles with tight cap, and then the rain collector was dried with a clean dry cloth before returning it to its collecting position. Collected samples were kept in a cool and dark place. Upper Blue Nile River is upstream of part of Nile River which starts flow in northern Ethiopia to 1,400 km to the Sudanese capital Khartoum from Lake Tana 1,800 meters above sea level in the Ethiopian Mountains. Precipitation occurs in summer season (June through September), in the mid spring season (February or March through May) and the remaining precipitation occurs in the dry season (October through January or February). More than 80% of annual flow in the Blue Nile Basin results from the summer season and is concentrated between July and October as shown in Fig. 1(c). The average annual rainfall varies between 1200 and 1800 mm/yr ranging from an average of about 1000 mm/year near the Ethiopia/Sudan border to1400 mm/yr in the upper part of the basin (Conway 2000).
The movement of Inter Tropical Convergence Zone (ITCZ) is controlling the flow of the monsoon rainfall and seasonal rainfall cycle in the Northern Ethiopia (Camberlin, 1997). It oscillates an extreme northward of 150N in July and an extreme southward of 150S in January along with the zones of maximum of low pressure and maximum tem­perature in Africa annually. ITCZ activates southwesterly and southerly moisture trajectories during summer (JJAS) and moving northward crossing Ethiopia in spring (MAM) and initiating northeasterly and easterly wind trajectories. In summer, the air masses are usually traveling from the Indian Ocean, the Gulf of Guinea and the region to the north of Ethiopia and converge above the Ethiopian high­lands (Korecha and Barnston 2007; Segele and Lamb, 2009; Viste and Sorteberg, 2011). The Lagrangian method of moisture identification (Viste and Sorteberg, 2012) showed westerly/easterly trajectories to the west and southwest of Ethiopia are the dominant and reliable moisture transport into the northern Ethiopian highlands with moisture from Congo Basin, the Gulf of Guinea and from the Indian Ocean. Westerly anomalies above Central Africa increase the amount of moisture entering Ethiopia with air coming from the Gulf of Guinea, while easterly anomalies reduce this contribution.
Moisture identification during July and August demonstrated a variety of moisture trajectories sources: flow from the Gulf of Guinea, flow from the Indian Ocean, and flow from the Mediterranean region across the Red Sea and the Arabian Peninsula and air coming from the south. The air entering the northern Ethiopian highlands through the African continent to the south comes from the Indian Ocean, regardless of its southwesterly entry direction suggesting the source from Atlantic ocean in July and August, and while the moisture transport is reduced from Central Africa whether from an Atlantic direction or Indian Ocean in June and September. In a spring a source of air masses is from the Northern Indian Ocean and either from Mediterranean and/or red sea. Between October and March the ITCZ is located in south of Ethiopia and the region is under influence of northerly flow of dry and cold air from the Arabian continent. The atmospheric condition affecting the transport from the north are less well understood, but may be as relevant as those in the southwest and also the contribution and importance of moisture transport from the locally evaporated moisture (Viste and Sorteberg, 2012).
Rainfall amount and Isotope Data Sources
Temporal and spatial variation of stable water isotopes (δ18O and δ2H) were used from observational samples of precipitation in 2014, data from the International Atomic Energy Agency-Global Network of Isotopes in Precipitation (IAEA-GNIP) for Addis Ababa (1961-2009) and IsoGSM simulations. Long-term measurements of available precipitation amount in observational stations were obtained from National Meteorological Agency of Ethiopia.
Laboratory Analysis
Oxygen-18 and deuterium measurements of observational precipitations have been carried out by LGR Laser Isotope Instrument of Stable isotopes in the laboratory of School of Earth Sciences, Addis Ababa University. Isotopic analyses were reported in δ units in per mill relative to Vienna Standard Mean ocean water (VSMOW) according to the following equation:
δ (‰) = ( \(\frac{\text{Rsample}}{\text{Rvsmow}}\) – 1) X 103 Where R refers to18O/16O or2H/1H ratios. The analytical error is ± 0.2 ‰ for 18O, and ± 2 ‰ for2H.
Isotope-enabled Global Spectral Model (IsoGSM)
IsoGSM is isotope enabled the spectrally nudged atmospheric GCM toward wind and temperature fields from the National Centers for Environmental Prediction (NCEP)–Department of Energy (DOE) Reanalysis 2 (R2) (Kanamitsu et al., 2002) in addition to being forced by sea surface temperature (SST). Spectral nudging technique with reanalysis datasets (e.g., horizontal winds from NCEP reanalysis) has been applied to some AGCMs to constrain large-scale atmospheric circulations and the dynamic fields were close to observations.
The model uses T62 horizontal resolution (about 200 km) and 28 vertical levels, and the temporal resolution of the output is 6 hourly. The convection scheme is the relaxed Arakawa–Schubert scheme (Moorthi and Suarez, 1992). Nudging the model winds by ERA-40 reanalysis improved the simulation of temperature and 𝛿18O at middle latitudes (Risi et al., 2012) and the simulated isotopic fields are also more accurate compared to observations from small to large scale time scales. The model reproducibility for daily to interannual timescales was evaluated by comparing it with the precipitation isotope ratio and the vapor isotopes ratio by (Yoshimura et al., 2008) and from satellite measurements (Yoshimura et al., 2011).
For this study, the IsoGSM was dynamically downscaled and interpolated using 0.125° spatial resolution and a forcing time step of 6 hours in the upper Blue Nile River Basin. The 6 hourly simulations of atmospheric precipitation were converted to daily and monthly δ18O for Addis Ababa and Debremarkos for evaluation with observation. Isotope processes were incorporated in the IsoGSM based on (Joussaume et al., 1984): isotopic fractionation takes place whenever phase transition occurs. Most fractionation can be assumed to occur at thermodynamic equilibrium, except for three particular cases: surface evaporation form open water, condensation from vapor to ice in supersaturation conditions under -200C, and evaporation and isotopic exchange from liquid raindrop into unsaturated air. The model assumes no isotopic fractionation of evapotranspiration fluxes from the land surface (Yoshimura et al., 2008).
  1. Results and discussions
  2. The temporal variations δ18O
  3. Monthly variability of δ18O
The monthly variations of 18O contents of precipitation collected at Addis Ababa, IsoGSM simulation and amounted weighted δ18O from GNIP data in Addis Ababa (Fig. 2) and monthly variations of 18O contents of precipitation in Debremarkos (Fig. 3) is given. The bar graphs in both graphs represent the monthly average of rainfall during observational sampling time (2014) and long year mean (1964-2005). The association between monthly weighted observed δ18O with model simulated (Fig. 2 and Fig. 3) shows suitability of model simulation in capturing seasonal pattern of precipitation isotopes in two stations with slight underestimation. The simulated δ18O demonstrates better suitability and small gap in Debremarkos compared with Addis Ababa except in September. A strong correlation of monthly weighted simulation against observation for existing δ18O in Addis Ababa (0.88) and Debremarkos (0.6) was found. A possible reason of a small bias in simulation could be related the IsoGSM incapability of capturing the local recycling effects as IsoGSM assumes no isotopic fractionation over the land surface. Even if the results of the simulation show slight gap in peak enriched isotopes of June and September and underestimation in Addis Ababa precipitation, it captures the monthly-scale variability reasonably well. This shows that the model is a good alternative to use for hydrological and hydrogeological investigation with minor bias adjustment which could be considered to be land surface fractionation.
The annual rainfall starts either in February/March and stops in October with two rainy seasons; small rains in spring (MAM) and huge rainfall in summer (JJAS) in the Upper Blue Nile River Basin. The rainfall pattern shows high amount of rainfall in July, August and September in the Basin. The general pattern of monthly weighted variation of observed and simulated δ18O in Addis Ababa (Fig. 2) and Debremarkos (Fig. 3) show seasonal variation for both simulation and observation with a relatively more enriched δ18O in June and September and more depleted at the heart of the great summer rainy season (August) and dry season. The annual heavy oxygen isotopes pattern in Addis Ababa shows slight increasing in 18O values from onset of rainfall in March until June and declining its composition until the lowest values in August, and then increasing in heavy oxygen composition until the end of rainfall in October and declining its heavy oxygen composition in dry season which starts in November. The simulated δ18O (May – September, 2014) and observed δ18O (July – October, 2014) in Debremarkos shows as similar pattern as Addis Ababa except the depletion of heavy oxygen isotopes starts in October. Such a clear correspondence between the degree of isotope enrichment in spring (FMAM) and depletion of precipitation progressively in summer (JJAS) rainy season reflects presumably the influence of the south north movement of ITCZ. The south north movement of ITCZ may trigger the variability of the wind direction which brings moisture either in the northern hemisphere: Mediterranean Sea/ red sea or Indian Ocean with time.
High enrichment of stable water isotopes in the area could be related to the mixing of transpired moisture from Congo Basin which is a relatively enriched and not fractionated than oceanic moisture sources. The monthly weighted δ18O variation of precipitation (Fig. 2 and Fig. 3) shows high amount of rainfall in July and September accompanied with high isotopic enrichment. This shows little impact of rainfall amount in isotopic signature. Isotopic signature in annual cycle is governed by the mixing of moisture sources and variation of moisture trajectories accompanied with the movement of ITCZ. The δ18O values in September for both Addis Ababa and Debremarkos (Table 1 and Table 2, respectively) show high amount of rainfall accompanied with more enriched 18O and low amount of rainfall accompanied with depleted 18O in October. This shows that the monthly isotopic variability in the upper Blue Nile River Basin is not related with the amount effect.
The reason of relative enrichment of spring rain with respect to summer could be the moisture travel path. In spring, ITCZ is moving northward crossing Ethiopia and initiating northeasterly and easterly wind trajectories in which moisture travels shorter distance from North Indian Ocean, moisture from the North either Mediterranean or red sea. In summer, ITCZ is in Northern hemisphere and trigger southwesterly and southerly moisture trajectories which usually bring the air masses traveling longer distance from the south Indian Ocean, the Gulf of Guinea and the Congo Basin and converge above the Ethiopian high­lands. The relative depletion of 18O could be related to degrees of rainout as ordered by the temperature along the air mass trajectory. It is controlled by the source of moistures and the mixing processes in the moisture trajectory which varies with season. The south north movement of ITCZ is controlling the progressive increase of flow of air masses and seasonal moisture sources in the Northern Ethiopia. The heavy oxygen isotope 18O composition is relatively enriched in spring rainfall than summer as shown in (Fig. 4, Table 2 and Table 3). The amount weighted mean of δ18O from GNIP data (1961 – 2009) as plotted (Fig. 4) shows a relative enrichment of the heavy oxygen isotope, 18O in spring (FMAM) than summer (JJAS) in Addis Ababa except in 1987. This could be related to the difference of moisture with seasons in addition to the low amount of rainfall during spring. This results in a progressive depletion of heavy isotopes from spring to summer particularly (June to August) with respect to the movement of ITCZ and the lowest 18O values being observed in August and slightly enriching in September and October. The moisture travels long distance from south Indian Ocean and mixes with moisture from Gulf of Guinea and the Congo Basin.
Daily variability of δ18O
The validation of daily IsoGSM reproducibility of δ18O based on their grid points vs. observational sample of precipitation isotopes in Addis Ababa (March 2014 to Oct 2014) (Fig. 5) and in Debremarkos (July 2014 to Oct 2014 in Debremarkos) (Fig. 6) is given. The IsoGSM reproduced reasonable δ18O and well captured the seasonal variation of isotopes in both Addis Ababa and Debremarkos stations with slight underestimation of isotopic enrichment. The daily variation shows high enrichment in June and September and more depleted in August in Addis Ababa station and also the simulation and observation of δ18O in Debremarkos shows similar pattern as Debremarkos. However, Addis Ababa samples show a relatively more enriched δ18O than Debremarkos samples. The maximum, minimum, amount-weighted mean values and the difference between extreme 18O in available observational data of daily precipitation and simulated δ18O of three observational stations in the upper Blue Nile Basin (March – October, 2014) is given (Table 3) shows clear seasonal variations of stable water isotopes. The δ18O is relatively more enriched in Addis Ababa (mean value of -0.45 ‰) than Entoto Hill (mean value of -2.03‰ ) and Debremarkos (mean value of -2.57 ‰).
The δ2H - δ18O relationships and Deuterium- Excess
The monthly deviations of δ18O in precipitation of Addis Ababa and Debremarkos (Fig. 2 and Fig. 3) and daily deviations (Fig. 5 and Fig. 6, respectively) could be related to the seasonal fluctuation of moisture trajectories and sources. A scatter relationship between δ2H and δ18O from observed samples of precipitation events during rainy season of 2014 was calculated. The δ2H and δ18O relationships from samples of 139 observed precipitation for Addis Ababa (δ2H = 7.83δ18O + 12.38) plotted in (Fig. 7 (a) and 80 observed precipitation for Debremarkos (δ2H = 8.23.δ18O + 21.24) plotted in (Fig. 7 (b)). Both stations show coefficients nearly 8 with large values of d-excess greater than 10. The effect of high d-values (greater 10) in rainfall and the reduced degree of evaporation from falling rain droplets (coefficients nearly 8) are the effects on both data sites. This reflects the influence of recycled continental moisture either from open water bodies or transpired moisture in both sites. The influence of recycling of moisture is more pronounced than evaporation from falling rain droplets for both locations with greater influence in Debremarkos. But relative comparison between two sites show more pronounced recycling of moisture in Debremarkos than Addis Ababa, and slight evaporative effects in Addis Ababa. The variation of these effects in relative to two sites may be explained with the Rayleigh rain out effect. The moisture from the source arrive Addis Ababa first and goes to Debremarkos. The high d-excess in Debremarkos accompanied with relative depletion of 18O in Debremarkos is due to rainout effect through the moisture trajectory from Addis Ababa to Debremarkos. More d-excess in Debremarkos is because of the mixture of more recycled local moisture throughout the moisture trajectory from Addis Ababa to Deberemarkos.
Monthly weighted mean deuterium excess (d = δ2H - 8.δ18O) in three observational stations (Table 4) and monthly weighted mean δ2H-δ18O relationships (Table 5) was calculated in observed precipitation events to understand spatial variations in the study sites. The δ2H and δ18O and d-excess variation show increasing in mixing of recycled moisture from March to July. However, the linear regression of rainfall sample of August in Addis Ababa is with coefficient of 7.84 and intercept 9.81 is nearly similar with Global Meteoric Water Line and d-excess is nearly 10.22 reveal that the moisture is nearly from oceanic source with initial condensation in Addis Ababa. However, d-excess is increasing from Addis Ababa to Entoto Hill, and Entoto Hill to Debremarkos. This shows the moisture recycling is increasing from Addis Ababa to Debremarkos in the moisture trajectory from Addis Ababa to Debremarkos with increasing d-excess and depletion of isotopes is due to Rain out effect and mixing of recycled local moisture in the trajectory. The rainfall of Addis Ababa seems an initial condensation with relatively more enriched δ18O with Debremarkos, and the mixing of continental recycling of moisture is increasing accompanied with depletion in 18O due to Rayleigh rain out effect in the moisture trajectory to Debremarkos.
A variety of moisture trajectories in the upper Blue Nile River Basin was reviewed. The trajectories of moisture sources are changing with seasons which are controlled by the north south modulation of ITCZ. The monthly fluctuation of δ18O in precipitation, the scatter relationships between δ2H and δ18O and deuterium excess variations with each month was carried out (Table 4 and Table 5) shows the mixing of different moisture sources which is changing with time. An isotopic fractionation is the reason of temporal variation of δ18O in precipitation of any location which is due to evaporation of water from ocean or land surface, rainout process during condensation in the moisture trajectory, the mixing of different moisture sources in the moisture trajectory. In total rainfall events of Addis Ababa collected in the year 2014, 30 events are from spring (March, April and May) and 109 events was from summer (June, July, August and September) and October. The δ2H–δ18O relationships of rainfall sample in Addis Ababa were calculated for spring and summer season (Fig. 8). The δ2H–δ18O relationships for spring (δ2H = 7.15.δ18O + 12.13) show relatively more influenced with evaporation of falling rain compared to summer (δ2H = 7.85.δ18O + 14.10) samples. This variation could be partially related to lesser amount of rainfall and high solar radiation during spring than summer season (also confirms the explanation by Kebede 2004). However, summer rainfall is relatively more influenced by recycling of moisture from continental surface than spring with greater d –excess and spring rainfall is better influenced by evaporative effects. This study also confirms the importance of moisture recycling demonstrated by (Seifu et al., 2012).
Monthly weighted average d-excess (Table 4) and Linear Regression between Hydrogen and Oxygen isotopes (Table 5) in the precipitation show that the effect of evaporation in the rainfall during rain drops is increasing from March to June, and addition of recycled moisture in the rainfall is relatively increasing from March to July. Evaporative effect is more pronounced in June and moisture recycling is pronounced in July. The coefficients and d-excess value from the relationship between δ2H and δ18O during August is 7.84 and 9.8 respectively. This show that, rainfall in August is not that much influenced by recycling as well as evaporation during rain drops. The rain bearing moisture in Addis Ababa during August is nearly from Oceanic source. However, it is recycling from Addis Ababa to Debremarkos as shown with greater d-excess value in Debremarkos compared to Addis Ababa and Entoto Hill. The d-excess variation also reveals the more mixing of moisture from recycled source in September and then the d-excess value drops in October.
The spatial variation signal of δ18O and Altitude Effect
The weighted mean 18O values of rain events (July 2014 to October 2014) from three sampling sites (Table 6) as a function of the inland distance (from south east to north west) in the upper Blue Nile River Basin (Fig. 1) is given. The weighted mean18O decreases from South East sampling sites (Addis Ababa) to North West sampling site (Debremarkos). This shows a relative depletion of δ18O from Addis Ababa to Entoto Hill and Entoto Hill to Debremarkos without regarding their altitude except in September and, decreasing in d-excess (as shown in Table 4) from Debremarkos to Entoto Hill and Entoto Hill to Addis Ababa. A relative depletion of weighted mean 18O from Addis Ababa to Entoto Hill could be related to both altitude effect and Rayleigh distillation effect since Entoto Hill and Addis Ababa are very close each other (less than 6 kms in distance) and 279m altitudinal difference.
Discussions
The appearance of inland gradients of 18O values likely indicates that variations in the heavy isotope contents of rain events are influenced either by different condensation mechanisms, varying sources of moisture or Rayleigh distillation effect. The heavy isotope depletion and increasing d-excess from Addis Ababa to Entoto Hill and Entoto Hill to Debremarkos could be explained by continuous removing of water vapour as rainwater by the Rayleigh distillation mechanism accompanied with local continental moisture mixing in the moisture trajectory farther during moisture transport from Addis Ababa to Debremarkos. The spatial variation signal of this study indicates that the moisture transport history and the degree of rainout in the site may be more important factors explaining monthly variations and spatial differences in δ18O. The source of local moisture is assumed to be either evapotranspiration of the Congo vegetation and/or evaporation of surface waters mixed with oceanic air masses.
This study shows the moisture bearing precipitable water in the Blue Nile River Basin is a mixture of different sources changing with seasons and months. The δ18O of rain events show the great importance of recycled moisture in the Basin which is influenced by rainout effect and a little amount effect within the Basin. The rainfall during spring is relatively affected by evaporation of rain droplets while the summer rainfall is more influenced by mixing of recycled moisture. In general, the rainfall sampled was admixture of re-evaporated and transpired moisture from surface, evaporative enrichment from falling raindrops and the local atmospheric condition with high deuterium excess record in the summer rainfall with two peaks of d-excess in July and September as also elucidated by Kebede 2004. The more enrichment of moisture in the Basin could be a return of flux of moisture from transpiration from Congo vegetation. Transpired moisture is non-fractionating, and thus when it mixes with the moisture coming from ocean sources the resultant vapor becomes enriched in heavy isotopes and injects high d-excess waters in to the clouds.
Conclusions
This study elaborates the temporal variation of stable water isotopes (δ2H and δ18O) in precipitation, their spatial variability signal and controlling features of variation with time over upper Blue Nile Basin using time series observational isotopes, GNIP data and IsoGSM simulation. IsoGSM model simulation with observed precipitation in Addis Ababa and Debremarkos shows a reasonable reproducibility of δ18O in daily and monthly weighted timescales and clear seasonal pattern of isotopes and well captured large scale circulation fields with slight bias which could be related to the significance of land surface fractionation. IsoGSM assumed no isotopic fractionation over the land surface. The model evaluation shows the importance of land surface fractionation to have more precise simulations of stable water isotopes. Simulations could be used for mapping distribution of stable isotope in the River Basin for hydrological and hydrogeological assessment and paleoclimate reconstruction.
The temporal variation shows a clear seasonal variability of δ18O in both simulated and observed precipitation in the study area. The highly enriched value of δ18O with minimum (-9.65‰) in Entoto Hill and maximum (7.79‰) in Addis Ababa was observed. Maximum, minimum and average of amounted weighted value of δ18O, respectively shows (7.79‰, -7.39‰ and -045‰) in Addis Ababa, (5.26‰ to -10.76‰, and -2.03‰) in Entoto and (4.08‰ to -9.65‰, and 2.41‰) in Debremarkos. The peak enrichments of heavy isotopes were observed in June and September and more depleted precipitation sample in August and October. The temporal variation in daily and monthly weighted variation of oxygen isotopes (δ18O) from both the model simulation and observed sample of 2014 showed an increasing of heavy oxygen (18O) composition in precipitation from March to peak enrichment in June, and then depletion starts until August and rises in September. The rainfall in dry season starts in October is relatively more depleted. The monthly analysis of d-excess showed continuous increasing from the onset of rainfall in Addis Ababa in March from 10.24 ‰ to July 15.44‰, and also high d-excess value of 20.02‰ and 22.39‰ respectively in Entoto Hill and Debremarkos was observed in July 2014. The d-excess is relatively dropped in August for all three stations and rises in September, and then drops again in October.
The monthly variability of δ18O and d-excess and the scatter relationship between δ2H and δ18O in the Blue Nile Basin of Ethiopia shows a strong influence of moisture source and trajectory history in variation of precipitation isotopes. The variation of d-excess shows the influence of recycled moisture which is increasing from March to July, and drops in August and rises in September. Recycling of moisture is pronounced with a little evaporative effect in three stations. The δ18O is relatively enriched in spring (FMAM) than summer (JJAS) precipitation. The amount effect is not much pronounced in affecting the monthly and seasonal variation of δ18O in the upper Blue Nile Basin. For example more enriched δ18O accompanied with high amount of rainfall in September is recorded for both Addis Ababa and Debremarkos. Spring (March, April and May) precipitation is relatively better affected by evaporative effect than summer (June, July, August and September) precipitation. The linear regression of δ2H versus δ18O shows that, the precipitation water is more affected by evaporation of rain droplets in June that is enhanced by low amount of rainfall.
The heavy isotopic composition of precipitation in spring is relatively enriched and low influence of recycled moisture compared to summer. During March, April and May, the ITCZ is moving northward crossing Ethiopia and the moisture in the Basin is shaped by northeasterly and easterly moisture flows which brings enriched moisture from short paths of Northern Indian Ocean, Mediterranean and Red Sea. During June, July, August and September, the ITCZ is located in northern Ethiopia and the region is under the influence of the southwesterly and south easterly monsoon flows. These flows bring moisture with varying trajectories paths from south Indian Ocean, Gulf of Guinea and Congo Basin. The depletion of moisture in this season is due to varying paths of moisture which is influenced by rain out effect. Moisture from low level trajectory from south Indian Ocean through Turkana channel pulls the transpired moisture from Congo vegetation Basin which is expected to be enriched in heavy isotopes. The monthly variation of isotopic composition in this season could be related to the variation of moisture paths from south Indian Ocean. Isotopic signature of rainfall during August show more depleted in 18O, low mixing of recycled water and evaporative effect compared to other months. The more depleted δ18O with d-excess value nearly 10 clearly show that the moisture is likely from south Indian Ocean through the air parcels traveling long pathways to the west before turning northeast­ward. The more depletion of water is due to rain out effect in the trajectory of moisture. Between October and March the ITCZ is located in south of Ethiopia and the region is under influence of northerly flow of dry and cold air which is more depleted water sourced from the Arabian continent and extra-tropical air masses.
More enriched δ18O accompanied with d-excess (greater than 10) in the upper Blue Nile River Basin could be related to mixing transpired moisture from Congo vegetation and evaporated water from open bodies through the moisture trajectory from Indian Ocean. The transpired moisture is non-fractionating, and consequently when it mixes with the moisture from ocean sources the resulting vapor becomes enriched in heavy isotopes and associated evaporation from open water bodies bring a high d-excess air masses. An association of monthly weighted isotopic composition of precipitation data from July, 2014 to October, 2014 in three sites; Addis Ababa, Entoto Hill and Debremarkos shows a relative depletion of δ18O from Addis Ababa to Entoto and Entoto to Debremarkos. The spatial variation of stable water isotopes in the Basin is controlled by the rain out effect. The depletion of isotopes from Addis Ababa to Entoto Hill and Entoto to Debremarkos accompanied with increasing value of d-excess. This shows that, moisture is transported from Addis Ababa to north of the Blue Nile River Basin and increasing of d-excess is due to rainout processes along moisture trajectory.
In general, the study shows that the temporal isotopic composition of precipitation in the upper Blue Nile River Basin is essentially shaped by the source of moisture which is controlled by the south north movement of ITCZ. The mixing of recycled continental moisture is pronounced and shown in the isotopic composition of stable water isotopes. The spatial variation of isotopic composition is influenced by Rayleigh rainout effect along the moisture trajectory.