Interactive comment on “ Observed and simulated hydroclimatology using distributed hydrologic model from in-situ and multi-satellite remote sensing datasets in Lake Victoria region in East Africa ”

The paper is well done and I hope this will add up the knowledge to understand the spatial-temporal distribution of hydro climatology conditions of East Africa using measured and satellite data as model inputs. I want to say just few things: 1. In the paper, the daily discharge is simulated at a low accuracy compared to the monthly discharge. Distributed models usually should simulate the daily discharge at a reasonable accuracy. Monthly values may not clearly show what is happening at a specific place and


Introduction
Climatologically most of East Africa is considered as a sub humid landscape that comprises arid and semi-arid regions, grasslands, savannahs, as well as a Mediterranean environment.East African climate is mainly influenced by the seasonal shift of the intertropical convergence zone (ITCZ).However other regional factors that influence the climate are topographical variations, large inland lakes, land cover/land use, as well as the proximity to the Indian Ocean.Oscillations in the ITCZ, shapes two rainy seasons in the equatorial East Africa, one from March to May and the second from October to December (Kaspar et al., 2008).This precipitation pattern can result in floods in this region with impacts on the food and agricultural security, human health, infrastructure, tourism, and other sectors.The rainy season that onsets from October through early December brings devastating floods in Uganda, Kenya, Tanzania, and other countries in East Africa almost every year.This region, surrounding Lake Victoria, is heavily populated with around thirty million people (Osano et al., 2003).These floods are a serious problem in East Africa, particularly in the Lake Victoria Basin, which impacts the livelihood of many people every year.Since the 1950's East African countries like Kenya, Uganda, and Tanzania showed an increase in population as well as unsustainable development.Due to economic pressure, much of the forest land is converted to agriculture or settlement purposes.The Lake Victoria region experiences rising demands for its depleting water resources.
Hydro-climatology deals with the interactions of climate with surface water.It recognizes that climate is the driving force of the hydrologic cycle.One of the main focuses Introduction

Conclusions References
Tables Figures

Back Close
Full of the hydro-climatic study is the interactions between precipitation, evapotranspiration, soil moisture storage, groundwater recharge, and stream flow (Shelton, 2009).
The study of the water budget at a given location and time period essentially deals with the components of hydro-climatology.Hydrologic modeling is one of the efficient and valuable approaches for understanding the relationship between climate, hydrologic cycle, and water resources.In East Africa, the current trend and future scenarios of unsustainable water resource utilization demands modeling studies that provide accurate spatial and temporal information on hydrological and climatological variables.The main obstacles for these investigations are the lack of sufficient geospatial data for distributed hydrologic model input and validation.Availability of observed data in regions with sparse ground based networks for hydrologic estimations is the key limitation in hydroclimatologic studies.Hydrologic modeling has been constrained by the difficulty in precisely estimating precipitation, the key forcing factor, over a range of spatial and temporal scales.However, advances in satellite remote sensing data can provide objective estimates on precipitation, evapotranspiration and land surface controlling factors for water budget calculations.The recently available and virtually uninterrupted supply of satellite-based rainfall estimates is becoming a cost-effective source of data for hydro climatologic investigations in many under-gauged regions around the world.
The question remains whether with the existing spatial and temporal coverage of satellite precipitation and other estimates, how can we achieve their optimal use to compute a less uncertain water budget?In this study, Coupled Routing and Excess Storage (CREST) (Wang et al., 2010) a distributed hydrologic model, is used to simulate the spatial and temporal variation of atmospheric, land surface, and subsurface water fluxes and storages.The goal of this paper is to study the hydro-climatology of Nzoia Basin, a sub catchment of the Lake Victoria region using observed and simulated data with particular emphasis on distributed hydrology of the watershed.(Fig. 1).More specifically, the objectives are to 1) quantify the hydrologic cycle of Nzoia Basin at decadal, annual, monthly and daily time scale using in-situ 21-year observational dataset; 2) model the rainfall-runoff rela-Introduction

Conclusions References
Tables Figures

Back Close
Full tionship using a distributed hydrological model, calibrated by long-term observations, in terms of predictability at the daily flood scale; 3) investigate the hydrological capability of remote sensing data (preliminary precipitation) in terms of the reconstruction of water cycle components.The use of remotely sensed spatially distributed datasets has made possible the transition from lumped to distributed hydrologic models that accounts for the spatial variability of the model parameters and inputs.
The paper follows with a brief description of the study basin, data, and model in Sect. 2. The hydroclimatology based on observational datasets are discussed in Sect.3, followed by Sect. 4 with a model set-up, calibration, and verification.The hydrological model reconstruction results are outlined in Sect.5, and finally summary and discussions are given in Sect.6.

Study area
The study area is the Nzoia River located at latitudes 34 • -36 • E and longitudes 0 • 03 -1 • 15 N in East Africa.It drains into the Lake Victoria and Nile river basins.Lake Victoria, with an area of 68 600 km 2 , is the second largest freshwater lake in the world (Swenson and Wahr, 2009).Nzoia, a sub-basin of Lake Victoria, is chosen as the study area because of its regional importance as it is a flood-prone basin and also one of the major tributaries to Lake Victoria (Fig. 1).The Nzoia sub-basin covers approximately 12 900 km 2 of area with an elevation ranging between 1100 to 3000 m.The Nzoia River originates in the southern part of the Mt.Elgon and Western slopes of Cherangani Hills (Li et al., 2009).Introduction

Conclusions References
Tables Figures

Back Close
Full

Gauged rainfall and discharge data
Daily observed rainfall data are obtained from the Africa Regional Centre for Mapping of Resources for Development (RCMRD) from 1985 to 2006 for the 12 rain gauge stations located within the Nzoia Basin.They are then interpolated to fit the model grid resolution using the Thiessen polygon method (Kopec, 1963).Also obtained are the daily discharge data (in m 3 /s) at the basin outlet for the same time period.

NASA TMPA
Precipitation is a critical forcing variable to hydrologic models, and therefore accurate measurements of precipitation on a fine space and time scale is very important for simulating land-surface hydrologic processes, and monitoring water resources, especially for semiarid regions (Sorooshian et al., 2005;Gebremichael et al., 2006).For the past decade, there have been several multi-satellite based precipitation retrieval algorithms for operational and research purposes (Hong et al., 2004;Huffman et al., 2007;Joyce et al., 2004;Sorooshian et al., 2000).For this study, we used one of the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) products, 3B42V6 given its 10+ year data availability.It was used to drive the CREST model to simulate the water budget components such as runoff, evapotranspiration and, change in storage for the study basin.The standard TMPA provides precipitation estimates from multiple satellites at a 3-hourly, 0.25 • ×0.25 • latitude-longitude resolution covering the globe between the latitude band of 50 • N-S (Huffman et al., 2007).This TRMM standard precipitation product has been widely used for hydrological applications such as flood and landslide prediction at the global and regional scope (Su et al., 2008;Hong et al., 2006;Hong et al., 2007;Yong et al., 2010).Introduction

Conclusions References
Tables Figures

Back Close
Full

Evapo-transpiration
In the model, Potential Evapotranspiration (PET) values from the global dataset based on the Famine Early Warning Systems Network.Further details on these estimates can be found at (http://earlywarning.usgs.gov/Global/product.php?image=pt).The PET are estimates of climate parameter data that is extracted from the Global Data Assimilation System (GDAS) analysis fields.

The CREST model
A distributed hydrologic model, Coupled Routing and Excess STorage (CREST) (Wang et al., 2010;Khan et al., 2010) is used to simulate the spatiotemporal variation of water fluxes and storages on regular grids.The model accounts for the most important parameters of the water balance component i.e. the infiltration and runoff generation processes.The main CREST components are briefly described as ; 1) data flow module based on cell to cell finite elements; 2) the three different layers within the soil profile that affect the maximum storage available in the soil layers.This representation within cell variability in soil moisture storage capacity (via a spatial probability distribution) and within cell routing can be employed for simulations at different spatiotemporal scales 3) coupling between the runoff generation and routing components via feedback mechanisms.This coupling allows for a scalability of the hydrological variables, such as soil moisture, and particularly important for simulations at fine spatial resolution.Further details on CREST model, can be found in Khan et al. (2010) and Wang (2010).Introduction

Conclusions References
Tables Figures

Back Close
Full The first and second maxima occurred in April-May and July-November, respectively.The maximum daily, monthly and yearly rainfall is 38, 10.87 and 5.19 mm/day, respectively (Fig. 2a-c).It is observed that for the given time period of 1985-2006, the basin average rainfall per annum was about 1500 mm.Observations of the rainfall since 1985 do not show any statistically significant trends.It is observed that half of the recorded rainfalls are below 5 mm/d and the rainfall histogram was skewed towards the lower ends (Fig. 3a and b).

Stream discharge
Discharge varies from low flows of 0.7 m 3 /s to a high of 600 m 3 /s.The highest river discharges occurred in the months of May through September while the lowest river discharges occurred in the months of January through March (Fig. 2b).From 1985-2006, the average daily discharge was 134 m 3 /s (Fig. 2a).The flow duration curve shows the average percentage of time that specific daily flows (Fig. 4a) are equaled or exceeded at Nzoia.The discharge histogram is skewed towards the lower values and more than half of the recorded daily discharges are less than 120 m 3 /s (Fig. 4b).

Return periods of rainfall and discharge
The annual peak discharge and precipitation for the given time period are shown in Fig. 5.The calculated return periods for both the discharge and rainfall are given in Table 1.The peak discharges of 1985, 1988, 1999, and 2006 were all above the 5year flow while 1985 and 1999 recorded discharges of 10-year return periods.In 1985, the recorded peak discharge was of the 100-year return period.
It is observed that the annual peak rainfall in the years 1985, 1988, 1990, 1994, 1998, and 2003

Annual mean discharge
The discharge time series provide information on the year-to-year variations of both low and peak discharges.Figure 6a shows the annual mean discharge for Nzoia River.1986, 1987, and 2002 are not the result of a single dry month but due to continuous low monthly discharges throughout the whole year (Fig. 6b).

Decadal monthly trend
The observed data are also analyzed for any trends over the past two decades: 1985-1994 (first decade) and 1995-2004 (second decade).Overall there is some decrease (−4.2%) in rainfall in the second decade compared to the first.Similarly there was a marginal increase (+2%) in discharge (Table 2, Fig. 7a and b).However, there is a more pronounced monthly variation both in rainfall and discharge.A maximum decrease in rainfall was recorded for the month of February (−55%) whereas December witnessed a maximum increase (+32%).Similarly, there is a maximum drop in stream discharge in the months of February and May (−13%) while a surge of +44% is observed in the month of January (Table 2).

Hydrologic model setup, calibration, simulation, and verification
A moderate resolution CREST model at a 30 arcsec resolution is implemented for the Nzoia Basin to retrospectively simulate the main components of water cycles with both Introduction

Conclusions References
Tables Figures

Back Close
Full in-situ and remote sensing data sets.The model is implemented over the Nzoia Basin using digital elevation data to generate flow direction, flow accumulation, and contributing basin area that are required as basic inputs to run the CREST model.The local drainage direction and accumulation are derived from the Digital Elevation processed from the Model Shuttle Radar Topography Mission (SRTM) (Rabus et al., 2003).The primary forcing datasets enabling the development of a distributed hydrological model using the long term rain gauge and observed streamflow data provided by the local authorities previously discussed in Sect.2.2.The CREST model is calibrated at the Nzoia Basin outlet (Fig. 1) for the given time period of 1985-1998.A spin up period of one year was assigned to produce reasonably realistic climate states.
The model utilizes global optimization approach to capture the parameter interactions.An auto-calibration technique based on the Adaptive Random Search (ARS) method (Brooks, 1958) was used to calibrate the CREST model.The ARS method is considered adaptive in the sense that it uses information gathered during previous iterations to decide how the simulation effort is expended in the current iteration.The two most commonly used indicators for the model calibration in order to get the best match of model-simulated streamflow with observations are the Nash-Sutcliffe Coefficient of Efficiency (NSCE) (Nash and Sutcliffe, 1970) and relative bias ratio (Bias).Therefore, these are used as objective functions for the automatic calibration.The ideal value for NSCE is 1 and Bias is 0%.Precipitation and streamflow time series corresponding to this optimized parameter combination during the calibration period are shown in Fig. 8a.CREST calibration with the optimized parameter combination yielded NSCE=0.87 and Bias=−0.23% as the final result (Fig. 8a).
The performance of CREST in discharge simulation at the drainage outlet was validated.The validation of the hydrological model was performed for the period 1999-2004.The simulation quality during the validation period is comparable, even with a decrease in model efficiency.One reason for the noise in the simulation might be due to the increase in human activities in the catchment area during the recent years.With this optimized parameter combination and model status at the last day from cal-Introduction

Conclusions References
Tables Figures

Back Close
Full ibration (Dec. 31, 1998), discharge from 1999 to 2004 was simulated and compared to observations (Fig. 8b). Figure 8 also shows the error metrics (NSCE=0.65 and Bias=1.04%)for this validation period which indicates that the CREST model can reproduce observed discharge in the Nzoia Basin with acceptable skill.Figure 8c shows the simulation results for Nzoia using TRMM 3B42V6 as precipitation forcing.It can be seen that from 1999 and 2003, the model simulated daily discharge with a Nash-Sutcliff efficiencies of 0.48 and bias of −4.57%.The model for the validation period captures peak and low flows and there is acceptable agreement between simulation and observation at different flow conditions throughout the simulation period.

Hydrological model reconstruction results
Basin-based water balance modeling studies are important in both hydrology and climate research since they provide information on the hydrological cycle and the amount of renewable water available for ecosystems at various land-atmosphere interaction scales ranging, in general, from daily, seasonal, annual, to decadal.Water balance for watersheds, lakes or over a unit land surface area is normally expressed as P +R+ET=dS/dt.Where P is Precipitation, R surface runoff, ET is evapotranspiration and dS/dt change in storage, all in per time (L/T) (Thornthwaite, 1948;V ör ösmarty et al., 1989;Willmott et al., 1985).In this equation, precipitation is the important climate variable for accurate water budget estimation and measured directly on a regular basis.and long-term mean monthly scale and are discussed hereunder.

Runoff
The results from the climatological water balance are show in Fig. 9.The basin average monthly analysis shows that the model produces nearly the same basin-wide runoff.
Model runoff is compared to the river discharge gauged at the catchment outlets of the basin.The location of the discharge measuring station is illustrated in Fig. 1.The runoff estimates are expressed in mm/month, to allow inspection of the relative contribution of the catchment.The overall comparison of runoff estimates are reasonably well matched in magnitude and time evolution (Fig. 9).The model slightly underestimates R for the months of June, July, August and September.The observed values still fall under the ±1 standard deviation (std dev) of monthly mean values.It is to be noted that there is fluctuation of observed streamflow which is an indication of water management practices on the Nzoia River; this is also depicted in Fig. 8b.

Precipitation
We utilized the TMPA 3B42V6 dataset as a forcing parameter to characterize the hydrologic variables at the study basin.As expected, 3B42V6 captures the seasonality of precipitation over the Nzoia Basin.The monthly distribution of 3B42V6 precipitation data also shows two rainy seasons that are comparable with the observed precipitation shown in Fig. 2. Figure 11 shows the spatial distribution of rainfall over the catchment.
The TMPA product showed fairly good agreement throughout the year; similar results are reported in Li et al. (2009).The 3B42V6 estimates fall under the ±1 std dev of monthly mean values throughout the year (Fig. 9).

Evapotranspiration
Estimation of evapotranspiration, a key hydrologic variable provides better understating of the relationships between water balance and climate.In arid and semi arid Introduction

Conclusions References
Tables Figures

Back Close
Full biomes, around 90% or more of the annual precipitation can be evapotranspired, and thus ET determines the freshwater recharge and discharge from aquifers in these environments (Wilcox et al., 2003).Moreover, it is projected that climate change will influence the global water cycle and intensify ET globally (Huntington, 2006;Meehl et al., 2007) consequently impacting the scarce water resources.Therefore, estimation of average monthly and annual evapotranspiration is important.Figure 9 shows the simulated evapotranspiration for the time period.Generally in the drier months, evapotranspiration equals rainfall amounts.The evapotranspiration, however, does not vary as much as rainfall does in a given year.

Summary and discussion
The countries of Eastern Africa are prone to extreme climatic events such as droughts and floods.These events have had severe, negative socio-economic impacts on most of the countries, causing widespread famine and economic hardships in the sub region (Li et al., 2009;Funk et al., 2005).According to the 2007 Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4), the annual precipitation is likely to increase in East Africa and Kenya.Changes are expected in daily, seasonal, inter-annual, and decadal variability.Most of the hydro-climatic studies on East African countries in the past were based mainly on global and regional climate models (Hulme, 2006;Camberlin and Philippon, 2002).In this study, we used observed data from 1985-2006, for the hydroclimatology of Nzoia Basin by studying 1) rainfall and stream discharge patterns, 2) return periods of rainfall and discharge, 3) annual mean discharge and decadal monthly trends of both rainfall and discharge.In addition, a distributed hydrologic model driven by satellite remote sensing data is used to study the water balance of the sub catchment.Runoff and precipitation observation have been used to evaluate the hydrologic model results.
The observed record at Nzoia showed that for the 1985-2006 time period the basin received quite consistently 2-, 5-and 10-years rainfall in totality for the past 21 years Introduction

Conclusions References
Tables Figures

Back Close
Full ( .The second decade (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004) however, received less 5-year and 10-year equivalent rainfalls compared to the first decade (1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994).The discharge data showed that the 2-years returned period equivalent discharge is observed more frequently in the second decade than in the first decade.There is only a marginal increase in annual mean discharge for the last 21 years.The 2-, 5-and 10-years peak discharges, if analyzed more closely, for the entire study period shows that more years since the mid 1990's have high peak discharges even with relatively less precipitation.This might have been the effect of changing land-use and land cover types or increased channelization of the Nzoia Basin over time.
The discharge data for the study period showed that the basin is dry and arid with no sustained base flow.The short spell of high discharge shows the rain caused flooding in the basin.With a decrease in rainfall, the primary input flux into the Nzoia Basin, the water budget situation might deteriorate over the coming years.Noticeable variations in monthly average rainfall and discharge were observed for the two decades (1985-1994 and 1995-2004).The rainfall fluctuated from as low as 55% (in February) to as high as 32% (in December) in drier months.Similarly, there are decreases in February and May monthly average discharge by 13% while January saw a surge of 44%.But overall, there is only a very slight increase (2%) in annual mean discharge suggesting an insignificant imbalance in water budget in the basin during the study period.
The study utilizes quasi-global satellite precipitation and other remote sensing data products.This helps to understand the utility of the remotely sensed data for hydroclimatology studies at a sub-catchment with sparse ground observations.Simulation of the key hydrological processes and their interconnection with climate and basin characteristics is a critical step in estimating catchment water balance.Therefore, a distributed hydrologic model evaluate the model results at the sub-catchment level.TMPA 3B42V6 showed good agreement with gauge observations (Fig. 9).Spatial distribution and time series of CREST modeling results for precipitation (P ), evapotranspiration (ET), runoff (R), and dS/dt are given in Figs. 10 and 11.In general, the model reproduces P , ET, and dS/dt fairly well.Considerable agreement was observed between the monthly model runoff estimates and gauge observations reported for the Nzoia River (Fig. 9).Runoff values respond to precipitation events occurring across the catchment during the wet season from March to early June.The hydrologic model reasonably captured the soil moisture storage variability.An important advantage of spatially distributed hydrologic model, such as CREST, is that it not only provides estimates of hydrological variables at the basin outlet, but also at any location as represented by a cell or grid within the given basin (Fig. 10).These spatially distributed model inputs, states, and outputs, are useful for visualizing the hydrologic behavior of a basin.These results reveal that relatively high flows were being experienced near the basin outlet from previous rainfall, with a new flood peak responding to the rainfall in the upper part of the basin.
Comparison of the model outputs such as evapotranspiration and soil moisture estimates against field measurements can help evaluate the model performance.The model developed from this study can be applied to poorly gauged catchments using satellite forcing data and also be used to investigate the catchment scale water balance.Implementing the CREST model resulted in spatiotemporally distributed hydrological variables that can be utilized in addressing issues pertaining to sustainability of the resources within the catchment.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | exceeded the 5-year return period values.Similarly, 1994, 1998 and 2003 have the peak rainfall of 10-year.Finally, 1985 and 1988 recorded rainfall of a 20-year return period.Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | CREST model simulates the spatio-temporal variation of water fluxes and storages on a regular grid with the grid cell resolution being user-defined.The model can output many variables as a raster grid for any time period.The hydrologic variables were generated from CREST retrospective simulation from 1999 to 2003 using TRMM 3B42V6.These four years were selected to minimize the model run time.Since simulation of the model involves thousands of iterations, model run time in particular is a critical factor to complete a simulation.Water balance basin average calculations were made at daily Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | (CREST) is implemented to simulate hydrological states and flux variables such as runoff, ET, precipitation and soil moisture at a spatial resolution of 30 arcsec at 3 hourly time steps.The CREST model was forced by satellite-based precipitation and evapotranspiration estimates, rain gauge observations, and other remote sensing products.Observations on runoff and precipitation have been used to Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 1 .
Fig. 1.Map of Nzoia river basin in Lake Victoria region, East Africa.

Fig. 1 .
Fig. 1.Map of Nzoia River Basin in Lake Victoria region, East Africa.
The lowest annual discharge is 65.90 m 3 /s in 1986 and the highest is 232.06 m 3 /s in 1994.The other wet years are 1998 and 2006 and the dry years are 1987 and 2002.Overall we can observe a slight increasing trend in annual mean discharge.Seasonal cycles included in annual discharge are noticeable with a greater variability of monthly mean stream flow as shown in Fig. 6b.The maximum monthly discharge is 420.96m 3 /s for May 1985.All the wet years of 1994, 1998 and 2006 are marked by high monthly discharges.The dry years of

Table 1 .
Discharge and rainfall return periods.