Your SlideShare is downloading. ×
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Streamflow simulation using radar-based precipitation applied to the Illinois River basin in Oklahoma, USA

495

Published on

This paper describes the application of a spatially distributed hydrological model WetSpa (Water and Energy Transfer between Soil, Plants and Atmosphere) using radar-based rainfall data provide by the …

This paper describes the application of a spatially distributed hydrological model WetSpa (Water and Energy Transfer between Soil, Plants and Atmosphere) using radar-based rainfall data provide by the United States Hydrology Laboratory of NOAA's National Weather Service for a distributed model intercomparison project. The model is applied to the
river basin above Tahlequah hydrometry station with 30-m spatial resolution and one hour time--step for a total simulation period of 6 years. Rainfall inputs are derived from radar. The distributed model parameters are based on an extensive database of watershed characteristics available for the region, including digital maps of DEM, soil type, and land use. The model is calibrated and validated on part of the river flow records. The simulated hydrograph shows a good correspondence with observation (Nash efficiency coeffiecient >80%, indicating that the model is able to simulate the relevant hydrologic processes in the basin accurately.

Published in: Education, Technology, Sports
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
495
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
16
Comments
0
Likes
1
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. Streamflow simulation using radar-based precipitation data applied to the Illinois River basin, USA Alireza Safari and F. De Smedt Department of Hydrology and Hydraulic Engineering,Vrije Universiteit Brussel Pleinlaan 2, 1050 Brussels, Belgium February 19, 2008 Abstract This paper describes the application of a spatially distributed hydrological model WetSpa (Water and Energy Transfer between Soil, Plants and Atmosphere) using radar-based rainfall data provide by the United States Hydrology Laboratory of NOAA’s National Weather Service for a distributed model intercomparison project. The model is applied to the Illinois river basin above Tahlequah hydrometry station with 30-m spatial resolution and one hour time–step for a total simulation period of 6 years. Rainfall inputs are derived from radar. The distributed model parameters are based on an extensive database of watershed characteristics available for the region, including digital maps of DEM, soil type, and land use. The model is calibrated and validated on part of the river flow records. The simulated hydrograph shows a good correspondence with observation, indicating that the model is able to simulate the relevant hydrologic processes in the basin accurately.Keywords: WetSpa, Physically-based Distributed Hydrologic Model, DMIP, NEXRAD Stage III rainfall data, Stream-flow Simulation, PEST, Illinois River basin.1 IntroductionRainfall–runoff models are used and developed by hydrologists to model rainfall–runoff processes. TheNOAA–sponsored Distributed Model Intercomparison Project (DMIP) provides a forum to explore theapplicability of distributed models using operational quality data in order to improve flow modeling andprediction along the entire river system [Smith, 2002, Smith, 2004, Smith et al., 2004].The US National Weather Service’s (NWS) Next Generation Weather Radar WSR-88D (NEXRAD) precip-itation products are widely used in hydrometeorology and climatology for rainfall estimation [Ciach et al.,1997, Seo et al., 1999, Krajewsk and Smith, 2002, Uijlenhoet et al., 2003], precipitation and weather fore-casting [Johnson et al., 1998,Grecu and Krajewski, 2000] and flood forecasting [Johnson et al., 1999,Younget al., 2000, Smith et al., 2005, Reed et al., 2007]. The most commonly NEXRAD product in hydrometeo-rological applications is the NEXRAD Stage III data, since it involves the correction of radar rainfall rateswith multiple surface rain gauges and has a significant degree of meteorological quality control [R.A.Fulton 1
  • 2. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 2et al., 1998]. However, because the Stage III products consist of high-resolution spatial-temporal precipi-tation data over large regions, it is difficult to use in conjunction with other geospatial products [Reed andMaidment, 1995, Reed and Maidment, 1999]. The nominal size of an HRAP grid cell is 4 km by 4 km andthe temporal resolution of these grids is 1-hour. The grid data is provided in binary XMRG format [Officeof Hydrologic Development, 2006].Distributed hydrological models are usually parameterized by deriving estimates of parameters from topog-raphy and physical properties of the soils, aquifers and land use in the basin. In recent years, a number ofmethods have been developed for the estimation of hydrologic model parameters. One frequently used andrelatively simple algorithm is the parameter estimation PEST method. This automatic calibration procedureuses a nonlinear estimation technique known as the Gauss-Marquardt-Levenberg method. The strength ofthis method lies in the fact that it can generally estimate parameter values using fewer model runs thanany other method. The program is able to run a model as many times as needed while adjusting parametervalues until the discrepancies between selected model outputs and a complementary set of field or labora-tory measurements is reduced to a minimum in a weighted least-squares sense. Numerous examples of theapplication of the PEST algorithm for the calibration of hydrologic models can be found in the literatureand PEST proves to be a time-saving tool compared to other model calibration techniques [Al-Abed andWhiteley, 2002, Baginska et al., 2003, Zyvoloski et al., 2003, Doherty and Johnston, 2003, Wang and Me-lesse, 2005, Liu et al., 2005, Bahremand and De Smedt, 2006, Bahremand and De Smedt, 2007, Goegebeurand Pauwels, 2007, Nossent and Bauwens, 2007].A few years ago, the United States Hydrology Laboratory (HL) (then the Hydrologic Research Laboratory)of NWS began a major research effort called Distributed Model Intercomparison Project (DMIP) to addressthe question: how can the NWS best utilize the NEXRAD data to improve its river forecasts? The resultsuggested that spatial rainfall averages derived from the NEXRAD data can improve flood prediction inmid/large basins as compared to gage-only averages [Bandaragoda et al., 2004, Reed et al., 2004, Smithet al., 2004,Reed et al., 2007]. Previous studies on some of the DMIP basins have shown that calibration ofdistributed hydrological models significantly improves simulation results [Ajami et al., 2004, Bandaragodaet al., 2004, Carpenter and Georgakakos, 2004, Ivanov et al., 2004, Butts et al., 2005].In this paper, a spatially distributed physically based hydrologic model, WetSpa, is applied to a subwater-shed of the Illinois River basin which forms part of the DMIP basins [Smith, 2002]. The paper is organizedas follows. First, a brief description of the hydrologic model used in this study is given. Next, model per-formance indicators are discussed. In section 3, the site and data used are described. Details of the modelcalibrations and a comparison of the results are given in section 4. Finally, conclusions that can be drawnfrom this work are presented.2 MethodologyThe WetSpa model is capable of predicting runoff and river flow at any gauged and ungauged location in awatershed on hourly time scale [Wang et al., 1997,De Smedt et al., 2000,Liu et al., 2003,Bahremand et al.,2006, Zeinivand et al., 2007]. Availability of spatially distributed data sets (digital elevation model, lan-duse, soil and radar-based precipitation data) coupled with GIS technology enables the WetSpa to performspatially distributed calculations. The hydrological processes considered in the model are precipitation,interception, depression storage, surface runoff, infiltration, evapotranspiration, percolation, interflow and
  • 3. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 3ground water drainage. The total water balance for each raster cell is composed of a separate water balancefor the vegetated, bare-soil, open water, and impervious part of each cell. A mixture of physical and em-pirical relationships is used to describe the hydrological processes. The model predicts discharges in anylocation of the channel network and the spatial distribution of hydrological characteristics. Hydrologicalprocesses are set in a cascading way. Starting from precipitation, incident rainfall first encounters the plantcanopy, which intercepts all or part of the rainfall until the interception storage capacity is reached. Excesswater reaches the soil surface and can infiltrate the soil zone, enter depression storage, or is diverted assurface runoff. The sum of interception and depression storage forms the initial loss at the beginning ofa storm, and does not contribute to the storm flow. A fraction of the infiltrated water percolates to thegroundwater storage and some is diverted as interflow. Soil water is also subjected to evapotranspirationdepending on the potential evapotranspiration rate and the available soil moisture. Groundwater dischargesto the nearest channel proportional to the groundwater storage and a recession coefficient. Possible evapo-transpiration from groundwater storage is also accounted for. For each grid cell, the root zone water balanceis modeled continuously by equating inputs and outputs: dθ D = P−I−S −E−R−F (1) dtwhere D [L] is root depth, θ [L3 L−3 ] soil moisture, t [T] time, I [LT−1 ] interception loss, S [LT−1 ] surfacerunoff, E [LT−1 ] evapotranspiration from the soil, R [LT−1 ] percolation out of the root zone, and F [LT−1 ]interflow. The surface runoff is calculated using a moisture–related modified rational method with a runoffcoefficient depending on land cover, soil type, slope, magnitude of rainfall, and antecedent soil moisture: α θ S = C(P − I) (2) θswhere θ s [L3 L−3 ] is water content at saturation, C[−] potential runoff coefficient, and α[−] a parameterreflecting the effect of rainfall intensity. The values of C are derived from a lookup table, linking values toslope, soil type and landuse classes [Liu and De Smedt, 2004]. The value of α reflects the influence of thesoil wetness on runoff and needs to be set by the user or optimized by model calibration, within the interval0 to 1.Evapotranspiration from the soil and vegetation is calculated based on the relationship developed by Thorn-thwaite and Mather [Thornthwaite and Mather, 1955], as a function of potential evapotranspiration, vege-tation type, stage of growth and soil moisture content.  E=0 θ < θw    f or     θ − θw     E = (cv E p − Ei − Ed ) θw ≤ θ < θ f  (3)  f or     θ f − θw      E =c E −E −E   f or θ ≥ θf v p i dwhere cv [−] is a vegetation coefficient, which varies throughout the year depending on growing stage andvegetation type, E p [LT−1 ] is the potential evapotranspiration, Ei [LT−1 ] and Ed [LT−1 ] are evaporationfrom interception storage and depression storage respectively, θw [L3 L−3 ] is the moisture content at per-
  • 4. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 4manent wilting point, and θ f [L3 L−3 ] is the moisture content at field capacity. The rate of percolation R orgroundwater recharge is determined by Darcy’s law [Hillel, 1980] in function of the hydraulic conductivityand the gradient of hydraulic potential. When the assumption is made that the pressure potential only variesslightly in the soil, the percolation is controlled by gravity alone [Famiglietti and Wood, 1994]. Therefore,the percolation out of root zone is simply the hydraulic conductivity corresponding to the moisture contentin the soil layer, which can be derived by the Brooks and Corey relationship [Eagleson, 1978]: 3+2/B θ − θr R = K (θ) = K s (4) θ s − θrwhere K(θ) [LT−1 ] is the unsaturated hydraulic conductivity, K s [LT−1 ] is the saturated hydraulic conduc-tivity, θr [L3 L−3 ] is the residual soil moisture content, and B[−] is the soil pore size distribution index. Thevertical transport of water through the unsaturated soil matrix is slow. It generally takes days or monthsbefore the percolating water reaches the saturated zone. Nevertheless, precipitation is followed by an al-most immediate rise of the groundwater table in consequence of a rapid transfer of increased soil-waterpressure through the unsaturated zone [Myrobo, 1997]. In addition, macro pores in the subsurface layersresulting from root and fauna activity may allow rapid bypassing of the unsaturated zone when the rate ofprecipitation is high [Beven and Germann, 1982].Interflow is assumed to become significant only when the soil moisture is higher than field capacity. Darcy’slaw and a kinematic wave approximation are used to determine the amount of interflow, in function of hy-draulic conductivity, moisture content, slope angle, and root depth: F = c f DS 0 K (θ) W (5)where S 0 [LL−1 ] is the surface slope, W [L] is the cell width, and c f [-] is a scaling parameter dependingon land use, used to consider river density and the effects of organic matter on the horizontal hydraulicconductivity in the top soil layer. Apparently, interflow will be generated in areas with high moisture andsteep slope.The routing of overland flow and channel flow is implemented by the method of the diffusive wave approx-imation of the St. Venant equation: ∂Q ∂Q ∂2 Q +c −d 2 =0 (6) ∂t ∂x ∂twhere Q [L3 T−1 ] is the discharge, t [T] is the time, x [L] is the distance along the flow direction, c [LT−1 ]is the kinematic wave celerity, interpreted as the velocity by which a disturbance travels along the flowpath, and d [L2 T−1 ] is the location dependent dispersion coefficient, which measures the tendency of thedisturbance to disperse longitudinally as it travels downstream. Assuming that the water level gradientequals the bottom slope and the hydraulic radius approaches the average flow depth for overland flow, c andd can be estimated by c = (5/3) v, and d = (vH) / (2S 0 ) [18], where v [LT−1 ] is the flow velocity calculatedby the Manning equation, and H[L] is the hydraulic radius or average flow depth. An approximate solutionto the diffusive wave equation in the form of a first passage time distribution is applied [Liu et al., 2003],relating the discharge at the end of a flow path to the available runoff at the beginning: 1 (t − t0 )2 U (t) = exp − (7) 2σ2 t/t0 σ 2πt3 /t0 3
  • 5. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 5where U (t)[T−1 ] is the flow path unit response function, serving as an instantaneous unit hydrograph (IUH)of the flow path, which makes it possible to route water surplus from any grid cell to the basin outlet orany downstream convergent point, t0 [T] is the average flow time, and σ [T] is the standard deviation of theflow time. The parameters t0 and σ are spatially distributed, and can be obtained by integration along thetopographic determined flow paths as a function of flow celerity and dispersion coefficient: t0 = c−1 dx (8)and σ2 = 2dc−3 dx (9)Hence, flow hydrographs at the basin outlet or any downstream convergent point are obtained by a convo-lution integral of the flow response from all contributing cells: t Q (t) = S (τ) U (t − τ) dτdA (10) A 0where Q (t) [L3 T−1 ] is the direct flow hydrograph, S (τ) [LT−1 ] is the surface runoff generated in a grid cell,τ [T] is the time delay, and A [L2 ] is the drainage area of the watershed.Because, groundwater movement is much slower than the movement of water in the surface and nearsurface water system, groundwater flow is simplified as a lumped linear reservoir on small GIS derivedsubwatershed scale. Direct flow and groundwater flow are joined at the subwatershed outlet, and the totalflow is routed to the basin outlet by the channel response function derived from equation (7).One advantage of WetSpa is that it allows spatially distributed hydrological parameters of the basin tobe used as inputs to the model simulated within a GIS framework. Inputs to the model include digitalelevation data, soil type, land use data, and climatological data. Stream discharge data is optional formodel calibration.Efficiency criteria are defined as mathematical measures of how well a model simulation fits the availableobservations [Beven, 2001]. The efficiency criteria used in this study are listed in Table 1. Criterion C1is reflecting the ability of reproducing the water balance; C2 is a proposed index [McCuen and Snyder,1975] which reflects differences both in hydrograph size and in hydrograph shape, C3 evaluates the abilityof reproducing the streamflow hydrograph [Nash and Sutcliffe, 1970], and finally C4 and C5 evaluate theability of reproducing low flows and high flows respectively. In all equations, Q s and Qo are the simulatedand observed streamflows at time step i, Qo is the mean observed streamflow over the simulation period, σo ¯and σ s are the standard deviations of observed and simulated discharges respectively, r is the correlationcoefficient between observed and simulated hydrographs, and N is the number of observations. To evaluatethe goodness of the model performance during calibration and validation periods, the intervals listed inTable 2 have been adopted [Andersen et al., 2001, Andersen et al., 2002, Henriksen et al., 2003]. Thesecriteria are not of the fail/pass type, but evaluate the performance in five categories from excellent to verypoor. The perfect value for C1 is 0 and for the other four criteria it is 1.
  • 6. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 6 Table 1: Performance criteria for model assessment Criterion Equation N i=1 (Q si −Qoi ) Model Bias C1 = N i=1 Qoi min{σo ,σ s } Modified correlation index C2 = max{σo ,σ s } ×r N 2 i=1 (Q si −Qoi ) Nash-Sutcliffe efficiency C3 = 1 − N ¯ 2 i=1 ( Qoi −Qo ) N 2 i=1 (ln Q si −ln Qoi ) Logarithmic version of Nash-Sutcliffe efficiency for low C4 = 1 − N ¯ 2 i=1 ( ln Qoi −ln Qo ) flow evaluation N (Qoi +Qo )(Qsi −Qoi )2 ¯ Adapted version of Nash-Sutcliffe efficiency for high flow C5 = 1 − i=1 N i=1 (Qoi +Qo )(Qoi −Qo )2 ¯ ¯ evaluation Table 2: Model performance categories to indicate the goodness of fit level Category Model bias criterion: C1 Model efficiencies’ criteria: C2 , C3 , C4 and C5 Excellent <0.05 >0.85 Very good 0.05-0.10 0.65-0.85 Good 0.10-0.20 0.50-0.65 Poor 0.20-0.40 0.20-0.50 Very poor >0.40 <0.203 Study Area and DataThe Illinois river is located in eastern Oklahoma and western Arkansas. Streamflow data from Tahlequahgauging station are used in this study. Figure 1 shows the location of this station and the correspondingbasin boundaries. The basin area is 2454 km2 . The average maximum and minimum air temperature inthe region are approximately 22 and 9◦C, respectively. Summer maximum temperatures can get as high as38◦C. The annual average precipitation is 1200 mm, and the annual average potential evaporation is 1050mm.The topographic data was obtained from the DMIP web site in raster form with a resolution of 30 m. Thetopography is gently rolling to hilly and the elevation ranges from 210 m to 600 m (Figure 1).Landuse and soil types information are important inputs to the WetSpa model, as these influence hydrolog-ical processes like evapotranspiration, interception, infiltration, runoff, etc. Soil types were derived fromPennsylvania State University STATSGO data [Soil Survey Staff, 1994a, Soil Survey Staff, 1994b] andlanduse from satellite images processed through the NASA Land Data Assimilation Systems (LDAS) pro-gram with an International Geosphere-Biosphere Program (IGBP) classification system [Eidenshink andFaundeen, 1994]. The spatial resolution of the soil types and landuse maps is 1 km. Figure 2 shows theland use and soil maps of the study area.Hourly radar-based rainfall data was acquired from the DMIP database. The data sets have to be untarred,uncompressed, transformed into ASCII format, and projected into a common coordinate system. The se-lected mesh of NEXRAD (pseudo-station network) consisting of 150 radar points is shown in Figure 3together with associated Thiessen polygons covering the study area.
  • 7. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 7 Elevation (m) 200 - 300 300 - 400 400 - 400 400 - 500 500 - 600 N W E S # # Tahlequah gauging station 10 0 10 20 Kilometers Figure 1: The Illinois river basin above Tahlequah on the border of Oklahoma and Arkansas, USA (a) Landuse categories: (b) Soil Texture Classes: Evergreen Needleleaf Forest Silt Loam Deciduous Broadleaf Forest Sandy Clay Loam Mixed Forest Silty Clay Loam Woody Savannah Silty Clay Croplands Urban and Built-Up Cropland/Natural veg. Mosaic N N W E W E S S 10 0 10 20 Kilometers 10 0 10 20 Kilometers Figure 2: The landuse and soil maps of the study area
  • 8. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 8 # # Rivers # # # Mesh of NEXRAD cells over the basin # # # # Thiessen polygons # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # N # # # # # # # # # # # # W E # # # # # # # # # S # # # # # # 10 0 10 20 Kilometers # # Figure 3: Thiessen polygons of the study area
  • 9. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 9 Table 3: WetSpa global parameters and their calibrated values Parameter Symbol Unit calibrated value Interflow scaling factor Ki - 4.0 Groundwater flow recession coefficient Kg d−1 0.00125 Correction factor for potential evapotranspiration Kep - 0.875 Maximum groundwater storage gmax mm 300 Surface runoff exponent Krun - 10.0 Table 4: Simulation statistics for the calibration and validation periods Criterion Calibration period Validation period C1 0.145 0.043 C2 0.875 0.765 C3 0.785 0.854 C4 0.745 0.738 C5 0.886 0.9064 Results and discussionTheoretically the parameters of the physically based WetSpa model need not to be calibrated. However,due to uncertainty of the model input and structure, a calibration of global model parameters (Table 3)improves the model performance. Global model parameters are time invariant and are either adjustmentcoefficients or empirical constants. These parameters and their calibration with PEST have been describedin other studies [Liu and De Smedt, 2004, Liu et al., 2005, Bahremand et al., 2006, Bahremand and DeSmedt, 2007]. The historical discharge record was divided into two periods: one for calibration and onefor validation. The calibration period covers the period October 1996 to September 1999 and the validationperiod the remainder until September 2002. Figures 4 and 5 give a graphical comparison between observed and calculated hourly flows for the calibra-tion and the validation periods, showing that the calibrated model simulates the timing and the magnitudeof the peak flows reasonably well. Table 4 presents summary statistics for the calibration and validationperiods. It is clear that the WetSpa model is performing well (Nash-Sutcliffe efficiency and modified cor-relation index > 0.75). The results show in particular that the model is able to simulate high flows witha very good accuracy (C5 > 0.85), while also performing well for the low base flows (C4 >0.7). Thelesser precision for low flows might be due to the simplified approach of modeling ground water storageand drainage in WetSpa. The use of distributed and physically based predicting models could improvebase flow predictions. There is also significant uncertainty in precipitation estimates derived from weatherradar [Smith et al., 1996, Stellman et al., 2001, Carpenter and Georgakakos, 2004], which influence modelsimulations significantly on all model output scales. Also, there is some uncertainty due to the inability ofthe model to represent the heterogeneous nature of hydrological processes on basin scale.5 Conclusions and recommendationsThis paper presents an application of the physically based distributed hydrologic WetSpa model, forcedwith radar-based precipitation data for a Distributed Model Intercomparison Project (DMIP) watershed.Coupling to GIS makes WetSpa a powerful tool to capture local complexities of a watershed and temporal
  • 10. Discharge (m3/s) Discharge (m3/s) 0 200 400 600 800 1000 1000 0 200 400 600 800 OCT 1999 OCT 1996 NOV 1999 NOV 1996 DEC 1999 DEC 1996 JAN 2000 JAN 1997 FEB 2000 FEB 1997 MAR 2000 MAR 1997 APR 2000 APR 1997 MAY 2000 MAY 1997 JUN 2000 JUN 1997 JUL 2000 JUL 1997 AUG 2000 AUG 1997 SEP 2000 SEP 1997 OCT 2000 OCT 1997 NOV 2000 NOV 1997 DEC 2000 DEC 1997 JAN 2001 JAN 1998 Mean areal precipitation Mean areal precipitation FEB 2001 FEB 1998 MAR 2001 MAR 1998 APR 2001 APR 1998 MAY 2001 MAY 1998 JUN 2001 JUN 1998 JUL 2001 JUL 1998 Observed Observed AUG 2001 AUG 1998 SEP 2001 SEP 1998 BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 OCT 2001 OCT 1998 NOV 2001 NOV 1998 DEC 2001 DEC 1998 JAN 2002 JAN 1999 Simulated FEB 2002 Simulated FEB 1999 MAR 2002 MAR 1999 APR 2002 APR 1999 MAY 2002 MAY 1999 JUN 2002 JUN 1999 JUL 2002 JUL 1999 AUG 2002 AUG 1999 SEP 2002 SEP 1999 0 0 80 60 40 20 80 60 40 20 100 100 Mean areal precipitation (mm/h) Mean areal precipitation (mm/h)Figure 5: Observed and simulated hydrographs for the validation period with hourly time scale Figure 4: Observed and simulated hydrographs for the calibration period with hourly time scale 10
  • 11. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 11variation of river flows, especially peak discharges. The model can predict not only the streamflow hydro-graph at any controlling point of the basin, but also the spatially distributed hydrological processes, suchas surface runoff , infiltration, evapotranspiration and the like, at each time step during a simulation. Allmodel parameters can be obtained from DEM, land use and soil type data of the watershed or combinationsof these three fundamental maps.Evaluation of flow simulations from WetSpa is presented in terms of summary statistics covering calibra-tion and validation periods. The goodness of the model performance during calibration and validationperiods was evaluated and shows that the calibrated WetSpa reproduces water balance and especially highstreamflows accurately, while this is somewhat less for low flows, due to the simplified model descriptionof ground water flow processes. Hence, the model performance is satisfactory for both calibration andvalidation periods. The model’s ability to reproduce observed hydrographs for the validation period showsthat the model can be used for prediction purposes, in particular for storm events that lead to flooding.A potential future research is to validate the model in watersheds where snow accumulation and abla-tion is significant (e.g. DMIP California Sierra Nevada watersheds). For these cases and especially inmountainous terrain, uncertainty in estimated distributed precipitation is high as weather radar data can besignificantly biased (e.g. partial beam filling, ground clutter, etc.).References[Ajami et al., 2004] Ajami, N. K., Gupta, H., Wagener, T., and Sorooshian, S. (2004). Calibration of a semi-distributed hydrologic model for streamflow estimation along a river system. Journal of Hydrology, 298:112–135.[Al-Abed and Whiteley, 2002] Al-Abed, N. A. and Whiteley, H. R. (2002). Calibration of the Hydrolog- ical Simulation Program Fortran (HSPF) model using automatic calibration and geographical informa- tion systems. Hydrological Processes, 16:3169–3188.[Andersen et al., 2002] Andersen, J., Dybkjaer, G., Jensen, K. H., Refsgaard, J. C., and Rasmussen, K. (2002). Use of remotely sensed precipitation and leaf area index in a distributed hydrological model. Journal of Hydrology, 264:34–50.[Andersen et al., 2001] Andersen, J., Refsgaard, J., and K.H.Jensen (2001). Distributed hydrological mod- elling of the senegal river basin–model construction and validation. Journal of Hydrology, 247:200–214.[Baginska et al., 2003] Baginska, B., Milne-Home, W., and Cornish, P. (2003). Modelling nutrient trans- port in Currency Creek, NSW with AnnAGNPS and PEST. Environmental Modelling & Software, 18:801–808.[Bahremand and De Smedt, 2006] Bahremand, A. and De Smedt, F. (2006). Sensitivity and uncertainty analysis of a GIS-based flood simulation model using PEST. Journal of WSEAS Transaction on Envi- ronment and Development, 2(1):29–37.[Bahremand and De Smedt, 2007] Bahremand, A. and De Smedt, F. (April , 2007). Distributed Hydro- logical Modeling and Sensitivity Analysis in Torysa Watershed, Slovakia . Journal of Water Resources Management, Published Online.
  • 12. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 12[Bahremand et al., 2006] Bahremand, A., De Smedt, F., Corluy, J., and Liu, Y. B. (2006). Application of WetSpa model for assessing land use impacts on floods in the Margecany-Hornad watershed, Slovakia. Journal of Water Science Technology, 53(10):37–45.[Bandaragoda et al., 2004] Bandaragoda, C., Tarboton, D. G., and Woods, R. (2004). Application of TOP- NET in the distributed model intercomparison project. Journal of Hydrology, 298:178–201.[Beven, 2001] Beven, J. K., editor (2001). Rainfall-Runoff Modelling –The Primer. John Wiley & Sons Ltd., Chichester.[Beven and Germann, 1982] Beven, K. and Germann, P. (1982). Macropores and Water Flow in Soils. Water Resources Research, 18, No 5:1311–1325.[Butts et al., 2005] Butts, M., Overgaard, J., Viaene, P., Dubicki, A., Stronska, K., Szalinska, W., Lewandowski, A., Olszewski, T., and Kolerski, T. (2005). Flexible process-based hydrological mod- elling framework for flood forecasting–MIKE SHE. In Proc. International conference on Innovation, advances and implementation of flood forecasting technology, Troms, Norway.[Carpenter and Georgakakos, 2004] Carpenter, T. M. and Georgakakos, K. P. (2004). Continuous stream- flow simulation with the HRCDHM distributed hydrologic model. Journal of Hydrology, 298:61–79.[Ciach et al., 1997] Ciach, G., Krajewski, W., Anagnostou, E., Baeck, M., Smith, J., McCollum, J., and Kruger, A. (1997). Radar Rainfall Estimation for Ground Validation Studies of the Tropical Rainfall Measuring Mission. Journal of Applied Meteorology, 36:735–747.[De Smedt et al., 2000] De Smedt, F., Liu, Y., and Gebremeskel, S. (2000). Hydrological Modeling on a watershed scale using GIS and remote sensed land use information. In C.A. Brebbia (ed), WTI press, Boston.[Doherty and Johnston, 2003] Doherty, J. and Johnston, J. M. (2003). Methodologies for calibration and predictive analysis of a watershed model. Journal of the American Water Resources Association, Volume 39 Issue 2:251–265.[Eagleson, 1978] Eagleson, P. (1978). Climate, soil, and vegetation, a simplified model of soil moisture movement in liquid phase. Journal of Water Resources Research, 14 (5):722–730.[Eidenshink and Faundeen, 1994] Eidenshink, J. and Faundeen, J. (1994). The 1-Km AVHRR global land data set: first stages in implementation. International Journal of Remote Sensing, 15:3443–3462.[Famiglietti and Wood, 1994] Famiglietti, J. and Wood, E. (1994). Multiscale modelling of spatially vari- able water and energy balance processes. Water Resources Research, 30(11):3061–3078.[Goegebeur and Pauwels, 2007] Goegebeur, M. and Pauwels, V. R. (2007). Improvement of the PEST parameter estimation algorithm through Extended Kalman Filtering. Journal of Hydrology, 337:436– 451.[Grecu and Krajewski, 2000] Grecu, M. and Krajewski, W. (2000). A large-sample investigation of statis- tical procedures for radar-based short-term quantitative precipitation forecasting. Journal of Hydrology, 239:69–84.
  • 13. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 13[Henriksen et al., 2003] Henriksen, H. J., Troldborg, L., Nyegaard, P., Sonnenborg, T. O., Refsgaard, J. C., and Madsen, B. (2003). Methodology for construction, calibration and validation of a national hydro- logical model for Denmark. Journal of Hydrology, 280:52–71.[Hillel, 1980] Hillel, D. (1980). Application of Soil Physics. Academic Press, New York.[Ivanov et al., 2004] Ivanov, V. Y., Vivoni, E. R., Bras, R. L., and Entekhabi, D. (2004). Preserving high- resolution surface and rainfall data in operational-scale basin hydrology: a fully-distributed physically- based approach. Journal of Hydrology, 298:80–111.[Johnson et al., 1999] Johnson, D., Smith, M., Koren, V., and Finnerty, B. (1999). Comparing mean areal precipitation estimates from NEXRAD and rain gauge networks. Journal of Hydrologic Engineering, 4 (2):117–124.[Johnson et al., 1998] Johnson, J., MacKeen, P., Witt, A., Mitchell, E., G.J.Stumpf, Eilts, M., and Thomas, K. (1998). The storm cell identification and tracking algorithm: an enhanced WSR-88D algorithm. Journal of Weather and Forecasting, 13:263–276.[Krajewsk and Smith, 2002] Krajewsk, W. and Smith, J. (2002). Radar hydrology: rainfall estimation. Journal of Advances in Water Resources, 25:1387–1394.[Liu and De Smedt, 2004] Liu, Y. and De Smedt, F. (2004). WetSpa extension, documentation and user manual. Department of Hydrology and Hydraulic Engineering, Vrije Universiteit Brussel, Belgium.[Liu et al., 2003] Liu, Y., Gebremeskel, S., De Smedt, F., Hoffmann, L., and Pfister, L. (2003). A diffusive transport approach for flow routing in GIS-based flood modeling. Journal of Hydrology, 283:91–106.[Liu et al., 2005] Liu, Y. B., Batelaan, O., De Smedt, F., Poorova, J., and Velcicka, L. (2005). Automated calibration applied to a GIS-based flood simulation model using PEST. In van Alphen, J., van Beek, E., and Taal, M., editors, Floods, from Defence to Management, pages 317–326, London. Taylor-Francis Group.[McCuen and Snyder, 1975] McCuen, R. H. and Snyder, W. M. (December 1975). A Proposed Index for Comparing Hydrographs. Water Resources Research, VOL. 11, NO. 6:1021–1024.[Myrobo, 1997] Myrobo, S. (1997). Tempral and spatial scale of response area and groundwater variation in till. Hydrologic Processes, 11:1861–1880.[Nash and Sutcliffe, 1970] Nash, J. and Sutcliffe, J. (1970). River flow forecasting through conceptual models, part 1. a discussion of principles. Journal of Hydrology, 10:282–290.[Nossent and Bauwens, 2007] Nossent, J. and Bauwens, W. (2007). Comparing SWAT and WetSpa on the River Grote Laak, Belgium. In The 4th International SWAT Conference, Delft, The Netherlands. UNESCO-IHE.[Office of Hydrologic Development, 2006] Office of Hydrologic Development, N. (2006). About the Mul- tisensor (NEXRAD and gauge) data. Technical report, National Oceanic and Atmospheric Administra- tion (NOAA), National Weather Service (NWS).
  • 14. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 14[R.A.Fulton et al., 1998] R.A.Fulton, Breidenbach, J., Seo, D., Miller, D., and OBannon, T. (1998). The WSR-88D rainfall algorithm. Journal of Weather and Forecasting, 13:377–395.[Reed et al., 2004] Reed, S., Koren, V., Smith, M. B., Zhang, Z., Moreda, F., Seo, D.-J., and DMIP Par- ticipants (2004). Overall distributed model intercomparison project results. Journal of Hydrology, 298:27–60.[Reed and Maidment, 1995] Reed, S. and Maidment, D. (1995). A GIS procedure for merging NEXRAD precipitation data and digital elevation models to determine rainfall-runoff modeling parameters. Online Report 95-3, Center for Research in Water Resources (CRWR), University of Texas at Austin.[Reed and Maidment, 1999] Reed, S. and Maidment, D. (1999). Coordinate transformations for using NEXRAD data in GIS-based hydrologic modeling. Journal of Hydrologic Engineering, 4:174–182.[Reed et al., 2007] Reed, S., Schaake, J., and Zhang, Z. (2007). A distributed hydrologic model and thresh- old frequency-based method for flash flood forecasting at ungauged locations. Journal of Hydrology, 337:402– 420.[Seo et al., 1999] Seo, D., Breidenbach, J., and Johnson, E. (1999). Real-time estimation of mean field bias in radar rainfall data. Journal of Hydrology, 223:131–147.[Smith et al., 2005] Smith, J. A., Miller, A. J., Baeck, M. L., Nelson, P. A., Fisher, G. T., and Meierdiercks, K. L. (2005). Extraordinary Flood Response of a Small Urban Watershed to Short-Duration Convective Rainfall. Journal of Hydrometeorology, 6:599–617.[Smith et al., 1996] Smith, J. A., Seo, D. J., Baeck, M. L., and Hudlow, M. D. (1996). An intercomparison study of NEXRAD precipitation estimates. Water Resources Research, 32:2035–2046.[Smith, 2002] Smith, M. B. (2002). Distributed model intercomparison project (DMIP). from the World Wide Web: http://www.nws.noaa.gov/oh/hrl/dmip/.[Smith, 2004] Smith, M. B. (2004). The distributed model intercomparison project (DMIP). Journal of Hydrology, 298:1–3.[Smith et al., 2004] Smith, M. B., Seo, D.-J., Koren, V. I., Reed, S. M., Zhang, Z., Duan, Q., Moreda, F., and Cong, S. (2004). The distributed model intercomparison project (DMIP): motivation and experiment design. Journal of Hydrology, 298:4–26.[Soil Survey Staff, 1994a] Soil Survey Staff (1994a). State Soil Geographic Database (STATSGO) Col- lection for the Conterminous United States. CD-ROM, USDA-NRCS, National Soil Survey Center, Lincoln, NE.[Soil Survey Staff, 1994b] Soil Survey Staff (1994b). State Soil Geographic Database (STATSGO) Data Users Guide Miscellaneous Publication 1492. Technical report, USDA Natural Resources Conservation Service, US Government Printing Office, Washington.[Stellman et al., 2001] Stellman, K., Fuelberg, H., Garza, R., and Mullusky, M. (2001). An examination of radar and rain gauge–derived mean areal precipitation over georgia watersheds. Weather and Fore- casting, 16:133–144.
  • 15. BALWOIS 2008-Ohrid, Republic of Macedonia- 27, 31 May 2008 15[Thornthwaite and Mather, 1955] Thornthwaite, C. and Mather, J. (1955). The water balance. Technical Report Publ. No. 8, Laboratory of Climatology, Centerton NJ.[Uijlenhoet et al., 2003] Uijlenhoet, R., Steiner, M., and Smith, J. (2003). Variability of Raindrop Size Distributions in a Squall Line and Implications for Radar Rainfall Estimation. Journal of Hydrometeo- rology, 4:43–61.[Wang and Melesse, 2005] Wang, X. and Melesse, A. M. (2005). Evaluation of the SWAT model’s snowmelt hydrology in a northwestern Minnesota watershed. Journal of American Society of Agri- cultural and Biological Engineers, 48(4):1359–1376.[Wang et al., 1997] Wang, Z., Batellan, O., and De Smedt, F. (1997). A distributed model for Water and Energy Transfer between Soil, Plants and Atmosphere (WetSpa). Journal of Physics and Chemistry of the Earth, 21:189–193.[Young et al., 2000] Young, C., Bradley, A. A., Krajewski, W., Kruger, A., and Morrissey, M. (2000). Evaluating NEXRAD multisensor precipitation estimates for operational hydrologic forecasting. Jour- nal of Hydrometeorology, 1:241–254.[Zeinivand et al., 2007] Zeinivand, H., De Smedt, F., and Bahremand, A. (2007). Simulation of basin runoff due to rainfall and snowmelt. In MODSIM 2007 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, pages 1796–1802, Christchurch, New Zealand.[Zyvoloski et al., 2003] Zyvoloski, G., Kwicklis, E., Eddebbarh, A. A., Arnold, B., Faunt, C., and Robin- son, B. A. (2003). The site-scale saturated zone flow model for Yucca Mountain: calibration of different conceptual models and their impact on flow paths. Journal of Contaminant Hydrology, 62–63:731–750.

×