An Investigation of the Effect of Drainage Density on Hydrologic

Turkish J. Eng. Env. Sci. 28 (2004) , 85 – 94. c TUB¨ ITAK˙

An Investigation of the Eﬀect of Drainage Density on Hydrologic Response

Osman YILDIZ Kırıkkale University, Faculty of Engineering, Department of Civil Engineering, Kırıkkale-TURKEY e-mail: [email protected]

Received 31.02.2002

Abstract The sensitivity of streamflow simulations to the drainage density of river basins was investigated. A physically based spatially distributed hydrologic model was used in the model experiments. The hydrologic model was applied to the Monongahela river basin in the United States of America for the simulation of 1988 and 1993 hydrologic regimes for selected periods between April and July. Model simulations of streamflows for 3 different drainage density scenarios (0.2, 0.24 and 0.38 km−1) were compared against the observations. Evaluation of the model results indicated that the hydrologic model response changes significantly for the prescribed drainage densities. In general, the hydrologic model overestimated the stream discharges in response to an increase in drainage density. This outcome was attributed to an increase in the number of channel pixels, and thus an increase in the subsurface flow contribution to the total streamflow.

Key words: Drainage density, Hydrologic model, Streamﬂow, River basin.

Introduction ally yield 3 different drainage densities, were used for the model simulations of streamflow hydro- Drainage density, a fundamental concept in hydrographs. The effect of drainage density was evaluated logic analysis, is defined as the length of drainage per through comparison of the model simulated stream- unit area. The term was first introduced by Horton flows against the observations on a daily basis at the (1932) and is determined by dividing the total length outlet of the river basin. of streams within a drainage basin by the drainage area. A high drainage density reflects a highly dis- The 1988 and 1993 extreme hydrologic sected drainage basin with a relatively rapid hydro- regimes logic response to rainfall events, while a low drainage The drought of 1988 and the flood of 1993 are among density means a poorly drained basin with a slow hy- the most severe occurrences of climatic extremes in drologic response (Melton, 1957). the continental United States during recent decades. The objective of this study was to investigate the The occurrence of these extremes has been linked to sensitivity of streamflow simulations to the drainage modifications in the general circulation induced by density of river basins. A physically based spatially pronounced sea surface temperature (SST) anoma- distributed hydrologic model developed by Yildiz lies in the tropical Pacific (Trenberth and Guillemot, (2001) was applied to the Monongahela river basin 1996). in the USA for the simulation of 1988 (a dry year) During the summer of 1988, a strong La Niña was and 1993 (a wet year) hydrologic regimes for a se- underway, with below normal SSTs in the eastern lected period between April and July. Three dif- tropical Pacific, while the summer of 1993 was char- ferent stream network configurations, which actu- acterized by conditions of a mature El Niño with pos-

85 YILDIZ itive SST anomalies over the same region. As noted Further, their physical parameterizations are valid by Trenberth and Guillemot (1996), these in turn af- in small-scale homogeneous media, and thus they fected the distribution of the extratropical jet stream only can be an approximate representation of the and mid-latitude storm track, thus causing anoma- hydrologic processes of a real landscape. Conse- lous circulations over the continental United States. quently, such models can not reproduce spatial het- Severe drought conditions during the summer of 1988 erogeneities in hydrologic system responses by us- afflicted much of the continental United States, espe- ing basin-averaged parameters (Abbott et al., 1986; cially the Great Plains and the Midwest. In the lower Beven, 1989). Mississippi Valley, rainfalls were at record lows from Incorporating detailed information on climate, April through June 1988. The most intense period soil, vegetation, and digital elevation a physically of drought and above normal atmospheric tempera- based spatially distributed hydrologic model devel- tures occurred in June 1988, but the conditions lead- oped by Yildiz (2001) was used in the model experi- ing up to the spring-summer drought were in place ments. With a simple, yet physically realistic repre- as early as March. The heat waves that accompa- sentation of surface-subsurface flow interactions, the nied the dryness extended throughout the summer, model couples an existing land surface model (De- although the weather patterns and rainfall returned vonec and Barros, 2002) with a surface flow routing to nearly normal in July. model and a lateral subsurface flow routing model The 1993 summer flooding in the Mississippi river (Figure 1). At the land-atmosphere interface water basin was produced by one of the largest rainfall and energy fluxes in the vertical direction are calcu- anomalies of the century. Heavy rainfall that per- lated by the land surface model through the use of sisted through June and July caused record high simplified conceptual descriptions of the physics, the river levels in the central United States. The total so-called parameterization schemes. A vertical soil rainfall over the summer period was twice as large as column is discretized into a number of layers with a the normal value. During the spring of 1993, rain- thin superficial layer at the top to function as the in- fall in the central United States was already above terface between the ground and the atmosphere, and normal, and the soil moisture levels were near satu- other deeper layers to store water and energy. The ration. Therefore, this region was poised for poten- surface of the soil is subdivided into vegetation and tially severe flooding prior to the onset of excessive bare soil areas. and localized rainstorms at the beginning of June.

Atmospheric Forcing Hydrologic model description

Physically based spatially distributed hydrologic DEM, Soil and Vegetation Interception and Information models have become an important tool for simu- Transpiration Sensible and Latent Heat Fluxes lating the effects of spatial heterogeneities in wa- Evaporation tersheds by utilizing physical parameters that have physical significance and represent spatial variability. They can easily incorporate detailed information [Surface Flow Routing Model] on topography, soil, vegetation, and climate from Infiltration digital and remotely sensed data resources. During Overland Flow recent decades, several physically based distributed hydrologic models (Abbott et al., 1986; Grayson et al., 1992; Johnson and Miller, 1997; Biftu and Gan, Interflow 2001, among others) have been developed for various Baseflow hydrologic applications in watersheds. [Subsurface Flow Routing Model] Due to the physical basis of the approach and the increasing availability of digital and remotely sensed Figure 1. Structure of the hydrologic model. spatial data, physically based distributed models have some advantages over simple lumped concep- Excess rainfall on the land surface is routed by tual models. Historically, conventional lumped mod- the surface flow routing model, which relies on an els generally have not incorporated spatially vari- algorithm using a single down-slope flow direction to able data including topography, soil and vegetation. facilitate the simulation of surface flow. Assuming a

86 YILDIZ linear flow surface across grid cells a one-dimensional Model study area kinematicwaveapproachisemployedinoverland routing to simulate the inflow and outflow discharges The Monongahela river basin is located on the west- for each grid cell. A modified Muskingum-Cunge ern slopes of the Appalachian Mountains (38.56N- method of variable parameters developed by Ponce 40.47N, 79.07W-80.76W) with portions in Pennsyl- and Yevjevich (1978) is applied to route the water vania and West Virginia. The basin is a tributary through the channel network to the basin outlet. Fi- of the Ohio river basin and has a drainage area of 2 nite difference approximations were used in numeri- approximately 13,875 km with outlet at Elizabeth, cal solutions of routing equations and the time-step PA. The actual stream network includes the West is adoptive, changing with hydraulic conditions on Fork, Tygart Valley, Cheat, and Monongahela rivers the hillslopes and in structures. and their tributaries. As part of the Appalachian Plateau, the basin Subsurface flow (i.e. interflow and baseflow) is is characterized by strong spatial variability in the routed in the lateral directions by the subsurface soil-terrain-hydrogeology system. Elevations in the flow routing model. A multicell approach proposed basin range from about 400 to 1200 m, being great- by Bear (1979) for aquifer systems was adopted for est in the southern mountainous areas and lowest subsurface flow routing. Therefore, water balance in the northern areas. At elevations above 400-500 equations are written for every grid cell and the sys- m, the bedrock is highly dissected, and consists of tem of equations is solved simultaneously for the en- sandstone with almost flat-lying layers of shale, clay, tire aquifer system by finite difference approxima- stone, and dense limestone. The soil layers above the tions. Given the river stage in the channel, the bedrock are very thin, and thus most of the rainfall flux between the channel and the ground water sys- runs off the slope. The small amounts of water that tem is determined at the end of each time step. infiltrate move vertically through fractures, and then In the model, groundwater divides are assumed to move horizontally through sandstone or coal layers correspond with the digital elevation model (DEM)- over large distances until they find another region of derived basin boundaries, and thus there is no inter- fractures, or an unconfined flow region such as collu- action between the local and regional groundwater vium and alluvium deposits. Accordingly, the base system. In addition, the water table is assumed to flow and interflow is very small during non-rainy pe- follow the topographic surface slope. riods in the warm season. At low elevations, pro- The current version of the model does not have ductive unconsolidated alluvial aquifers ensure sig- a dynamic vegetation component but vegetation can nificant and sustained baseflow and interflow contri- be dynamically introduced into the model simula- butions during summer months (Trapp and Horn, tions through the adaptive assimilation of remotely 1997). sensed or digital data. The vegetation cover in the watershed area also The stream network of the watershed is con- presents significant spatial variability with a predom- structed from DEM using a threshold value of the inance of deciduous trees at high altitudes and short flow contributing area and is optimized through the grass and crops at low altitudes. A small fraction of visual comparison with the actual stream network. the southeastern part is covered by coniferous trees, Specifically, a pixel with a flow contributing value while a narrow band of bare ground can be found lower than the threshold value is treated as a plane along the northeast-southwest direction. pixel; otherwise it is treated as a channel pixel. The regional climate is humid to temperate, with The choice of threshold value is important in ap- topographic difference influences leading to local proximating the actual shape of the stream network anomalies. The average annual temperature is about as well as in obtaining accurate streamflow hydro- 9 ◦C. Mean monthly temperatures range from -2 to graphs. Several techniques are presented in the lit- 22 ◦C. Average annual precipitation is 1067 mm and erature for simulating stream networks from DEMs ranges from 940 mm in northern areas to 1524 mm in (Montgomery and Foufoula-Georgiou, 1993). The the southern mountainous areas. Precipitation dur- most common technique is to choose an arbitrary ing the winter is cyclonic in origin, whereas thunder- threshold value on the basis of the visual similar- storms are responsible for most of the summer rain- ity between the extracted network and topographic fall. The average annual runoff (1951-1980) ranges maps. The reader is referred to Yildiz (2001) for from 635 to 1016 mm in the mountainous southeast- further details on the model’s structure. ern areas and from 458 to 660 mm elsewhere. The

87 YILDIZ average annual recharge is estimated to range from rived from the ancillary data using digital and re- 200 to 378 mm. The remainder of the average an- motely sensed data resources. Specifically, soil pa- nual precipitation is estimated as evapotranspiration rameters including hydraulic conductivity, porosity, ranging from 90 to 410 mm across the north-south field capacity and wilting point were obtained from direction (McAuley, 1995). the STATSGO data base, which was designed pri- marily for regional, multi-county, river basin, state, Data description and multi-state resource planning, management, and monitoring (USDA, 1995). The dominant soil tex- Using 3-arc second DEM data (approximately 100 ture in the basin was silt loam, while loam and sandy m) from the United States Geological Survey loam were found scattered across the river basin, es- (USGS) watershed boundary delineation and stream pecially in the south. network construction were performed at 1-km spa- Vegetation was included dynamically in the tial resolution. Therefore, the original DEM data hydrologic model utilizing time-series of remotely (i.e. 3-arc second) were aggregated into 1-km spatial sensed data. Vegetation characteristics including leaf scale. area index (LAI) and fractional vegetation cover- The hydrologic model was driven by atmospheric age (Fr) were estimated by parameterizations (LAI: forcing data including air temperature, pressure, hu- Choudhury et al., 1994; Fr: Carlson and Ripley, midity, wind velocity, and shortwave and longwave 1997) using normalized vegetation difference index radiations obtained from regional climate forecasts. (NDVI) data from the Advanced Very High Resolu- The data were produced by the National Center tion Radiometer (AVHRR). Given the soil and vege- for Atmospheric Research (NCAR) Regional Climate tation information, the other model parameters were Model (RegCM2) for spring and summer 1988 and selected from the literature (albedo: Dingman, 1994; 1993 periods over the Midwest United States. The roughness length and minimum stomatal resistance: climate model was driven at the lateral boundaries Dickinson et al., 1993; Manning’s roughness coeffi- by European Center for Medium Range Forecast cients: Chow, 1959). (ECMWF) data analyses and model outputs were produced at a temporal resolution of 6 h for the pressure and 3 h for the remaining data sets at 25-km Streamflow simulations of the 1988 and 1993 spatial scale (Jenkins and Barron, 1997). The cli- hydrologic regimes mate forecast data were downscaled from 25-km to 1-km spatial resolution with a bilinear interpolation Hydrologic model simulations of streamflow hydro- scheme. The downscaled data were further linearly graphs in the basin were performed at 1-km spatial interpolated into a 1-h temporal scale. scale at an hourly time step for selected spring and Although the RegCM2 precipitation exhibited a summer periods. As depicted in Figure 2, 3 differ- close temporal correlation with the basin averaged ent stream network configurations were developed for observed precipitation, the climate model simulated both 1988 and 1993 simulations using flow contribut- excessive precipitation during the entire simulation ing threshold values of 25, 15 and 5 km2, which pro- − period. Therefore, observed precipitation of 14 point duced drainage densities of 0.2, 0.24 and 0.38 km 1, measurements within the basin for two 5-month pe- respectively. The simulated streamflows were com- riods between April and August at an hourly time pared against the observations at the outlet of the step were used in model simulations. Spatially dis- river basin on a daily basis. tributed precipitation over the entire river basin For both 1988 and 1993 simulations, the soil col- was obtained by interpolating techniques using a umn was assumed fully saturated at the beginning of modified Thiessen polygon approach in which each the simulation period. Soil moisture can be updated Thiessen polygon is represented by a raingauge, and as new soil moisture information becomes available. thus, at a given time step, rainfall is uniform over The hydrologic model was initialized for a period of a Thiessen polygon but spatially variable over the 1 month (spin-up period) at the beginning of the entire river basin. The standard Thiessen poly- simulation in order to allow the state variables to gon method was modified in order to include oro- reach equilibrium conditions. The model was not graphic precipitation effects, especially during the calibrated; that is, the physically based model pa- spring months. rameters extracted from the ancillary data were not The physically based model parameters were de- submitted to optimization, because the simulation

88 YILDIZ years of 1988 (a dry year) and 1993 (a wet year) rep- showed that the model’s sensitivity to model param- resented 2 extreme hydrologic regimes. Bindlish and eters changes as the climate regime changes. It also Barros (2000) showed that the calibration of model showed that spatial and temporal variability can af- parameters is particularly sensitive to the underlying fect sensitivity significantly. During the dry year of climate regime, and thus calibration does not lead to 1988, vegetation properties of fractional vegetation an improved model response. coverage and leaf area index had significant effects on Using the fractional factorial design method (Box model results, suggesting that hydrologic processes et al., 1978) sensitivity testing of the hydrologic of evaporation and transpiration are expected to play model to selected model parameters listed in Table an important role in such a dry climate regime. Dur- 1 showed that the impact of vegetation is significant ing the wet year of 1993, in addition to the vegetation on the hydrology of the Monongahela river basin to parameters of Fr and LAI, soil hydraulic conductivity different hydroclimatological extremes. Evaluation was of primary importance due to higher soil water of the model sensitivity analysis in 1988 and 1993 availability (Yildiz, 2001).

Figure 2. Comparison of the stream networks for the threshold values of the ﬂow contributing area used in the delineation of the stream network: (a) 25 km2,(b)15km2,and(c)5km2.

89 YILDIZ

0 1

/s) 1200 2 3 100 observed streamflow 3 simulated streamflow 4 800 5 600

400 (a) Precipitation (cm) 200

Stream Discharge (m Stream Discharge 0 01 Apr. 15 Apr. 01 May 15 May 01 June 15 June 27 June

0 1

/s) 1200 2 3 100 3 4 800 5 600

400 (b) Precipitation (cm) 200

Stream Discharge (m Stream Discharge 0 01 Apr. 15 Apr. 01 May 15 May 01 June 15 June 27 June

0 1 2

/s) 1200 3 3 100 4 800 5 600 400 (c) Precipitation (cm) 200

Stream Discharge (m Stream Discharge 0 01 Apr. 15 Apr. 01 May 15 May 01 June 15 June 27 June Time (Day) Figure 3. Observed and simulated streamﬂow hydrographs at Elizabeth in 1988 for the threshold values of the ﬂow contributing area used in the delineation of the stream network: (a) 25 km2,(b)15km2,and(c)5km2.

Table 1. Selected model parameters for the model sensitivity analysis.

Parameter Name Classiﬁcation Leaf area index Land use/Land cover Fractional vegetation coverage Land use/Land cover Root depth Vegetation Minimum stomatal resistance Vegetation Albedo Land use/Land cover Roughness length Land use/Land cover Soil ﬁeld capacity Soil hydraulics Soil wilting point Soil hydraulics Hydraulic conductivity Soil hydraulics

90 YILDIZ

Discussion of Results and Conclusions of mean, standard deviation, root mean square error and bias, but relatively lower coefficients of variation The 1988 and 1993 model simulations of streamflow (Table 2). As shown in the figures, substantial dif- hydrographs performed for the prescribed stream ferences between the observed and simulated peak network configurations along with the observations flows, especially during the spring season, were ob- are shown in Figures 3 and 4, respectively. Compar- tained as a result of a drainage density increase. ison of the simulated and observed streamflow hy- This outcome can be attributed to an increase in drographs clearly reveals that the hydrologic model the number of channel pixels, and thus an increase in response changes significantly for drainage densities the subsurface flow contribution (i.e. interflow and of 0.2, 0.24 and 0.38 km−1.Inbothyears,thesimu- baseflow combined) to the total streamflow. In fact, lated streamflows steadily increased as the drainage comparison of the ratio of the model simulated sub- density of the river basin increased. The hydro- surface flow to the total streamflow for the prescribed logic model generally overestimated the stream dis- stream network configurations indicated that an in- charges in response to an increase in drainage den- crease in drainage density resulted in an increase in sity, producing relatively higher streamflow statistics subsurface flow response (Figure 5).

0 1

/s) 1200 2 3 100 observed streamflow 3 simulated streamflow 4 800 5 600

400 (a) Precipitation (cm) 200

Stream Discharge (m Stream Discharge 0 15 Apr. 01 May 15 May 01 June 15 June 01 July

0 1

/s) 1200 2 3 100 3 4 800 5 600

400 (b) Precipitation (cm) 200

Stream Discharge (m Stream Discharge 0 15 Apr. 01 May 15 May 01 June 15 June 01 July

0 1 2

/s) 1200 3 3 100 4 800 5 600 400 (c) Precipitation (cm) 200

Stream Discharge (m Stream Discharge 0 15 Apr. 01 May 15 May 01 June 15 June 01 July Time (Day)

Figure 4. Observed and the simulated streamﬂow hydrographs at Elizabeth in 1993 for the threshold values of the ﬂow contributing area used in the delineation of the stream network: (a) 25 km2,(b)15km2,and(c)5km2.

91 YILDIZ

(01 April-27 June 1988) 100

60 (a) Ratio (%) 40 threshold: 25 km2 threshold: 15 km2 threshold: 5 km2 20 01 Apr. 15 Apr.01 May 15 May 01 June 15 June 27 June

(09 April-01 July 1993) 100

60 Ratio (%)

40 (b)

20 15 Apr.01 May 15 May 01 June 15 June 01 July Time (Day) Figure 5. Comparison of the ratio of subsurface flow to streamflow for the threshold values of the flow contributing area of 25, 15, and 5 km2 in (a) 1988 and (b) 1993.

Table 2. Streamflow statistics of the observed and simulated streamflow hydrographs for flow contributing threshold values of 25, 15, and 5 km2.

1988 1993 Obs. 25 km2 15 km2 5km2 Obs. 25 km2 15 km2 5km2 Mean1 143.5 174.2 205.4 278.3 140.5 172.5 196.6 252.3 Std. Dev.2 82.4 85.3 86.7 99.5 100.5 91.5 102.5 129.7 CV3 0.57 0.49 0.42 0.36 0.72 0.53 0.52 0.51 RMSE4 67.5 84.5 151.5 57.5 68.4 127.4 Bias5 30.5 62.5 135.5 33.1 44.5 110.5

1 Arithmetic average 2 Standard deviation

3 Coeﬃcient of variation (Std. Dev./Mean) n 1/2 4 1 È − 2 Root mean square error deﬁned by RMSE = n [Qs(i) Qo(i)]

i=1

n n È 5 1 È − 1 Bias deﬁned by Bias = n Qs(i) n Qo(i) in which Qs(i) and Qo(i) are the simulated and observed streamﬂow i=1 i=1 rates respectively, and n is the number of items of data.

92 YILDIZ

This simple test of drainage density effect on hy- model results suggested that drainage density can drologic response indicated the importance of stream significantly affect model response, and that it can network construction for use in hydrologic model be a source of ambiguity in model calibration. simulations of river basins. The evaluation of the

References

Abbott, M. B., Bathurst, J.C., Cunge, J.A., Grayson, R. B., Moore, I. D. and McMahon, T. O’Connell, P.E. and Rasmussen, J., “An Introduc- A., “Physically Based Hydrologic Modeling 1. A tion to the European Hydrological System-Systeme Terrain-based Model for Investigative Purposes”, Hydrologique Europeen, ‘SHE’, 1: History and Phi- Water Resources Research, 20, 10, 2639-2658, 1992. losophy of a Physically-Based Distributed Modeling Horton, R.E., “Drainage Basin Characteristics”, System”, Journal of Hydrology, 87, 45-59, 1986. Transactions,American Geophysical Union, 13, 350- Bear, J., Hydraulics of Groundwater, McGraw-Hill, 361, 1932. Inc., London, New York, 1979. Jenkins G.S. and Barron, E.J., “Global Climate Beven, K., “Changing Ideas in Hydrology: The Case Model and Coupled Regional Climate Model Sim- of Physically Based Models”, Journal of Hydrology, ulations over the Eastern United States: GENESIS 105, 157-172, 1989. and RegCM2 Simulations”, Global and Planetary Biftu, G.F. and Gan, T.Y., “Semi-Distributed Phys- Change, 15, 3-32, 1997. ically Based Hydrologic Modeling of the Paddle River Basin, Alberta Using Remotely Sensed Data”, Johnson, D.L. and Miller, A.C., “A Spatially Dis- Journal of Hydrology, 244, 137-156, 2001. tributed Hydrologic Model Utilizing Raster Data Structures”, Computer and Geosciences, 23, 3, 267- Bindlish, R. and Barros, A.P., “Disaggregation of 272, 1997. Rainfall for One-Way Coupling of Atmospheric and Hydrological Models in Regions of Complex Ter- McAuley, S.D., “National Water Quality Assess- rain”, Global and Planetary Change, 25, 111-132, ment Program: The Allegheny-Monongahela River 2000. Basin”, NAWQA Fact Sheet, FS-137-95, United Box, G.E.P., Hunter, W.G. and Hunter, J.S., Statis- States Geological Survey, 1995. tics for Experimenters: An Introduction to Design, Melton, M. A., “An Analysis of the Relations among Data Analysis and Model Building, Wiley and Sons, the Elements of Climate, Surface Properties, and New York, 1978. Geomorphology”, Technical Report 11, New York: Carlson, T.N. and Ripley, D.A., “On the Relation Department of Geology, Columbia University, 1957. between NDVI, Fractional Vegetation Coverage, and Montgomery, D.R. and Foufoula-Georgiou, E., Leaf Area Index”, Remote Sensing of Environment, “Channel Network Source Representation Using 62, 241-252, 1997. Digital Elevation Models”, Water Resources Re- Choudhury, B.J., Ahmad, N.U., Idso, S.B., Regi- search, 29, 12, 3925-3934, 1993. nato, R.J. and Daughtry, C.S.T., “Relations Be- tween Evaporation Coeﬃcients and Vegetation Indi- Ponce, M.V. and Yevjevich., V., “Muskingum- cies Studied by Model Simulations”, Remote Sens- Cunge Method with Variable Parameters”, Jour- ing of Environment, 50, 1-17, 1994. nal of the Hydraulics Division, ACSE, 104:12, 1663- 1667, 1978. Chow, V.T., Open Channel Hydraulics, McGraw- Hill, New York, 1959. Trapp, H., Jr. and Horn, M.A., “Groundwater Atlas of the United States: Delaware, Maryland, New Jer- Devonec, E. and Barros, A.P., “Exploring the Trans- sey, North Carolina, Pennsylvania, Virginia, West ferability of a Land Surface Hydrology Model”, Virginia”, HA 730-L United States Geological Sur- Journal of Hydrology, 265, 258-282, 2002. vey, 1997. Dickinson, R.E., Henderson-Sellers, A. and Kennedy, P.J., “Biosphere-Atmosphere Transfer Trenberth, K.E. and Guillemot, C.J., “Physical Pro- Scheme (BATS) Version 1e as Coupled to the cesses Involved in the 1988 Drought and 1993 Floods NCAR Community Climate Model”, National in North America”, Journal of Climate, 9, 1288- Center for Atmospheric Research (NCAR), USA, 1298, 1996. Technical Note, NCAR/TN-387+STR., 1993. USDA (United States Department of Agriculture), Dingman, D.R., Physical Hydrology, Prentice Hall, “State Soil Geographic (STATSGO) Data Base: Inc., New Jersey, 1994. Data Use Information”, Pub. No. 1492, 1995.

93 YILDIZ

Yildiz, O., “Assessment and Simulation of Hydro- Pennsylvania State University, University Park, PA, logic Extremes by a Physically Based Spatially 2001. Distributed Hydrologic Model”, PhD. Thesis, the