THE BIGGER THE BETTER? HOW SPATIAL RESOLUTION AFFECT RUNOFF MODELING AND WATERSHED DELINEATION David Alvarez CDM 1715 North Westshore Boulevard Suite 875 Tampa, FL 33607 email: alvarezd@cdm.com Barrett Goodwin CDM 8140 Walnut Hill Lane, Suite 1000 Dallas, Texas 75231 e-mail: goodwinba@cdm.com ABSTRACT The modeling of Hydraulic Systems has become increasingly more sophisticated with the advancement in technology. This technology is becoming more GIS based for use in hydraulic modeling. The purpose of this study is to understand how different spatial resolutions in a DEM affect the delineation of watersheds, stream networks and runoff. This study area is located in the Dallas County and Collin County portion of the Rowlett Creek watershed. For this paper HEC-HMS, developed by the Corps of Engineers, is been use to calculate runoff. HEC- GeoHMS and ArcHydro are extensions for Arcview which derives the parameters necessary for the HEC-HMS model. Several terrain models with different pixels sizes are generated from Lidar data. Each model's results were compared based on magnitude of differences between various delineations of the watershed, stream network and runoff. Understanding how the pixel size affects the model will give us an understanding if is worth investing in high accuracy data for simple watershed modeling studies. INTRODUCTION When rain or snow falls onto the earth, it just doesn't sit there -- it starts moving according to the laws of gravity. A portion of the precipitation seeps into the ground to replenish Earth's ground water. Most of it flows downhill as runoff. Surface runoff is the volume of excess water that runs off a drainage area. Rainfall is the primary source of water that runs off the surface of small rural watersheds. The main Factors affecting the volume of rainfall that runs off are the kind of soil and the type of vegetation in the watershed, but Factors that affect the rate at which water runs off are the watershed is topography and shape along with conservation practices on a watershed. Runoff is extremely important in that not only does it keep rivers and lakes full of water, but it also changes the landscape by the action of erosion. Water professionals need to be able to manage surface and groundwater resources over the scale of an entire watershed. The effects of land cover, vegetation, soil type, topography, geology, water quality, and other factors must be considered in order to make sound management decisions. This decision has been in a way simplify by the use of geographic information system (GIS). Put simply, a GIS is a system of computer software, hardware, and data, combined with qualified people to assist with manipulation, analysis, and presentation of information that is tied to a spatial location.. The GIS user can access and manipulate information associated with Geographic features and look for spatial and temporal patterns and relationships. The application of Geographic Information systems (GIS) in the field of hydrology has grown significant in the past decade. Advances in the technology increased professional awareness have greatly improve the accuracy, functionality and commonality of GIS in water resources management and related fields Today in the market there are several applications that can be use to implement the runoff part of the project. (WMS, HEC-HMS, BASINS, TRC-55, Mike11 and more). Each of this application has it strength and weakness. In our case we decided to go with the free side of the market The Hydrologic Modeling System (HEC-HMS). For the GIS part we used ArcGIS Desktop 9.2 with the extensions (Spatial Analyst, Geostatistical and 3D Analyst).
As mention before one of the factors that affect the runoff is the topography. Since the USGS started the release of the National Elevation Dataset (NED) to the public. The NED has a resolution of 1 arc-second (approximately 30 meters) or 1/3 arc-second (approximately 10 meters) The GIS community has used this data for runoff modeling and more, but in the last few year a new source of data came alone, that change our way of obtaining elevation data. Light Detection and Ranging (LIDAR). This data is more accurate and more detail than the NED. In this paper we want to try to understand how spatial resolution in the Digital Elevation Model (DEM) Topography- will affect the relationship with the runoff if all other variables are kept constant. LOCATION The majority of the Rowlett Creek watershed is within Collin County Texas with a small portion in Dallas County Texas as shown in figure 1. Collin County is located in northeastern Texas, approximately thirty miles south of the Red River. McKinney, the county seat, is thirty-four miles northeast of Dallas. The county is bordered by the following counties: Grayson to the north, Fannin, to the northeast, Hunt to the east, Rockwall to the southeast, Dallas to the South, and Denton to the west. The county is approximately 889 square miles in size and has a population of 615,200 in 2004. There are 29 communities within Collin County with the larger communities located in the southwest portion of the county. All of the Rowlett Creek watershed is considered urban or developed. Primary landuse/landcover consist of concrete, housing, businesses, and industry. The county lies within the Blackland Prarie region. The surface of the county is generally level to gently rolling, with an elevation ranging from 450 to 700 feet above sea level. Deep clayey soils over marl and chalk surface the central and western part of the county. Dark loamy alluvial soils, subject to flooding during the rainy season, lie in the eastern section. Limestone and sand for making cement are the only mineral resources. Temperatures range from an average high of 96 F in July to an average low of 34 in January. Rainfall averages just under thirty-five inches a year, and the growing season extends for 237 days. Figure 1. Watershed general location.
SOFTWARE Watershed Delineation Hydro was developed by Dr. David Maidment (University of Texas, Austin Center for Research for Water Resources), in collaboration with several prominent universities (Consortium of Universities for the Advancement of Hydrologic Sciences) and ESRI, as a mapping software for water resource professionals. It generates watershed boundaries based on hydrology, and it allows for geospatial representation of surface water bodies and integration with hydrologic and hydraulic modeling. The Geospatial Hydrologic Modeling Extension (Geo-HMS) is an ArcGIS application that allows developing a number of hydrologic modeling inputs. Analyzing digital terrain information, HEC-GeoHMS transforms the drainage paths and watershed boundaries into a hydrologic data structure that represents the watershed response to precipitation. Runoff HEC-HMS is designed to simulate the precipitation-runoff processes of dendritic watershed systems. The program is a generalized modeling system capable of representing many different watersheds. A model of the watershed is constructed by separating the hydrologic cycle into manageable pieces and constructing boundaries around the watershed of interest. Any mass or energy flux in the cycle can then be represented with a mathematical model. In most cases, several model choices are available for representing each flux. Each mathematical model included in the program is suitable in different environments and under different condition. INPUTS Creation of the Digital Elevation Model (DEM) Digital Elevation Models (DEMs) are important tools in hydrologic research and water resources management owing to the relevance that geo-morphological features intrinsic in the DEMs have for the simulation of important water flow processes such as surface runoff, evaporation and infiltration. Digital For the creation of the Digital Terrain Model (DEM) we used the Geo-Statistical extension from ESRI. This extension has several interpolation methods that can be use to generate surface models. There is a huge amount of literature that compares the different interpolation and how it does affect result when the terrain model is created. Each of the interpolation methods has it advantages and it drawbacks. Irrespective of the landscape morphology and surface area, few differences existed between the techniques under study provided that the sampling density was high. This could have been foreseen since the greater the sampling density is, the lower the impact of the interpolation technique is, simply due to the mechanical decrease of space between known values * This implies that for lower values of sampling density, the accuracy of height estimation is more dependent on the choice of interpolation techniques.. This has been said the use IDW is a easy choices. This is due that it was a simpler and more accurate interpolation method than kriging for DEM development, likely due to the high density of the lidar data points. The technique takes into account only some adjacent data points, and thus performs well even for complex landforms if data density is high **. Soils Soil Survey Geographic (SSURGO)."The STATSGO data base was designed primarily for regional, multi-state, river basin, state and multi-county resource planning, management, and monitoring ***. The SSURGO data base provides the most detailed level of information and was designed primarily for farm and ranch, landowner/user, township, county, or parish natural resource planning and management. Using the soil attributes, this data base serves as an excellent source for determining erodible areas and developing erosion control practices; reviewing site development proposals and land use potential; making land use assessments and chemical fate assessments; and identifying potential wetlands and sand and gravel aquifer areas. * Fisher, N.I., Lewis, T., Embleton, B.J.J., 1987. Statistical Analysis of Spherical Data. Cambridge University Press, Cambridge. 329 pp. ** ASPRS Vol 72 No 11 *** http://www.ftw.nrcs.usda.gov/pdf/ssurgo_db.pdf
Land Cover/Land Use The Land Use file was created form base on photo-interpretation. Features were rectified to the latest aerial photography available for each county. Collin, Dallas, Denton, Rockwall & Tarrant counties were based on orthos with a relative accuracy of 2-foot. RUNOFF MODEL HMS computes the runoff volume by computing the volume of water that is intercepted, infiltrated, stored, evaporated or transpired, and subtracts it from the precipitation ****. HEC-HMS is divided in to three components. Basin model - contains the elements of the basin, their connectivity, and runoff parameters Meteorologic Model - contains the rainfall and evapotranspiration data Control Specifications - contains the start/stop timing and calculation intervals for the run. Infiltration, interception, evaporation, storage, and transpiration are collectively known as losses in HMS. HMS has the following loss models to account for cumulative losses: Initial/Constant SCS (Soil Conservation Service) Curve Number Deficit/Constants Green and Ampt Gridded SCS Curve Number Soil Moisture Accounting Method (SMA) Gridded SMA No Loss rate HMS computes the precipitation loss for each computation time interval. This loss is subtracted from the total precipitation depth for that interval *. This gives precipitation excess that is considered constant over the entire study area and hence gives a volume of runoff. SCS Curve Number Loss Method The Soil Conservation Service (SCS) developed the SCS Curve Number method to compute abstractions from storm rainfall. This model is used to estimate the excess in the precipitation, as a function of cumulative precipitation, soil cover, land use and antecedent moisture. This is defined as: 2 ( P Ia) Pe = P Ia + S where: Pe=Cumulative excess rainfall depth P= Cumulative depth of precipitation Ia= initial Loss/ abstraction S= potential maximum retention SCS also developed an empirical relation for initial abstraction from an analysis of small watersheds and is defined as: (1) I a = 0.2* S (2) The following relationship relates the maximum retention, S, to soil and cover conditions of the watershed to the curve number, CN: **** U.S Army Corps of Engineering (2001) Hydrological Modeling System, HEC-HMS, User manual * U.S Army Corps of Engineering (2001) Hydrological Modeling System, HEC-HMS, User manual
1000 S ( in) = 10 (3) CN CN Calculations The Soil Conservation Service (SCS) Curve Number was developed as an index that represents the combination of hydrologic soils group, land use, and treatment class. The Curve Number defines the impervious character of a watershed with 100 being impervious and 0 being totally pervious. CN is a function of three factors, which are soil group, cover complex, and antecedent moisture conditions. More that 4000 soils have been classified based on their runoff potential and have been grouped into four Hydrologic Soils Groups (HSG). The soils are grouped according to the intake of water when the soils are thoroughly wet and receive precipitation from long duration storms. The four hydrologic soil groups are A, B, C and D as shown in Table 2 and Table 3. GROUP CHARACTERISTICS Table 1. HSG and soil characteristics * Group A B C D Characteristics Having a higher filtration rate (low runoff potential) when thoroughly wet. These consist mainly of deep, well drained to excessively drained sand or gravely sand. These soils have a high rate of water transmission. Having a moderate infiltration rate when thoroughly wet. These consist chiefly of moderately deep or deep, moderately well drained or well drained soils that have moderately fine texture to moderately coarse texture. These soils have a moderate rate of water transmission. Having a slow infiltration rate when thoroughly wet. These consist chiefly of soils having a layer that impedes the downward movement of water or soils of moderately fine texture or fine texture. These soils have a slow rate of water transmission. Soils have a very slow infiltration rate (high runoff potential) when thoroughly wet. These consist chiefly of clay that has high shrink-swell potential, soils that have a permanent high water table, soils that have a clay pan or clay layer at or near the surface, and soils that are shallow over nearly impervious material. Table 2. HSG and their runoff potential * Group A B C D Infiltration Rate Greater than 0.30 in/hr 0.15-0.30 in/hr 0.05-0.15 in/hr These soils have a very slow rate of water transmission (0-0.05 in/hr). As a result of urbanization, the soil profile may be considerably altered and the listed soil group classification may no longer apply. In these circumstances, Table 3 should be used to determine Hydrologic Soils Groups (HSG) according to the texture of the new surface soil, provided that significant compaction of the soil has not occurred: * http://www.chathamtownship.org/natres_5_soils.pdf
Table 3. HSG and Texture * Group A B C D Texture Sand, Loamy Sand or Sandy Loam Silt Loam or Loam Sandy Clay Loam Loam, Silty Clay Loam, Sandy Clay, Silty Clay or Clay An exaggeration in the estimation of the curve numbers may yield a result that does not accurately represent the actual impervious character of the watershed. Inaccurate curve number estimation may lead to an inaccurate S (equation 3) calculation, which will directly affect the depth calculations. The direct estimation of the CN works fine if the area of the project is small that the land use and soils cover can be assumed to be homogenous over the entire area. In a watershed containing a wide variety of land use and soils types, such as this study area, a composite CN must be calculated using the following equation: i A CNi CN composite = (4) A i where: CNcomposite= The Composite CN that is used for runoff volume calculations in HEC-HMS i= an index of watershed subdivisions of uniform land use and soil type Ai= drainage area of subdivision i. Table 4. Cover types, HSGs and their respective numbers Cover Type Curve Number for HSG A B C D Residential High 77 85 90 92 Residential Low 53 69 80 85 Agriculture 72 81 88 91 Transportation 98 98 98 98 Forest 25 55 70 77 Poor Wood 45 66 77 83 Commercial 89 92 94 95 Grass 39 61 74 80 Industrial 81 88 91 93 After the soil and land use are merge the CN composite is calculate for each of the Subbasins in each of the three models as shown in Table 5.
Table 5. CN composite by subwatershed for the three models 10 ft Model 30 ft Model 90 ft Model Subbasin ID CN Composite CN Composite CN Composite 1 81.9 82.1 82.1 2 80.7 80.6 80.6 3 82.9 82.9 78.7 4 79.8 79.7 83.0 5 81.3 78.6 79.7 6 78.6 81.5 81.4 7 81.5 77.0 81.5 8 78.9 81.4 75.1 9 79.4 79.5 79.4 10 81.5 81.5 81.5 11 83.0 83.0 83.0 12 80.4 80.9 80.6 13 78.8 78.7 78.6 14 84.2 84.2 84.2 15 77.7 77.7 84.3 16 84.4 84.4 77.7 17 83.7 83.7 83.8 18 78.9 79.2 79.0 19 75.2 75.2 75.0 20 83.7 83.7 83.7 21 81.3 81.3 81.3 22 80.6 80.5 80.2 23 79.0 79.2 79.2 24 82.9 82.9 81.2 25 81.9 82.1 82.9 26 83.5 83.5 83.5 27 79.5 79.5 79.7 28 76.3 76.4 76.1 29 74.3 77.9 77.9 30 77.9 74.2 74.5 31 71.9 71.4 66.6 32 79.5 79.2 78.7 33 82.0 82.1 81.8 In this case due that each model delineated the subbsains some what different, the relation between the ID linking the three models could not be the same. The table below (table 6) show which subbasin have the same subbasin ID in the three models. As show on the table the difference are not significant.
Table 6. Common Subbasins ID between the three models 10 ft Model 30 ft Model 90 ft Model CN Area Km 2 CN Subbasin ID Composite Composite Area Km 2 CN Composite Area Km 2 1 81.9 82.1 82.1 2 80.7 80.6 80.6 9 79.4 79.5 79.4 11 83.0 83.0 83.0 14 84.2 84.2 84.2 17 83.7 83.7 83.8 18 78.9 79.2 79.0 19 75.2 75.2 75.0 20 83.7 83.7 83.7 21 81.3 81.3 81.3 22 80.6 80.5 80.2 23 79.0 79.2 79.2 24 82.9 82.9 81.2 25 81.9 82.1 82.9 26 83.5 83.5 83.5 27 79.5 79.5 79.7 28 76.3 76.4 76.1 29 74.3 77.9 77.9 30 77.9 74.2 74.5 31 71.9 71.4 66.6 32 79.5 79.2 78.7 33 82.0 82.1 81.8 DIRECT RUNOFF MODEL (TRANSFORM) Excess rainfall, after flowing over the watershed surface, becomes direct runoff at the watershed outlet *.The difference between the observed total rainfall hyetograph and the excess rainfall hyetograph are abstractions or losses. Losses result from water absorption by infiltration. Losses can be determined from the observed hydrograph. There are six methods to model the direct runoff in HMS: Clark Kinematic Wave ModClark Snyder SCS User-Specific S-Graph User Specific UH Graph All watersheds have some response time that represents how long it takes for runoff to reach the main watershed outlet after it begins to rain. The most commonly used watershed response time is the basin time of concentration. The Soil Conservation Service (SCS) has developed a relationship between the basin lag time and the watershed time of concentration, which is useful for ungauged watersheds. This is defined as: * U.S Army Corps of Engineering (2001) Hydrological Modeling System, HEC-HMS, User manual
where: Tl: Basin Lag Time (Hrs) Tc: Watershed Time of Concentration (Hrs) T = 0. 6 l T c Time of Concentration The time of concentration is defined as the time required for a particle of water to travel from the hydraulically most remote part of the watershed to the main outlet. Frequently the point where the time of concentration is measured corresponds to the furthest point from the watershed outlet*. However, because travel velocities are a function of watershed slope and roughness, it is possible that the point where the time of concentration is measured may not be located the furthest away from the watershed outlet. The Soil Conservation Service has developed a time of concentration formula that includes the effect of the watershed runoff characteristics. This is defined as: where 0.8 L ( S + 1) T C = 1.667( 1900 Y 1000 S = 10 CN TC - Watershed time of concentration (Hrs) L Hydraulic length of watershed (length from the hydraulically most remote part of watershed) (ft) Y Average watershed slope (%) S - Maximum soil retention. CN Curve number Reach Routing Routing refers to the process of calculating the passage of a flood hydrograph through a system. The reach element in HMS represents the flow of water through the stream system. When a runoff hydrograph enters a channel reach, the shapes of the hydrograph will most likely change as flow travels along the channel. Generally speaking, the peak discharge is at the downstream end and is less than the peak flow at the upstream end. Furthermore, the time to peak is typically greater for the downstream hydrograph than for the upstream hydrograph, as shown in Figure 2 and Figure 3. 0.7 Figure 2. Hydrograph behavior. The hydrograph entering the channel is usually called an inflow hydrograph. The time history of flow at the downstream end of the channel is typically the outflow hydrograph. The reduction in peak discharge is call flow attenuation. The difference in time between peak discharges is usually referred to as the lag time. If there are no
flow losses (infiltration) or flow gains (lateral flow) along the channel length, then the volume of runoff will be the same for both hydrographs. D: Duration of the unit hydrograph Tl: Basin lag time (Hrs) Figure 3. Hydrographs. The primary reason that the peak discharge is attenuated is due to storage effects in the channel. The smaller the amount of storage in the channel, the less the amount of flow attenuation. The major reason that there is a lag between flow peaks is due to the travel time through the channel. Shorter channels or channels having high velocities will have a smaller lag time (see Figure 4). Figure 4. Lag time Concepts. HEC-HMS has several different reach routing models that can be used in the simulation model. Also, HEC- HMS permits runoff hydrographs to be routed through channels using the following approaches: Modified plus method Muskingum method Simple Lag model Muskingum-Cunge method Kinematic Wave Method In this case the simplest method is the Simple Lag model, which is appropriate in cases where the channels have little or no storage. This method translates the inflow hydrograph by a specified amount of time. The outflow hydrograph has exactly the same runoff values as the inflow hydrograph. The only difference is that they occur at a time that is equal to the time at which the inflow occurs, plus the lag time, as shown in Figure 3. For this project I used the Muskingum-Cunge method, which is based mostly on the geometry of the channel. In the late 1960 s Cunge theorized that the Muskingum method was actually a solution to the kinematic wave problem using a linear scheme. Furthermore, he asserted that the attenuation of the flood wave was due to a numerical diffusion associated with the method itself. The major difference between the Muskingum and Muskingum-Cunge method is in the way the coefficients are determined. In the Muskingum method the routing and the weighting
coefficients are assumed to remain constant over the duration of the flood event *. Thus coefficients on the update formula also remain constant. DIFFERENCE IN THE DELINEATIONS OF THE WATERSHEDS First we need to compare the stream network that was generated by each model. Doing a visual inspection of the stream network for each of the models we can not differentiate between them. The three networks look the same as shown in Figure 5. Now if a more detail visual inspection is done, we can see there is difference between three stream networks as shown in Figure 6. Figure 5. Comparison between the stream networks generated from the 10, 30 and 90 ft DEM s. * Chow V.T. and others. (1988). Applied Hydrology
Figure 6. Comparison between the stream networks generated from the 10, 30 and 90 ft DEM s. This difference can be more drastic in the meandering part of the river. As shown in Figure 6 and Figure 7 the Stream Network that was generated from the 90 ft DEM is loosing the curvature of the streams. But in a general sense the streams follow the same pattern. This simplification or generalization on the streams will reduce the Time of concentration for the model. The increase in resolution will simplified the morphology of the area, but at the same time the terrain constitution is will define; even in the ninety feet DEM most of the terrain features are preserved. Maybe in a flat area this will not be the case. In relation to the length of the stream the difference between the 10 ft and 30 ft DEM in not significant enough to conclude that it will affect result of the analysis. But that different between the 10ft and 90 ft is something to consider. This difference can be clarified in the table below. See table 7.
Figure 7. Profiles between the stream networks generated from the 10, 30 and 90 ft DEM s. Table 7. Difference between the three models DEM resolution Total Length of the streams (ft) Mean Standard deviation Number of Streams 10 ft 506422.20 15346.127488 11387.49432 33 30 ft 498469.98 15105.151046 11104.46017 33 90 ft 475158.48 14398.742119 10854.621298 33 These changes in length between each model can be associated with the increase or decrease of flow in each model, but cannot be directly correlated with the amount of runoff that passes through the outlet. Now if we compare the watershed delineation between the 10ft and the 30 ft model, there differences visually are more notable but can be assume to be minimum as shown in figure 8 and figure 9. The main differences are located at the edges of the subbsains and some in the interior subbasin. This could be due to the difference in the pixel size. Figure 8. Difference between the 10ft and 30 ft delineations.
Figure 9. Difference between the 10ft and 30 ft delineations. The difference in areas is shown on table 8. This difference between the total areas can be translated into an increase of the value of the runoff. This increase has to be related not only to the difference is area, but also to the land cover and soil type that the difference adds. Table 8. Difference between the subbasin delineations DEM resolution Total Area (ft 2 ) Mean Standard deviation Number of Streams 10 ft 3363129200.00 101913006.06 78593963.99 33 30 ft 3359814300.00 101812554.54 78999948.08 33 90 ft 3357927899.99 101755390.90 81296442.56 33
By examining the hydrograph for each of the models, each one has very similar behavior, except for the small drop in the 30 and 90 ft models as shown in figures 10, 11 and 12. This strange drop could be due to the difference in the areas, or even in the length of the stream network or possible due to the simplification of the terrain morphology due to the change in pixel size. The only significant difference will be the amount of flow that each model reaches at each peak which is significant. Figure 10. 10 ft DEM Hydrograph. Figure 11. 30 ft DEM Hydrograph.
Figure 12. 90 ft DEM Hydrograph. The total outflow will vary between models, but its variation is almost insignificant as shown in table 9 and figure 13. This will provide the understanding that the spatial resolution in this project does not affect the runoff model. More research is needed. As a general conclusion, we can infer that the resolution does affect the end results, but is not the only factor that affects the runoff. This result has to be quantified to determine how much the outflow difference is related to the DEM resolution. Table 9. Outflow variation between the models Model Total Outflow (in) Peak Outflow (cfs) 10 ft 36.28 1269518 30 ft 36.73 445571 90 ft 36.76 416829 Model vs Outflow 36.8 36.7 Total Outflow 36.6 36.5 36.4 36.3 36.2 0 20 40 60 80 100 Spatial Resolution Figure 13. Outflow variation between the models.
CONCLUSIONS In general, the resolution does affect the end result in how the watershed is delineated and the stream network is determined, but it is not the only factor that affects the runoff. This result has to be quantified, i.e., how much of the outflow difference is related to the DEM resolution. Structures also are a key element to keep in mind when hydrologic modeling is taking place. Structures affect the delineation of the watershed and the stream network. Based on the results of this project, a definitive conclusion based on the assumption that the better the resolution the better the result is not possible. REFERENCES Chaplot, Vincent, et al. Accuracy of interpolation techniques for the derivation of digital elevation models in relation to landform types and data density, Geomorphology, v. 77 issue 1-2, p. 126-141. Chow V.T. and others, 1988. Applied Hydrology, McGraw-Hill Education. Fisher, N.I., Lewis, T., Embleton, B.J.J., 1987. Statistical Analysis of Spherical Data, Cambridge University Press, Cambridge. 329 pp. Handbook of Texas Online http://www.tshaonline.org/handbook/online/index.html http://www.hec.usace.army.mil/software/hec-hms/. Su, Jason, et al., 2006. Influence of vegetation, slope and lidar sampling angle on DEM accuracy, Photogrammetric Engineering and Remote Sensing, v72, 11, pp1265-1274. U.S Army Corps of Engineers, 2001. Hydrological modeling system, HEC-HMS, User Manual. USGS web page (standards for DEM) http://geology.er.usgs.gov/eespteam/gislab/cyprus/dem_standards.htm. Vazquez, R.F, Feyen, J. Assessment of effects of DEM gridding on the predictions of basin runoff using MIKE SHE and a modeling resolution of 600 m, Journal of Hydrology, 334, 73-87.