1998 Esri User’s Conference
July 25-31, 1998
San Diego, California

 

 

HEC-PrePro v. 2.0: An ArcView Pre-Processor for
HEC’s Hydrologic Modeling System

 

Francisco Olivera, Seann Reed and David Maidment
University of Texas at Austin - Center for Research in Water Resources
Austin, Texas

  

 

Abstract

Table of Contents

1. Introduction

    HEC-PrePro v. 2.0 -- a system of ArcView scripts and associated controls -- has been developed to extract hydrologic, topographic and topologic information from digital spatial data of a hydrologic system, and to prepare an input file for the Hydrologic Modeling System (HMS) developed by the Hydrologic Engineering Center (HEC) of the United States Army Corps of Engineers.

    HMS is a computer program used to model rainfall-runoff processes in a watershed or region, and is an improved Windows version of the well-known HEC program HEC-1. One component of an HMS model is the basin file, which stores hydrologic, topographic and topologic information for the system. Based on data from GIS layers, HEC-PrePro v. 2.0 prepares a basin file in ASCII format, which when opened by HMS automatically creates a topologically correct schematic network of sub-basins and reaches, and attributes each element with selected hydrologic parameters.

    HMS is a very flexible program that allows the user to choose among different loss rate, watershed routing (i.e., unit hydrograph), and baseflow models for the sub-basins, as well as different routing methods for the streams. However, because some of these models and methods depend on hydrologic parameters that can not be extracted from readily available spatial data, HEC-PrePro v. 2.0 does not estimate parameters for all methods supported by HMS. At the moment, HEC-PrePro v. 2.0 is capable of determining parameters required by the Soil Conservation Service (SCS) curve number method for loss rate calculations, the SCS unit hydrograph model for watershed routing, and either the Muskingum or lag method for flow routing in the streams (depending on the reach length).

    Using HEC-PrePro v. 2.0, the determination of physical parameters for HMS is a simple and automatic process that accelerates the setting up of a hydrologic model for HMS and leads to reproducible results.

2. Previous work

    HEC-PrePro v. 2.0 is the synthesis of several GIS and hydrologic modeling applications developed over the last years. In this section, the existing GIS technology and HEC hydrologic modeling programs, used in Hec-PrePro v. 2.0, are reviewed.

    The raster-based GIS environment is very suitable for hydrologic modeling, mainly because raster systems have been used for years, and a mature understanding has been achieved and efficient and useful algorithms have been developed for terrain analysis. Grid systems are ideal for modeling topographically driven flow, because a characteristic of this type of flow is that flow directions do not depend on any time dependent variables, say flow or water depth. Consequently, raster GIS software includes hydrologic functions as part of its capabilities, which allow one to determine flow direction and drainage area from digital elevation models (DEM’s), and to delineate stream networks and watersheds. DEM’s are available at different resolutions: 30 meters (USGS a), 3 arc-seconds (approximately 90 m) (USGS b), and 500 meters (USGS c) for the entire United States, and 30 arc-seconds (approximately 1 Km) (USGS d) for the whole earth.

    Jensen and Domingue (1988) and Jensen (1991) outlined a grid scheme to delineate watershed boundaries and stream networks. The scheme uses digital elevation data to assign a flow direction from each cell in a grid to one of its eight neighboring cells according to the path of the steepest descent (i.e. each cell of the watershed is connected to the lowest of its neighbor cells). The cells contributing flow to the pour point can be counted, representing area, and the cells having no contributing flow define drainage boundaries. Cells having a flow accumulation in excess of a threshold value are classified as stream network cells.

    Functions to delineate streams and watersheds that use the Jensen and Domingue algorithms are available through Avenue requests in ArcView 3.0a Spatial Analyst 1.1. The Hydrologic Modeling ArcView extension, distributed by Esri with the Spatial Analyst , makes this functionality accessible to users who are not familiar with Avenue. A more specialized set of procedures that allows the user to interactively delineate watersheds is available through the Watershed Delineator ArcView extension. The Watershed Delineator was developed by the Applications Programming group at Esri for the Texas Natural Resources Conservation Commission (TNRCC). The Watershed Delineator pre-processes the terrain data, and delineates watersheds to any point, line segment, or polygon selected interactively by the user from the map. Pre-processing includes defining flow directions and drainage areas for each cell, and defining a sub-watershed base map at a specified drainage area threshold. Using pre-processed data, when the end user makes a delineation at a point, only those DEM cells within the base sub-watershed containing the selected point need to be processed, even if the selected pour-point is on the main river and many sub-watersheds are upstream of it. To provide this functionality, the drainage sequence of sub-watersheds in the base map is determined within the vector domain. The Spatial Analyst functionality to make raster-to-vector conversions (to convert strings of cells into lines and groups of cells into polygons) makes this possible. Using vector objects to represent streams (lines) and watersheds (polygons) facilitates the transfer of information to a lumped hydrologic model.

    The Hydrologic Engineering Center’s Hydrologic Modeling System (HMS) provides a variety of options for simulating rainfall-runoff processes. The basic framework for simulation of basin runoff is similar to that in HEC-1. Hydrologic elements are arranged in a dendritic network, and computations are performed in an upstream-to-downstream sequence. HEC-1, also developed by HEC, calculates discharge hydrographs for either historical or hypothetical events for one or more locations in a basin. To account -- to a certain extent -- for the spatial variability of the system, the basin can be subdivided into sub-basins with unique hydrologic parameters. Precipitation excess is transformed into direct runoff using either unit hydrograph or kinematic wave techniques. Different unit hydrograph options are available: unit hydrograph ordinates may be supplied directly by the user, or the unit hydrograph may be expressed in terms of Clark, Snyder, or Soil Conservation Service unit hydrograph parameters. The kinematic wave option permits depiction of sub-basin runoff with elements representing one or two overland-flow planes, one or two collector channels, and a main channel (DeVries and Hromadka, 1993).

    HMS is comprised of a graphical user interface (GUI), integrated hydrologic analysis components, data storage and management capabilities, and graphics and reporting facilities. The GUI provides a means for specifying model elements (i.e., sub-basins, sources, reaches, junctions, reservoirs, diversions and sinks) and their interconnectivity, inputting data for the elements, and viewing hydrographs. The HEC Data Storage System (DSS) is used for storage and retrieval of time series, and gridded data, in a manner largely transparent to the user. The execution of a simulation requires specification of three sets of data. The first, labeled basin file, contains parameter and connectivity data for hydrologic elements. The second set, labeled precipitation file, consists of meteorological data and information required to process the data. The third set, labeled control specifications file, specifies time-related information for a simulation. An HMS Project can consist of a number of data sets of each type, and a "run" is configured by selecting a basin file, a precipitation file and a control specifications file.

    Hellweger and Maidment (1997) present a GIS pre-preprocessor for HMS, HEC-PrePro v. 1.0, which identifies the seven hydrologic elements defined in HMS, and establishes their interconnectivity. The purpose of this pre-processor is to summarize the spatial data in stream and sub-basin vector GIS layers, and prepare a basin file for HMS. The program is written in both ArcInfo Arc Macro Language (AML) and ArcView Avenue. HEC-PrePro v. 1.0 identifies the intersection points of the stream and sub-basin layers as sources, sinks, or sub-basin outlets. Sub-basin elements are defined by the centroid of the polygons in the sub-basin layer, and reaches are identified as the downstream elements of sources and sub-basin outlets. Diversions are located at any point connecting one upstream reach to two or more downstream reaches, and junctions at points connecting two or more upstream reaches to one downstream reach. Enclosed polygons in the stream layer are identified as reservoirs. Elements previously classified as sub-basin outlets are combined with junctions, because they serve the same function. The ArcView version of HEC-PrePro v. 1.0 also provides the capability of transferring element attributes from the attribute tables of the sub-basin and stream layers to the HMS basin file. However, it does not include routines for computing attributes specific to a chosen hydrologic method.

    HEC-PrePro v. 2.0 combines the terrain analysis capabilities of the Watershed Delineator with the topologic analysis capabilities of HEC-PrePro v. 1.0, and adds the ability of computing parameters of the hydrologic elements. All these components together conform a very valuable tool for hydrologic modeling that allows the user to prepare the basin file for HMS from already available digital spatial data. HEC-PrePro v. 2.0 uses codes within the Watershed Delineator and HEC-PrePro v. 1.0. Modifications to the borrowed codes have been made to meet the specific objectives of this system.

3. Methodology

    HEC-PrePro v. 2.0 is an ArcView system for hydrologic modeling that performs two operations: pre-process of the entire hydrologic system, and isolation and process of a sub-system. In the pre-process, which is run only once, the stream-watershed network of the entire system is defined, both in the raster and vector domain, and attributed with hydrologic parameters. The isolation and process consists of clipping out a hydrologic sub-system and preparing an HMS basin file for it. Based on the pre-processed data layers, an unlimited number of processes can be performed. HEC-PrePro v. 2.0 can be divided into four conceptual components: (1) raster-based terrain analysis and network definition; (2) vectorization of the hydrologic elements; (3) computation of the hydrologic elements parameters; (4) isolation of a hydrologic sub-system; and (5) topologic analysis and preparation of an HMS basin file. The first three components correspond to the pre-process, and the last two to the process of a sub-system.

3.1. Pre-process of the entire hydrologic system

    Topographic analysis required to define the hydrologic system is based on the DEM. By running the flowdirection GIS function, a single downstream cell -- in the direction of the steepest descent -- is defined for each terrain cell, so that a unique path from each cell to the basin outlet is determined. This process produces a cell-network, with the shape of a spanning tree, that represents the paths of the watershed flow system. However, because a flow direction can not be determined for cells that are lower than their surrounding neighbor cells, a process of filling the spurious terrain pits is necessary before running the flowdirection function. In most cases, the existence of pits in the DEM is explained by numerical errors introduced in the process of interpolation of observed values to estimate values for each grid cell. Filling the DEM pits consists of increasing the value of the pit cells to the level of the surrounding terrain, so that water is able to flow out of the area. Once the pits have been filled and the flow directions are known, the drainage area – in units of cells – is calculated with the flowaccumulation GIS function. The flowaccumulation function counts the number of cells located upstream of each cell (the cell itself is not included) and, if multiplied by the cell area, equals the drainage area. Figure 1 shows an example of how the flowdirection and flowaccumulation functions work when applied to a DEM.

    Grid functions

    Figure 1: Grid functions for terrain analysis for hydrologic purposes.

    The stream and watershed network is determined so that there is a single stream segment for each watershed that is modeled. The DEM cells that form the streams are defined as the union of two sets of grid cells. The first set consists of all cells whose flow accumulation is greater than a user-defined threshold value. This set identifies the streams with the largest drainage area, but not necessarily with the largest flow because flow depends on other variables that are not related exclusively to topography. The second set is defined interactively by the user by clicking a certain point on the map, which results in an automatic selection of all downstream cells. This tool was included in HEC-PrePro v. 2.0 because it was observed that under specific circumstances users are interested on particular streams, which might have a small drainage area (low flow accumulation). To include these streams using the threshold criterion, it would be necessary to lower the threshold value for the entire system, thus defining unnecessarily a much more dense stream network.

    Sub-basin outlets are also defined as the union of two sets of grid cells. The first set, based on the stream network, consists of all cells located just upstream of the junctions. Consequently, at a junction, two outlet cells are identified, one for each of the upstream branches. The system outlet is also identified as an outlet. The second set is defined interactively by the user by clicking on any cell on the stream network such as those associated with gages or other water control points. After the sub-basin outlets have been defined, a unique identification code is assigned to each stream segment connecting a headwater cell with a sub-basin outlet, or two sub-basin outlets.

    The watershed GIS function is used to delineate the areas draining to each sub-basin outlet. A one-to-one relation between stream segments and sub-basins is maintained because a unique segment has been identified for each sub-basin outlet.

    Vectorization of the hydrologic elements

    After the stream segments and their corresponding drainage areas have been delineated in the raster domain, a vectorization process is performed using raster-to-vector conversion functions included in ArcView Spatial Analyst 1.1. This process consists of creating a line data set of streams, and a polygon data set of sub-basins. The reason for this vectorization is that the number of hydrologic elements (streams and sub-basins) in the system is usually small compared with the number of grid cells, and further processing and modeling is faster in the vector domain. Following the raster-to-vector conversion, vector-processing steps are included to preserve the one-to-one relationship between stream lines and sub-basin polygons, and to determine the connectivity between polygons. When vectorizing sub-basins, it is important to verify that each sub-basin is represented by a single polygon. A sub-basin represented by more than one polygon is a common problem when the raster representation of the sub-basin includes a dangling set of cells (a group of cells that is connected to the main set of cells only through a corner), because the raster-to-vector converter groups into discrete polygons cells with the same value and a common side (see Figure 2). In such a case, the dangling set of cells will be assigned to a different polygon, thus creating a second polygon for the same sub-basin (same identification code). A program has been included to merge all polygons – sometimes more than two –, that correspond to the same sub-basin, into a single polygon.

    Dangling polygons

    Figure 2: Watershed polygon with dangling polygons. Dangling polygons are merged to the main watershed polygon by running an Avenue script.

    HEC-PrePro v. 2.0 also identifies, for each sub-basin polygon, all the sub-basin polygons located upstream of it, and merges them so that they can be easily retrieved when delineating a watershed from a point, as will be explained below.

    Computation of the hydrologic element parameters

    The sub-basin parameters calculated by HEC-PrePro v. 2.0 are area, length and slope of the longest flow-path, average curve number, and lag-time. Since area and lag-time are the only parameters required for the SCS dimensionless unit hydrograph, HEC-PrePro v. 2.0 is capable of generating all the necessary information for routing flow in the sub-basin.

    The sub-basin area is calculated automatically in the process of vectorizing the sub-basin polygons.

    The longest flow-path is identified as the set of cells of the sub-basin for which the sum of the downstream flow length to the watershed outlet and the upstream flow length to the sub-basin boundary is a maximum (Smith 1995). Before presenting other sub-basin parameters, it is important to discuss the physical meaning of the downstream flow length to the watershed outlet and upstream flow length to the sub-basin boundary. The downstream flow length to the watershed outlet is equal to the distance along a flow path from a grid cell to the outlet of the sub-basin in which it is located. After running the (downstream) flowlength GIS function -- which calculates the flow distance to the border of the analysis window or to a nodata cell (whichever is found first) -- the downstream flow length to the watershed outlet is calculated as the difference between the flow length value of the cell and the flow length value of its corresponding outlet cell. The upstream flow length to the sub-basin boundary is equal to the distance along a flow path from a grid cell to the most upstream location within its watershed, and does not necessarily follow the main channel. After assigning nodata values to all sub-basin outlet cells, the (upstream) flowlength GIS function -- which calculates the flow distance to the most upstream cell of the analysis window or to a nodata cell (whichever is found first) – is used to calculate the upstream flow length to the sub-basin boundary. nodata values are assigned to the sub-basin outlets to keep the flowlength function from searching for longer flow-paths in the upstream sub-basins.

    The length of the longest flow-path is equal to the maximum value of the sum of the downstream flow length to the watershed outlet and the upstream flow length to the watershed boundary.

    The slope of the longest flow-path is determined as the elevation drop between two arbitrarily defined points of the flow path, divided by their distance along the channel. The points can be located at any user-defined distance from the sub-basin outlet, expressed as a percentage of the length of the longest flow-path. For instance, percentages of 10% and 85% refer to that 75% portion of the channel located 10% of the channel length upstream of the sub-basin outlet.

    The average SCS curve number of the sub-basin is calculated as the average of the curve number values within the sub-basin polygon. A curve number grid is calculated using land use data described by Anderson's land use code, percentage of each hydrologic soil group (A, B, C and D) according to STATSGO soils data, and a look-up table that relates land use and soil group with curve numbers (Smith 1995).

    The sub-basin lag-time is calculated with the SCS formula and is given by (Chow, Maidment and Mays 1988)

    where tp (minutes) is the sub-basin lag-time -- measured from the centroid of the hyetograph to the peak time of the hydrograph --, Lw (feet) is the length of the longest flow-path, S (%) is the slope of the longest flow-path, and CN is the average curve number in the sub-basin. However, because in HMS the analysis time-step has to satisfy the condition of being smaller than 0.29 times the lag-time of the basin (HEC 1990), the lag-time is taken as the value given above or 3.5 times the analysis time-step, whichever is greater. Therefore, the lag-time is redefined as

    where Dt is the analysis time-step. Although this modification artificially delays the flow in the sub-basin, it only affects the very small watersheds (with lag-times smaller than 3.5 times the analysis time step) in which the volume of runoff produced is small compared with the size of the entire system. Moreover, mass conservation is not affected by altering the sub-basin lag-time.

    The stream parameters determined by HEC-PrePro v. 2.0 are the length, the routing method – either Muskingum or pure lag --, the Muskingum K and the number of sub-reaches into which the stream is subdivided in case Muskingum is used for routing, and the flow time in case pure lag is used for routing. Other stream parameters like the flow velocity and the Muskingum X can not be computed by HEC-PrePro v. 2.0 and must be calculated externally and supplied as input.

    The reach length is determined automatically in the process of stream vectorization, and flow time is calculated as , where L is the reach length and v is the flow velocity.

    The Muskingum method is used for stream routing in all reaches long enough not to present numerical instability problems. In short reaches, in which the flow time is shorter than the time-step, the pure lag method is used as will be explained later.

    To avoid numerical instability when using the Muskingum method, long reaches are subdivided into shorter equal-length sub-reaches, so that the flow-time in each of them satisfies the condition (Fread 1993), where k is the flow time in the sub-reach. Since the flow time in the sub-reaches is equal to , where K is the flow time in the reach, and n (an integer value greater than zero) is the number of sub-reaches, then it follows

    Moreover, because n should be at least equal to 1, should be greater than Dt to satisfy the Muskingum method constraints, otherwise the pure lag method must be used. Additionally, the minimum number of sub-reaches into which the reach can be subdivided is given by:

    where int takes the integer part of the argument (int does not round the number).

    K for the Muskingum method and the lag time for the pure lag method are both equal to the flow time which, as indicated above, is .

3.2. Isolation and process of a hydrologic sub-system

4. Application to the Llano River in Central Texas

    HEC-PrePro v. 2.0 has been run for the Llano River, tributary of the Colorado River in Central Texas. A study area of 200 Km (North – South) by 180 Km (West – East), which also includes the Concho and San Saba Rivers, was identified (see Figure 3). A total of 145,000 500-meter DEM cells were used in the terrain analysis.

    Study area

    Figure 3: The study area is located in Central Texas and includes the Concho, San Saba and Llano Rivers, tributaries of the Colorado River.

    The pre-process of the study area consists of running in sequence all the options of the pull-down menu shown in Figure 4, from Fill Sinks to Calculate Attributes. The process of a sub-system consists of running the Run HECPREPRO option after isolating the sub-system.

    Pull down menu

    Figure 4: HEC-PrePro pull-down menu. Options should be run in sequence.

    After running the terrain analysis – Fill Sinks, Flow Direction, and Flow Accumulation –, all cells draining more than 750 Km2 (3,000 grid cells) were identified as stream cells, and six additional streams and one sub-basin outlet were added interactively. Figure 5 shows the streams defined by the threshold criterion (blue), those added interactively (red), and the sub-basin outlet (green).

    Added streams and outlets

    Figure 5: Streams and sub-basin outlets can be added interactively. Blue streams have been defined by the threshold criterion, red streams and the green sub-basin outlet have been added by clicking on the map.

    Once streams and outlets are identified, sub-basins are delineated in the raster domain. Figure 6 shows the resulting stream - watershed delineation after vectorization, in which the Concho River, San Saba River, and Llano River can be identified at the Northern, Central and Southern part of the study area respectively.

    Stream - watershed delineation

    Figure 6: Stream - watershed delineation after vectorization. Note the one-to-one relation between streams and watersheds.

    Finally, hydrologic parameters for each of the elements are calculated and attached to their corresponding attribute table, as can be seen in Figures 7 and 8.

    River attribute table

    Figure 7: Attribute table of the stream vector data set. Stream parameters have been attached to the table.

    Watershed attribute table

    Figure 8: Attribute table of the watershed vector data set. Watershed parameters have been attached to the table.

    After isolating the Llano River basin by clicking near its junction with the Colorado River, a basin file -- readable by HMS -- is created using the information of the hydrologic system obtained in GIS. This basin file, in ASCII format (Figure 9), includes the hydrologic parameters of the elements, and their interconnectivity.

    HMS basin file

    Figure 9: HMS basin file in ASCII format. Hydrologic parameters calculated in GIS and stored in the attribute tables are transferred to the basin file.

    Figure 10 shows the schematic of the Llano River basin in the HMS - Schematic window, and Figure 11 shows hydrologic parameters in the HMS – Basin Model Editor window.

    Schematic in HMS

    Figure 10: HMS display of the schematic of the Llano River basin.

    HMS element editor window

    Figure 11: HMS element editor window. Stream parameters calculated in GIS have been transferred to HMS.

5. Conclusions

Acknowledgements

References

Chow, V.T, D.R. Maidment and L.W. Mays (1988), Applied Hydrology, McGraw-Hill Inc., New York, 1988.

DeVries, J.J., and T.V.Hromadka (1993), Computer Models for Surface Water in Handbook of Hydrology, ed. by D.R. Maidment, McGraw-Hill Inc., New York, 21.1-21.39.

D.L.Fread (1993), Flow Routing in Handbook of Hydrology, ed. by D.R. Maidment, McGraw-Hill Inc., New York, 10.1-10.36.

HEC (1990), "HEC-1 Flood Hydrograph Package," User's Manual, p.24, Hydrologic Engineering Center, U.S. Army Corps of Engineers, Davis, CA.

Hellweger, F., and D.R.Maidment (1997), Definition and Connection of Hydrologic Elements Using Geographic Data, accepted for publication in the ASCE – Journal of Hydrologic Engineering.

Jensen, S.K., and J.O. Domingue (1988), Extracting Topographic Structure from Digital Elevation Data for Geographic Information System Analysis, Photogrammetric Engineering and Remote Sensing 54 (11).

Jensen, S.K. (1991), Applications of Hydrologic Information Automatically Extracted from Digital Elevation Models, Hydrologic Processes 5(1).

Smith, P. (1995), Hydrologic Data Development System, Master Thesis, Department of Civil Engineering, University of Texas at Austin.

USGS a, 7.5-Minute Digital Elevation Model Data, http://edcwww.cr.usgs.gov/glis/hyper/guide/7_min_dem as of March 1998.

USGS b, 1-Degree Digital Elevation Models, http://edcwww.cr.usgs.gov/glis/hyper/guide/1_dgr_dem as of March 1998.

USGS c, Metadata for GCIP Reference Data Set (GREDS), http://nsdi.usgs.gov/nsdi/wais/water/gcip.HTML as of March 1998.

USGS d, Global 30 arc-second Elevation Data Set, http://edcwww.cr.usgs.gov/landdaac/gtopo30/gtopo30.html as of March 1998.

Author Information

Francisco Olivera, PhD
Research Associate
University of Texas at Austin
Center for Research in Water Resources
J.J.Pickle Research Campus # 119
Austin, TX 78712
Telephone: (512) 471-0570
FAX: (512) 471-0072
folivera@mail.utexas.edu

Seann Reed
Graduate Research Assistant
University of Texas at Austin
Center for Research in Water Resources
J.J.Pickle Research Campus # 119
Austin, TX 78712
Telephone: (512) 471-0073
FAX: (512) 471-0072
seann@mail.utexas.edu

David Maidment, PhD
Professor of Civil Engineering
University of Texas at Austin
Center for Research in Water Resources
J.J.Pickle Research Campus # 119
Austin, TX 78712
Telephone: (512) 471-0065
FAX: (512) 471-0072
maidment@mail.utexas.edu