University of
Texas at Austin
Center for Research in Water Resources
Project: System of GIS-Based Hydrologic and Hydraulic Applications for
Highway Engineering
Sponsored by: Texas Department of Transportation (TxDOT)
Principal Investigator: David R. Maidment, Ph.D.
Project Manager: Francisco Olivera, Ph.D.
Graduate Research Assistant: Juling Bao
1. Flow Hydrographs and Peak Discharges
2. Computer-Based Hydrologic and Hydraulic Modeling
Broadly speaking, this review consists two parts: (1) review of published material dealing with estimating the hydrologic parameters used for designing highway drainage structures, and (2) review of Internet sites that supply information about geospatial data and hydrologic records. With this approach in mind, this review has been subdivided into the following sections: (1) Flow Hydrographs and Peak Discharges, (2) Computer-Based Hydrologic and Hydraulic Modeling, and (3) Geospatial Data of the State of Texas. The first section deals with flow estimation, and special attention has been given to flood peak discharges because of the amount of literature that addresses this topic, and because peak discharges constitute the key design parameter for highway drainage structures. The second section addresses the use of computers for hydrologic/hydraulic analysis, the use of geographic information systems (GIS) for accounting for the spatial variability of the terrain, and the use of GIS-based hydrologic models for designing highway drainage structures. The last section deals with the availability of geospatial data of the State of Texas for hydrologic modeling.
The unit hydrograph, a method for estimating storm runoff, was first proposed by L.K. Sherman in 1932 (Chow et al. 1988, p.213), and since then has been used as a key concept. The unit hydrograph is defined as the watershed response to a unit depth of excess rainfall, uniformly distributed over the entire watershed and applied at a constant rate for a given period of time. In 1938, after studying watersheds in the Appalachian mountains of the United States, Snyder proposed relations between some of the characteristics of the unit hydrograph, i.e. peak flow, lag time, base time, and width (in units of time) at 50% and 75% of the peak flow (Chow et al. 1988, p.224). Snyder's method was enhanced with the regionalization of the watershed parameters developed in 1977 by Espey, Altman and Graves (Chow et al. 1988, p.227). A significant contribution to the unit hydrograph theory was given by Clark (1945), who proposed a unit hydrograph which is the result of a combination of a pure translation routing process (plug flow) followed by a pure storage routing process (completely stirred tank reactor). Although Clark does not develop a spatially distributed analysis, the translation part of the routing is based on the time-area diagram of the watershed. The storage part consists of routing the response of the translation through a single linear reservoir located at the watershed outlet. The detention time of the reservoir is selected in order to reproduce the falling limb of observed hydrographs. Note that the actual travel time of a water particle, according to this approach, is the travel time given by the time-area diagram plus the detention time of the reservoir, which is somewhat inconsistent. Some years later, Nash (1957) proposed a unit hydrograph equation which is a gamma distribution, i.e. the response of a cascade of identical linear reservoirs to a unit impulse. It is important to notice that the method proposed by Nash did not model the watershed itself, and was just a fitting technique based on the first and second moments of the calculated and observed hydrographs. In 1972, the Soil Conservation Service (SCS) of the US Department of Agriculture (USDA) proposed a unit hydrograph model based on a single parameter: the lag time between the center of mass of the excess precipitation hyetograph and the peak of the unit hydrograph. The shape of the hydrograph is given by an average pre-computed dimensionless unit hydrograph curve or, as a simplification, by triangular dimensionless unit hydrograph (Chow et al. 1988, p.228).
Yet, studying the storm rainfall-runoff relation involves much more than studying the unit hydrograph. Consequently, in trying to relax the unit hydrograph assumptions of uniform and constant rainfall, and to account for spatial variability of the catchment, considerable research has been done in recent years, and many articles dealing with these topics can be found in the literature.
Pilgrim (1976) carried out an experimental study consisting in tracing flood runoff from specific points of a 0.39 Km2 watershed, near Sydney, Australia, and measuring the travel time of the labeled particles to the outlet. A conclusion of his study is that "at medium to high flows the travel times and average velocities become almost constant, indicating that linearity is approximated at this range of flows". Pilgrim also states that time-variations of the tracer activity time curves "make an additional contribution to the non-linearity of the runoff process".
An important attempt to linking the geomorphological characteristics with the hydrologic response of a watershed is given by Rodriguez-Iturbe and Valdes (1979). In their paper, Horton's empirical laws, i.e. law of stream numbers, lengths and areas, are used to describe the geomorphology of the system. The instantaneous unit hydrograph is defined as the probability density function of the time a rainfall drop chosen at random takes to reach the outlet. This time is given by the sum of the times spent in each state (order of the stream in which the drop is located) on its way to the outlet. The time spent in each state is taken as a random variable with an exponential probability density function whose parameter depends on Horton's length ratio, mean velocity of the stream flow (dynamic parameter), and a scale factor.
Mesa and Mifflin (1986), Naden (1992) and Troch et al. (1994) present similar methodologies to account for spatial variability when determining the watershed response. The catchment response is calculated as the convolution of a network response and a hillslope response.
To calculate the network response, Mesa and Mifflin (1986) use the solution of the advection-dispersion equation, weighted according to the normalized width function of the network. In their paper, the normalized width function is defined as the number of channels at a given distance to the outlet, divided by the total length of all channels in the network. For the hillslope response, Mesa and Mifflin suggest a double travel time function, related to fast and slow flow, in the form of two isosceles triangles. The two functions are weighted, according to the probability that a water drop would take either path to the channel system, and added to give the final hillslope response. From the physical viewpoint, fast and slow hillslope responses are related to surface and subsurface flow respectively. Their model was tested in a 1.24 Km2 sub-basin of the Goodwin Creek watershed in Mississippi. For the stream network, an average velocity of 1 m/s and a dispersion coefficient of 9.06 m2/s were found. For the hillslope response, the average velocities of the fast and slow components were 0.25 m/s and 0.0046 m/s respectively, and the fraction of the slow flow was taken equal to 90% of the total hillslope response.
For the network response, Naden (1992) suggests also the solution of the advection-dispersion equation, but weighted by a standardized width function of the network. In her paper, the standardized width function is defined as the number of channels at a given distance to the outlet, divided by the total number of channels in the network. She also recommends an additional weighting of the width function by the excess rainfall spatial distribution. Naden does not give a specific methodology to determine the hillslope response, and the one used in her paper "was selected by eye" as a single peak, reflecting the quick response, followed by an exponentially decaying curve for the slow component. For the case of the River Thames at Cookham in United Kingdom, a stream flow velocity of 0.6 m/s and dispersion parameter of 1 m2/s were found. Additionally, because of the slow component of the hillslope response, which yields about 80% of the flow volume, the rainfall spatial variability is smoothed out resulting in almost identical watershed responses for different rainfall spatial patterns. The ratio of the average velocities of the fast and slow components was found to be around 20.
Troch et al. (1994) propose the same stream network response as Mesa and Mifflin (1986). However, for the hillslope response they suggest a function given by the solution to the advection-dispersion equation, applied this time to the overland flow, and weighted according to a normalized hillslope function. The normalized hillslope function is interpreted as the probability density function of runoff generated at a given overland flow distance from the channel network. Contrary to Mesa and Mifflin's and Naden's hillslope response functions, Troch et al.'s does not account for the slow component.
An interesting approach to model the fast and slow responses of a catchment is presented by Littlewood and Jakeman (1992, 1994). In their model, the watershed is idealized as two linear storage systems in parallel, representing the surface and the subsurface water systems. The surface system is faster and affects mainly the raising limb of the resulting hydrograph, while the subsurface system is slow and determines the tail of the response.
Although the linear unit hydrograph model has been used for more than 60 years, it is well know that flow, specially in the streams, exhibits a non-linear behavior. Flow velocity, as modeled by Manning's or Chezy's equations for example, is a function of the water depth which implies that the duration of the watershed response depends on the volume of water flowing. Therefore, in principle, superposition, a well-known property of linear systems, does not apply to flow systems.
Many distributed flow routing methods can be found in the literature (Chow et al. 1988, Lettenmier and Wood 1993). Based on the Saint Venant continuity and momentum equations, the dynamic wave model, diffusion wave model and kinematic wave model can be derived. The simplest among them, the kinematic wave model, neglects pressure and inertial forces in the flow and leaves friction equal to gravity forces. The diffusion wave model considers, additionally, pressure force terms; while the dynamic wave model includes also inertial terms. These models can be defined as linear or non-linear, depending on the way the original equations were set.
In non-linear systems, the terrain shall be analyzed continuosly, because its hydrologic behavior changes with time, and superposition can not be used. Not using superposition, though, implies determining the continuously changing response of the watershed, which might be complicated for uniform systems and eventually inapplicable for spatially variable systems.
Many computer programs for hydrologic and hydrologic modeling are available to the engineering community. Some of these programs have been developed by the government and are of public domain. DeVries and Hromadka (1993) have prepared a comprehensive summary of available software, in which programs are grouped in the following categories: (1) single-even rainfall-runoff and routing models, (2) continuous-stream-flow simulation models, (3) flood-hydraulics models, and (4) water-quality models. Because of the widespread use of HEC-1, HEC-2 - and its Windows version HEC-RAS - and TR-20, overviews of the programs are included in this review; however, the reader is referred to the software manuals for detailed information about them.
HEC-1 is a computer model for rainfall-runoff analysis developed by the Hydrologic Engineering Center of the U.S. Army Corps of Engineers. The program develops discharge hydrographs for either historical or hypothetical events for one or more locations in a basin. To account - to a certain extent - for spatial variability of the system, the basin can be subdivided into subbasins with specific hydrologic parameters.
The program options include: calibration of unit hydrograph and loss-rate parameters, calibration of routing parameters, generation of hypothetical storm data, simulation of snowpack processes and snowmelt runoff, dam safety applications, multiplan/multiflood analysis, flood damage analysis, and optimization of flood-control system components. Uncontrolled reservoirs and diversions can also be accommodated.
Precipitation excess is transformed into direct runoff using either unit hydrograph or kinematic wave techniques. Several unit hydrograph options are available: it may be supplied directly by the user, or it may be expressed in terms of Clark, Snyder, or Soil Conservation Service unit hydrograph parameters. The kinematic wave option permits depiction of subbasin runoff with elements representing one or two overland-flow planes, one or two collector channels, and a main channel.
HEC-2 was developed by the Hydrologic Engineering Center of the U.S. Army Corps of Engineers to compute steady-state water surface elevation profiles in natural and constructed channels. Its primary use is for natural channels with complex geometry such as rivers and streams. The program analyzes flow through bridges, culverts, weirs, and other types of structures.
The encroachment computation option has been widely used in the analysis of floodplain encroachments for the U.S. Federal Emergency Management (FEMA) flood insurance program. There are several types of encroachment calculation procedures, including the specification of encroachments with fixed dimensions and the designation of target values for water surface increases associated with floodplain encroachments. The program requires that three flow path distances be used between cross sections: a channel length and left and right overbank lengths.
HEC-2 uses the standard direct step method for water surface profile calculations, assuming that flow is one-dimensional, gradually varied, and steady. The program computes water surfaces as either a subcritical flow profile or a supercritical profile. Mixed subcritical and supercritical profiles are not computed simultaneously. If the computations indicate that the profile should cross critical depth, the water surface elevation used for continuing the computations to the next cross section is the critical water surface elevation.
HEC-2 computes up to 14 individual water surface elevation profiles in a given run. Usually a different discharge is used for each profile, although when the encroachment or channel improvement options are used, the section dimensions are changed rather than the discharge. The discharge can be changed at each cross section to reflect tributaries, lateral inflows, or diversions.
In the last years, HEC has developed HEC-RAS, for River Analysis System, which has the same features as HEC-2 but with a Windows interface. Besides the user interface, no major differences between the programs have been observed.
The U.S. Soil Conservation Service (SCS) TR-20 computer model is a single-event rainfall-runoff model that is normally used with a design storm as rainfall input. The program computes runoff hydrographs, routes flows through channel reaches and reservoirs, and combines hydrographs at confluences of the watershed stream system. Runoff hydrographs are computed by using the SCS curve number method, based on land-use information and soil maps indicating soil type, and the SCS dimensionless unit hydrograph defined by a single parameter, the watershed lag. TR-20 utilizes the SCS methods given in the Hydrology section of the National Engineering Handbook.
Watersheds are usually divided into subbasins with similar hydrologic characteristics and which are based on the location of control points through the watershed. Control points are locations of tributary confluences, a structure, a reservoir, a diversion point, a damage center, or a flood-gauge location.
Historical or synthetic storm data are used to compute surface runoff from each subbasin. Excess rainfall is applied to the unit hydrograph to generate the subbasin runoff hydrograph. Base flow can be treated as a constant flow or as a triangular hydrograph. A linear routing procedure is used to route flow through stream channels. The modified Puls method (storage-indication routing) is used for reservoir routing. As many as 200 channel reaches and 99 reservoirs or water-retarding structures can be used.
TR-20 has been widely used by SCS engineers in the United States for urban and rural watershed planning, for flood insurance, and flood hazard studies, and for design of reservoirs and channel projects. The SCS methodology is accepted by many local agencies also. TR-55 is a simplified version of TR-20 that does rainfall-runoff modeling for a single watershed.
In a relatively short time, Geographic Information Systems (GIS) have gained fairly widespread use in a variety of engineering applications. Originally envisioned (and used) as a geographic mapper with integrated spatial database, GIS are increasingly being used in modeling applications, where geographic data can be readily accessed, processed and displayed. GIS have been implemented mostly by large entities such as Federal, State, and local government agencies, being the predominant use mapping and management of spatial data. However, there is increasing interest in the potential application of GIS in engineering design and analysis, especially in hydrology and hydraulics.
Because of the infancy of the use of geographic information systems for hydrologic modeling, the practicing engineering community has had only limited exposure to such field. This was verified by a survey, developed as part of Smith's master project (1995), sent out to the fifty state highway agencies to assess the current use (state of the practice) and expected use of GIS for hydraulics-related highway work. From the thirty two responses that were received, it became evident that those state highway agencies who have implemented GIS (ten states) are using it just for mapping and data management. Most of them recognize the potential of GIS for engineering analysis, but only the state of Maryland has implemented a system that supports hydrologic analysis, i.e., GISHYDRO (Ragan 1991). To some extent, the distinction between GIS and Computer Aided Design (CAD) seems to be blurred. GEOPAK, for example, listed by one responder as a GIS, is a roadway design CAD package which has digital elevation model (DEM) capability.
In the hydrologic environment, GIS is a tool that allows one to jump from lumped pre-GIS models to spatially distributed models. The border between lumped and distributed models is not sharp, and there are pre-GIS attempts to deal with spatially distributed terrain attributes. For example, the US Army Corps of Engineers flood model HEC-1, well known as a lumped model, allows the user to subdivide the watershed in smaller sub-basins for analysis purposes, and route their corresponding responses to the watershed outlet. In this case, the concept of purely lumped model does not apply, although the model can not be considered a fully spatially distributed model either. It is therefore advisable to keep in mind the extent to which a given model is lumped or distributed.
Several pioneers are worthy of note for their foresight and work in the development of hydrology-related application of GIS for engineering applications. DeVantier and Feldman (1993) presented a general review of the connection between GIS and hydrologic modeling, which "summarizes past efforts and current trends in using digital terrain models and GIS to perform hydrologic analysis". The link between GIS and hydrologic modeling becomes more natural as the concern about spatially distributed terrain parameters and the use of computers for hydrologic analysis turns more widespread. Digital terrain models (DTM) are the means used by GIS to describe the spatially distributed attributes of the terrain, which are classified as topologic and topographic data; although, strictly speaking, topographic is part of topologic data. Digital elevation models (DEM), in particular, refer to the topographic data, while all other attributes, not related to elevation, constitute the topologic data. It can be expected that, because of the large amount of information required to describe the terrain, GIS is a memory and computationally intensive system. However, storing and handling the data is not necessarily the critical point when working with GIS, because the acquisition and compilation of the information can be an even more difficult task.
Terrain data can be handled in different ways, depending on the type of model to be used. The grid approach consists in subdividing the terrain into identical square cells arranged in rows and columns; triangular irregular networks (TIN) consist in selecting a set of representative irregularly distributed points and connecting them by straight lines producing triangles; and digital line graphs (DLG) consist in digitally representing the elevation contour lines as a set of point-to-point paths. Accordingly, it is expected that grid data, because of its geometric structure, leads to finite difference methods, while TIN data to finite element methods of runoff computation. The extra effort required for working with TIN's and finite element methods is compensated by the fact that TIN's are less memory demanding, because their resolution is not fixed and can be suited to the local terrain characteristics. On the other hand, although modeling with grid and finite difference methods is less complex, because of its fixed geometric structure, it is more memory demanding. DLG's appear mainly as a natural way to store information, and as a data source for analysis with grid or TIN.
Much research has focused on stream-watershed delineation and, in general, on watershed analysis based on topographic data, say DEM's or DLG's. Hutchinson (1989) presents an interpolation algorithm to determine the DEM from elevation data points and stream lines. This algorithm produces DEM's that are consistent with the stream lines and has proven to produce more accurate DEM's than the ones obtained with previous methodologies. Jensen and Domingue (1988) and Jensen (1991) outlined a grid scheme to delineate watershed boundaries and stream networks to defined outfalls (pour points). The scheme uses digital elevation data to determine the hypothetical direction of flow from each cell in a grid to one of its eight neighboring 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 establish stream network cells. Tarboton et al. (1991) computed stream slopes and stream lengths using a similar grid system. In addition, the authors proposed criteria for proper selection of the threshold based on statistical properties of the terrain. Jones et al. (1990) employed a triangulation scheme on digital elevation data to determine watershed boundaries and flow paths. Procedures for delineating streams and watersheds from DEM's, as well as for correcting DEM's depressions produced by data noise, can be found in Maidment (1994), Meijerink et al. (1994), ESRI (1992), Garbrecht and Martz (1995 a, 1995 b) and Martz and Garbrecht (1992).
Maidment et al. (1996 b) present the watershed delineation of the Niger river basin based on a 1 Km DEM. In this delineation, a stream is identified on the DEM wherever the upstream drainage area exceeds 10,000 Km2, and subwatershed boundaries are delineated from outlets at each stream junction, which produces a drainage network with a single stream for each subwatershed. To avoid long reaches between junctions, outlets were also placed on long streams each 250 Km. A total of 167 streams with their corresponding drainage areas were determined in this way. Before delineating the watersheds, the DEM had to be corrected to account for the Lake Chad inland catchment, at the North-East of the Niger basin. Since the standard delineation process consists in filling up terrain depressions, a pour point at the lowest point of the Lake Chad basin was defined to avoid filling up the whole catchment and making it overflow towards the Niger basin.
Rinaldo et al. (1992) analyzed the similarities between stream networks derived from DEM's and optimal channel networks (OCN) obtained by minimizing the energy spent in the system. Likewise, an automated procedure, fully based on topographic data, for subdividing catchments into smaller elements and for calculating hydrologically relevant attributes of the elements is described by Moore et al. (1988) and Moore and Grayson (1991). This catchment partitioning is done in order to apply lumped models, that represent particular hydrologic processes, at an element level. The integration of the element responses gives the spatially-distributed response of the entire catchment.
Grid-based GIS appears to be a very suitable tool for hydrologic modeling, mainly because "raster systems have been used for digital image processing for decades and a mature understanding and technology has been created for that task" (Maidment 1992 a). The ESRI Arc/Info-GRID system and the United States Army Corps of Engineers GRASS system, use a grid data structure. Grid systems have proven to be 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 variable. This characteristic is what makes topographically driven flow easily modeled in a grid environment and, consequently, grid systems include hydrologic functions as part of their capabilities. At present, hydrologic functions, available in GRID and GRASS, allow one to determine flow direction and drainage area at any location, stream networks, watershed delineation, etc. (Maidment 1992 a).
Recently, there have been attempts to take advantage of GIS capabilities for runoff and non-point source pollution modeling. Vieux (1991) presents a review of water quantity and quality modeling with GIS and, as an application example, employs the kinematic wave method to an overland flow problem. GIS is used to process the spatially variable terrain and the finite elements method (FEM) to solve the mathematics. Maidment (1992 a, 1992 b, 1993) presents a grid-based methodology for determining a spatially distributed unit hydrograph that assumes a time-invariant flow velocity field. According to him, the velocity time-invariance is a requirement for the existence of a unit hydrograph with a constant time base and relative shape. This concept is also explained in this research, in the light of the conditions for linearity of a routing system. In Maidment's articles, from a constant velocity grid, a flow time grid is obtained and subsequently the isochrone curves and the time-area diagram are determined. The unit hydrograph is obtained as the incremental areas of the time-area diagram, assuming a pure translation flow process. A more elaborated flow process, accounting for both translation and storage effects, is presented by Maidment et al. (1996 a). In their paper, the watershed response is calculated as the sum of the responses of each individual grid-cell, which is determined as a combined process of channel flow (translation process) followed by a linear reservoir routing (spreading process). Although an approximate method, the model shows a good fit for the unit hydrograph of the Severn watershed at Plynlimon in Wales. Olivera et al. (1995) and Olivera and Maidment (1996) present a grid-based, unsteady-flow, linear approach that uses the diffusion wave method to model storm runoff and constituent transport. According to this paper, the routing from a certain location to the outlet is calculated by convolving the responses of the grid-cells of the drainage-path.
Sensitivity of model results to the spatial resolution of the data has been addressed by Vieux (1993), who discusses how the grid-cell size affects the terrain slope and flow-path length, and accordingly the surface runoff. Vieux and Needham (1993) conclude that increasing the cell size shortens the streams length and increases the sediment yield.
A water balance of the State of Texas, using GIS, was prepared by Reed and Maidment (1996), in which a 5 Km precipitation grid, a 500 m digital elevation model (DEM), gauged streamflow data, and other spatial data sets were used to generate spatially distributed maps of mean annual runoff and evaporation. 166 gauged watersheds were delineated from a 500 m DEM and hydrologic attributes were compiled for each of them. To estimate the runoff in ungauged locations, plots of watershed average annual rainfall (mm) versus annual runoff per unit watershed area (mm) were analyzed. By eliminating watersheds with a large amount of reservoir evaporation, urbanization, recharge, or springflow, a clear trend emerged in this rainfall-runoff data, and an expected runoff function was derived. Because runoff values were normalized by watershed area, this expected runoff function is scale-independent, and represents watersheds with drainage areas ranging from 270 to 50,000 Km2. An expected runoff grid was created by applying the expected runoff function to the precipitation grid. Finally, a grid of actual runoff was created on a 500 m grid by combining gauged streamflow data with expected runoff information. By applying a flow accumulation function to the runoff maps, the expected and actual flows were calculated at each 500 m DEM cell. Flow maps created using these results show statewide spatial trends such as the increased density of stream networks in East Texas, and also capture localized phenomena such as large springflows and agricultural diversions. A map of the differences between actual and expected runoff shows where human activities have altered natural runoff.
In the area of floodplain management, the US Corps of Engineers has developed an integration between HEC-2, a widely used floodplain determination package, and GRASS, a software developed to work with raster data (Walker et al. 1993). The integration package accesses HEC-2 output in tabular form, and converts it into GRASS format. For floodplain determination, Talbot et al. (1993) developed a GIS application that takes water elevations as input. Their approach is intended to be non-specific, accepting stage values from any model that can determine water elevations along a stream channel. HEC-1 and HEC-2 are mentioned as valid sources of stage values. The application involves the intersection two TIN's, one representing the terrain, and the other the channel's water elevations, so that the banks of the floodplain can be established. The authors indicate that the resulting floodplain is locally reasonable and indicative of the overall floodplain. Beavers (19**) has developed ARC/HEC2, a set of AML's (Arc/Info Macro Language) and C programs, which work to extract terrain information from contour coverages, insert user-supplied information (such as roughness coefficients, or location of left and right overbanks), and format the information into HEC-2 readable data. Following HEC-2 execution, ARC/HEC2 is capable of retrieving the HEC-2 output (in the form of water elevations at each cross section) and creating an Arc/Info coverage of the floodplain. This process allows the resulting floodplain to be stored in a coverage format which is readily accessed by users who wish to use the floodplain information in conjunction with other Arc/Info coverages. ARC/HEC2 requires that a terrain surface be generated so that accurate cross section profiles are provided to HEC-2. These terrain surfaces, in the format of TIN's or grids, are created within Arc/Info based on contour lines, survey data, or other means of establishing terrain relief. The accuracy of the surface representation is crucial for accurate floodplain calculations.
Watershed Modeling System (WMS) is a hydrologic software package, developed at the Engineering Computer Graphics Laboratory (ECGL) at Brigham Young University, divided logically into six integrated and task-oriented modules: TIN Module, DEM Module, Tree Module, Grid Module, Scatter Point Module, and Map Module.
Tin Module. A TIN is a set of elevation points which have been connected to form a network of triangles. The triangles fit together in a manner which simulates the face of the land. This allows simple calculation of basin areas, slopes, channels, and many other geometric parameters, such as contours and oblique terrain views. These geometric parameters can then be combined with hydrologic analyses.
The TIN module is used for terrain modeling and automated basin delineation. Points used to create TINs can be obtained by digitizing a contour map, or generated automatically from feature arcs and polygons, using DEMs or existing TINs as background elevation maps. TINs are used for basin delineation and drainage analysis.
If stream feature lines are used to create a TIN, a stream network of TIN points is automatically created as part of the TIN. An initial outlet point is established at the end of each stream network. Watershed boundaries for each outlet point can then be determined by grouping together all triangles whose flow paths pass through the outlet. Additional oulet points can be added at any point along the stream in order to subdivide the watershed into smaller sub-basins. Once basin boundaries have been delineated, geometric attributes such as area, slope, and runoff distances can be calculated for each basin.
Because no two terrain models are alike, WMS has a variety of tools for manual and automatic TIN editing including: breaklines, swapping edges, coordinate editing, smoothing, pit removal, and point/triangle insertion/deletion. TINs can be contoured, displayed in oblique view with hidden surfaces removed, and other display options can be set to visualize and understand the terrain surface better.
DEM Module. Digital Elevation Models (DEM) from the United States Geological Survey (USGS) can be imported and used in WMS to define elevations in a hydrologic model. The USGS has made these models available on the Internet. Besides standard USGS DEMs, other regular gridded elevation models such as those produced by Arc/Info and other common GIS software can be imported into WMS and used as background elevation information. In order to minimize storage requirements, most DEM elevations are stored as integer values by WMS. In other words, all elevations are rounded to the nearest whole number. While there is nothing fundamentally wrong with rounding elevations in this manner (typically the accuracy to which elevations are known is greater than unity), it can lead to complications in runoff modeling because "artificial" flat regions can occur. Such flat regions can be remedied in WMS using DEM smoothing operations.
Map Module. The Map module is used to define stream channels, ridges, boundaries, and any other important terrain feature present in the model. The points, nodes, arcs, and polygons have been structured after the Arc/Info data model. TINs can be constructed from these feature objects using an existing TIN or a DEM as a background elevation map. Triangle edges are enforced along all arcs. Streams for 2-D analysis using CASC2D are also created and parameters assigned using tools in the Map module.
Within the Map module there are several other tools which can be helpful in either setting up models or presentation of results to a client. Tools for reading and writing of DXF files, mapping TIFF images, and text annotation are part of this module.
Tree Module. Traditionally HEC-1 and TR-20 models are developed around a topologic representation, or tree diagram of a watershed. Nodes or icons for each component, such as outlet points (confluences), basins, diversions, and reservoirs are linked together according to the underlying stream network of the watershed. Using WMS, tree diagrams can be established in one of two ways: automatic creation from TIN geometry, or manually defining outlets, basins, diversions, etc., and linking them together. Preferably a TIN is used since it can help supply important geometric information which would otherwise have to be determined manually from maps.
Once the tree diagram for a watershed is established, all necessary data to run a HEC-1 simulation can be defined using "point and click" techniques and a series of user-friendly dialogs. Any of the different precipitation, loss, unit hydrograph, routing, etc., methods available in HEC-1 can be defined.
Potential errors such as undefined basin/outlet data can be detected and corrected prior to running the model using the model checker. WMS automatically creates a properly formatted HEC-1/TR-20 input file and then launches the HEC-1/TR-20 program. After successfully running, hydrograph results can be viewed simultaneously inside of WMS to aid in model calibration.
Grid Module. The Grid module is used for surface visualization and for the development of a CASC2D rainfall/runoff analytical model. For example, the user can discretize a watershed into a number of grid cells and then define important rainfall, infiltration, and channel properties at grid cells in preparation for running CASC2D. Any parameter such as hydraulic conductivity or rainfall intensity may be interpolated from a set of scattered data points to the grid. Results of the 2D analysis can then be contoured on the grid or displayed with hidden surface removal and color fringes to display the variation in the computed results.
Scatter Point Module. The Scatter Point module is used to interpolate from groups of scattered data points to grids. The Scatter Point module can be used to interpolate from a set of scattered (x,y) points representing something like rain gages to a finite difference grid or to basin centroids for establishing rainfall curves for HEC-1. A variety of interpolation schemes are supported. NEXRAD data can be used in the scatter point module to contour rainfall time-steps, from which rainfall animation is possible.
WMS also provides an interface to five hydrologic models: HEC-1, TR-20, CASC2D, Rational Method, and National Flood Frequency (NFF).
HEC-1 and TR-20. WMS includes a comprehensive interface to the HEC-1 and TR-20 hydrologic models used by many hydrologic engineers to model the rainfall-runoff process. The interfaces have been created in such a way that models can be built from TINs used to delineate basin boundaries and compute geometric data, or by manually constructing a series of outlets and basins to form a topologic representation of the watershed.
CASC2D. Originally developed at Colorado State University under the direction of Dr. Julien, its development and refinement has been done by Dr. Barahm Saghafian (CERL). CASC2D couples both overland and channel routing using a finite difference approach. It accounts for antecedent moisture conditions, evapotranspiration and infiltration losses, and spatial and temporal rainfall distributions.
Rational Method. The Rational Method is one of the simplest and best known methods routinely applied in urban hydrology. Peak flows are computed from the simple equation Q = kCiA, where Q is the peak flow, k is a conversion factor, C is the runoff coefficient, i is the rainfall intensity, and A is the catchment area.
With WMS's capability to create TINs from feature arc data, roads, railroads, canals, and other urban features which control runoff are easily incorporated into the model so that an urban catchment area can be delineated and the variable A can be automatically computed.
National Flood Frequency (NFF). The National Flood Frequency (NFF) Program was developed by the USGS in cooperation with the Federal Highway Administration (FHA) and the Federal Emergency Management Agency (FEMA). The computer program evaluates regression equations for estimating T-year flood-peak discharges for rural and urban watersheds.
For TxDOT, as well as for other highway agencies, a continuing concern is the need to apply current engineering hydrologic and hydraulic design and analysis procedures that balance simplicity with accuracy. Although most hydrologic and hydraulic calculation procedures are now available in computer programs, the use of which has substantially reduced the mathematical effort involved, a substantial effort is still necessary to establish and manipulate the data required for input into those programs. In trying to simplify the process of determining these input data, the Departments of Transportation of Texas and Maryland have developed GIS systems that calculate spatial hydrologic parameters that can then be used by standard hydrologic software packages.
GISHYDRO, a geographic information system (GIS) structured for hydrologic analysis, was developed and installed in the Maryland State Highway Administration's (MSHA) Division of Bridge Design in Baltimore in 1991 (Ragan 1991). The objective of GISHYDRO was to improve the efficiency and quality of hydraulic design by allowing the user to quickly assemble the landuse, soil and slope data for any watershed in the state, and then make the necessary interfaces to define the required input parameters and run the SCS TR-20 hydrologic model for existing or proposed watershed conditions. A digitizer was used to delineate watershed and subwatershed boundaries, define details of the stream, swale and overland flow paths, and enter areas proposed for landuse change. GISHYDRO then sets up the files for entry into the Soil Conservation Service computer program TR-20, so the model can be run for existing or proposed conditions. The same files are used to run a nonpoint source pollution model that estimates BOD, nitrate, phosphate and other loadings in terms of the watershed landuse and soil types.
HDDS is a prototype system intended to demonstrate the potential capabilities of using GIS for highway-based hydrologic data development and analysis. His system employs data that are now widely available or will become more prevalent.
The focus of Hydrologic Data Development System (HDDS) is on the development of an integrated set of Arc/Info programs and associated data. Though the HDDS programming is specific to Arc/Info, the data are transferable and the general methodology should be applicable to any GIS package that has similar capabilities. The system provides the user with the capability of establishing some of the most important hydrologic parameters used in hydrologic analysis methods, such as the drainage basin boundaries, areas and subareas, the maximum flow path length, the estimated travel time, the watershed average slope, the hydrologic soil group, the design rainfall, the weighted runoff coefficients, and other hydrologic parameters of a catchment defined by a highway/stream crossing. The data may be passed automatically from HDDS to the TxDOT Hydrologic and Hydraulic System (THYSYS), to calculate design flood frequency relationships. The resulting data may then be manipulated to create drainage area maps, tables and other documentation.