catchments and watersheds

7/31/2019 Catchments and Watersheds

1/33

Catchments and Watersheds 1

4

Catchments and Watersheds

Catchments and watersheds define drainage areas of the land surface thatcontribute flow to particular edges on the Hydro Network. An Edge Catchment is the

local drainage area associated with a particular Hydro Edge. A watershed is a

collection of Edge Catchments capturing the entire land surface area drainingthrough a connected set of Hydro Edges. Hydro Response Units define the

hydrologic properties of the land surface.

This chapter includes:

Catchment and Watershed definitions in the ArcGIS Hydro Data Model

Raster based techniques for drainage delineation

Sample raster based drainage delineations

Drainage delineation from the Hydro Network

Hydro Response Units


2/33

2 ArcGIS Hydro Data Model

Catc hments and Wate rsheds

Catchments and watersheds are land areas that drain to a Hydro Network. The determination of

their boundaries is necessary when modeling a hydrologic system. Drainage boundaries are used inwater availability studies, water quality projects, flood forecasting programs, as well as many otherengineering and public policy applications. Accurate drainage boundaries are essential for accuratemodeling studies. Through the ArcGIS Hydro Data Model and associated ArcInfo functionality,

watershed and catchment boundaries may be determined accurately and repeatedly in an automatedfashion.

Manual Delineation of drainage areas on a Topographic Map

When manually delineating watersheds from a topographic map, drainage divides are located byanalyzing the contour lines. Arrows representing the flow directions are drawn perpendicular to

each contour, in the direction of the steepest descent. The location of a divide is taken to be whereflow directions diverge, or where the arrows point in opposite directions. Manually locating these isa difficult process, and slight errors are unavoidable. The extent of these errors is dependent upon

the worker, and different workers will likely present different results. On the contrary, watershedsand catchments determined through the use of raster data will produce consistent results, regardless

of the user. The accuracy of these results will also only be a function of the accuracy and resolutionof the raster data. Even with this accuracy, manual editing of the raster-defined drainage boundariesis still necessary in order to obtain the best possible agreement with topographic data. This chapter

explains the raster-based delineation techniques, and then describes how the watersheds and

catchments derived from these techniques may be incorporated into the ArcGIS Hydro Data Model.The Upper Washita watershed in Oklahoma is presented as an example application.

Def in i t ion o f Catc hment s

A catchment is defined as the land area that contributes runoff to a givenHydro Edge. In theArcGIS Hydro Data Model, catchments are intimately associated with the Hydro Edge about which

they are formed, and as such are referred to asEdge Catchments. Any raindrop falling on an Edge


3/33


Catchment has a unique path to a Hydro Edge and thus to being routed down through the Hydro

Network. Edge Catchments are polygon features in the data model.

Edge Catchments (Polygon Features) Each distinct area (color) is a separate Edge Catchment

Edge Catchments can be created by three methods, one each for catchments associated with Flow

Edges, Virtual Flow Edges, and Shoreline Edges. The respective catchments are named Flow EdgeCatchments, Virtual Catchments, and Shoreline Catchments.

Flow Edge Catchments Catchments that contribute runoff directly to a Flow Edge. The above figure depicts several

Flow Edge Catchments, one for each Flow Edge

.


4/33


Virtual Catchments associated with Virtual Flow Edges in a waterbody

Virtual Catchments These are pseudo-drainage areas associated with a given Virtual Flow Edgeinside a waterbody. The sum of the areas of the Virtual Catchments of a waterbody must equal thearea of the waterbody to maintain a one to one relationship between Hydro Edges and EdgeCatchments throughout the network. The area inside a given Virtual Catchment is closer to its

corresponding Virtual Flow Edge than to any other Virtual Flow Edge. The ArcInfo Grid functionCostallocation accomplishes this subdivision of the waterbody.

Shoreline Catchments associated with Shoreline Edges along the waterbody boundary

Shoreline Catchments - These are catchments associated with a Shoreline Edge. Runoff fromthese catchments flows directly into the waterbody through the Shoreline Edge. Flow Edges do notexist in these catchments.


5/33


The final type of catchment, the Waterbody Catchment, is defined as the local area that directly

contributes rainfall or runoff to a waterbody. Such a catchment consists of all the ShorelineCatchments associated with the waterbody.

Waterbody Catchment A) Waterbody (Blue) with Shoreline Catchments, Shoreline Edges shown in contrasting colors;

B) Waterbody Catchment (Green) Note: Flow Edges are not included in the Waterbody Catchment, Dam shown in

brown

For representation in the ArcGIS Hydro Data Model, each of the catchments described above hasthe following attributes:

OID object ID used to identify each object in the network

GridCode ID field connecting each catchment to the grid cells from which it was formed

Wshed_ID ID field connecting each catchment to the watershed that contains it.

AreaInSqMeters Field containing the area (in m2) of the catchment

Def in i t ion o f Watersheds

A watershed is a polygon feature formed by synthesizing the Edge Catchments of a connected set ofHydro Edges. As such, watersheds are distinct from catchments, defined as areas draining to a given

Hydro Edge. Watersheds are created at Hydro Junctions (intersections of Hydro Edges) on the

Hydro Network, and they define the drainage area of each junction. New Hydro Junctions may becreated by adding Hydro Points along Hydro Edges. The Hydro Edges are then split at the Hydro

Points and new Edge Catchments are created which drain to the edges above and below the points.


6/33


A Watershed defined from a new Hydro Junction A) Hydro Network with multiple Edge Catchments;

B) Watershed (Orange bounded in Yellow) defined from a new Hydro Junction (Pink). Note that the green Edge

Catchment has been divided into two Edge Catchments, one upstream and one downstream from the Hydro Point.

The watershed is the synthesis of all the Edge Catchments draining to a specific Hydro Junction onthe network. Subwatersheds are thus defined as the incremental drainage areas of a set of Hydro

Junctions on the Hydro Network within the watershed. The watershed outlet point may be createdfrom any type of Hydro Point, and must be defined by the model user.

Watershed and Subwatersheds The watershed (yellow outline) is the area draining to the outlet Hydro Junction (red).

It contains several subwatersheds (green outlines), each of which drains either to a non-outlet Hydro Junction (purple)

or directly to the outlet Hydro Junction.

For representation in the ArcGIS Hydro Data Model, a watershed has the following attributes:

OID object ID used to identify each object in the network

Wshed_ID ID field identifying the watershed


7/33


Through the Watershed OID, there is a one to many relationship with Edge Catchment.

Subwatersheds are identified as parts of the overall Watershed using Wshed_ID in a many to onerelationship.

Watersheds are such convenient boundaries for modeling and regulatory purposes that manystandard watershed data sets and coverages currently exist. These data sets vary in scale andaccuracy, factors that must be considered before a given data set is selected for use. Some widely

used current data sets are now described.

Hydrologic Units Developed by the US Geological Survey, Hydrologic Units describe watersheds

from the continental scale to the city scale. These coverages were developed for the United Statesand are separated into 4 classes: Region, Sub-Region, Accounting Units, and Cataloging Units.Each class describes a successively smaller land area. Each Region (the largest area class) is

referred to with a two digit Hydrologic Unit Code (HUC). The Sub-Regions contained within eachRegion are referred to by 4-digit HUCs, with the first two digits equal to that of the Region. In this

scheme, Accounting Units have 6-digit HUCs, and Cataloging Units have 8-digit HUCs. The figurebelow1 shows the Regions of the United States. For more information, see the reference section atthe end of this chapter.

Hydrologic Units Regions of the United States1

Watershed Boundary Dataset Also known as the Hydrologic Unit Boundary dataset, this dataset iscurrently under development by the USDA Natural Resources Conservation Service and the USGS.

It further refines the HUC dataset by adding two data levels below the Cataloging Unit, namely the

Watershed and Subwatershed.


8/33


Watershed Boundary Dataset for Kansas Cataloging Units (Red outline) surrounding Subwatersheds (shaded

polygons) Note that the drainages are artificial ly truncated at the State boundary.

Hydro1K Derivatives This dataset describes continental scale watershed boundaries for every

landmass across the world. The boundaries were developed from 30-arcsecond elevation grids.Continental scale basins represent the lowest level basins in the Pfafstetter basin classification

system. This system, which categorizes basins at increasingly finer scales, is based upon thetopographic control of areas drained on the Earth's surface, as well as the topology of the resultinghydrographic network2.

Pfafstetter Level 1 Basins of North America as derived from the Hydro1K elevation dataset3


9/33


National Elevation Dataset Hydrologic Derivatives (NED-H) This dataset is the result of an

interagency effort between the USGS and the National Weather Service, with the goal of using theNational Elevation Dataset to delineate much smaller catchments than even the Watershed

Boundary Dataset. More information about NED-H is presented subsequently in this chapter.

Local Watershed Coverages Many regulatory agencies use high-resolution elevation data todetermine the watersheds within a small local area. These watersheds are often highly accurate,

depending on the accuracy of the elevation data. For more information on the use of such datasets,see Chapter 9Application to the City of Austin.

Use of Dig i t a l Elevat ion Model Dat a in Drainage Del ineat ion

The standard method for delineating drainage areas (either catchments or watersheds) involves theuse of a digital elevation model, or DEM. DEMs are digital records of terrain elevations for groundpositions at regularly spaced horizontal intervals. These grids are derived from standard topographic

quadrangle maps through the use of hypsographic data and /or photogrammetric methods4. Suchgrids are easily processed with the ArcInfo Grid functions, or with the ArcView Spatial Analyst

functions. The following discussion describes the basic theory behind the watershed delineationfunctions in ArcInfo. These functions have been combined and packaged into the CRWR Pre-Pro5application, which allows watershed delineation in ArcView (with the Spatial Analyst Extension).

Obtaining DEM data will be briefly discussed in the next section.

The Grid operations involved in watershed delineation are all derived from the basic premise that

water flows downhill, and in so doing it will follow the path of with the largest gradient (steepestslope). In a DEM grid structure, there exist at most 8 cells adjacent to each individual grid cell.(Cells on the grid boundary will not be bounded on all sides) Accordingly, water in one cell travels

in 1 out of at most 8 different directions in order to enter another cell. This concept is deemed the 8-direction pour point model.

8-Direction Pour Point Model for Grid Operations

In this grid representation, water at the center of the grid may flow only along one of the eight pathsdepicted by arrows. The number in each cell represents the direction water travels to enter the

nearest downstream cell, and the numbering scheme has been set by convention. The numbers were

determined from the series { }7,...1,02 =xx , which when written in the binary representation usedby computers corresponds to 1, 10, 100,10000000.


10/33


Watershed delineation with the 8-direction pour point model is best explained with an illustration.

For demonstration purposes, assume a section of a sample DEM grid is given in Figure A, B below.The numbers in each grid cell represent the cell elevation.

Slope Calculations with the 8-direction Pour Point Model A) Slope calculated for diagonal cells; B) Slope calculated

for cells with common sides.

Focusing on the center cell (value = 44), only 2 of the 8 adjacent cells contain values less than 44.

This limits the possible flow directions in that water will not flow to a cell with a greater elevation.The water will flow in the direction in which the greatest elevation decrease per unit distance isobtained. In A), this slope is calculated along the diagonal by subtracting the destination cell value

from the original cell value, and dividing by 2 , the distance between the cell centers assuming

each cell is 1 unit long on each side. In B), the slope is calculated to the non-diagonal cell. It isequal to the elevation difference because the distance between the cell centers is unity. In this case,the diagonal slope is greatest, and water will flow toward the bottom right cell. The center cell is

then assigned a flow direction value of 2. This process is then repeated for each of the cells in theDEM grid, and a new grid is created to store the results of the calculations. This new grid, called theFlow Direction grid, contains cells with only the numerical values dictated by the 8-direction pour

point model.

Grid Operations A) DEM Grid; B) Flow Direction Grid. Note: Area in red is from the previous figure


11/33


Physical Representation a Flow Direction Grid A) with directional arrows; B) As a flow network

It is from the flow direction grid that the flow accumulation grid is calculated. This grid records the

number of cells that drain to an individual cell in the grid. Note that the individual cell itself is notcounted in this process.

Flow Accumulation number of cells draining to a given cell (blue) along the flow network

At this point, it is necessary to consider the possibility that flow might accumulate in a cell in theinterior of the grid, and that the resulting flow network may not necessarily extend to the edge of thegrid. An example of such a situation is the Great Salt Lake in Utah, which is an interior sink. None

of the precipitation that falls on the Great Salt Lake watershed travels through a river networktoward the ocean. This situation poses a problem for automated delineation, for the flow that

accumulates in an inland sink does not reach an outlet from which the delineation process maytake place. A second potential problem arises, however, if the DEM grid itself contains artificiallows in the terrain, due to errors in elevation determination or grid development. These artificial

sinks must be eliminated in order to accurately delineate watersheds.

Any artificial sinks or inland catchments in the DEM are removed through the use off the Fill

function in ArcInfo. This function alters the elevations of the offending cells through the use of aninterpolation function, as shown below.


12/33


Filling an artificial pit in the DEM

To allow for the existence of known inland catchments in the DEM, special processing is neededbefore running the Fill function. One such method is to assign a NODATA value to the lowest

elevation cell in the inland catchment; such a cell would be treated as an outlet in that it wouldallow water to flow out of the system. In the network model, this is a Hydro Edge that ends in asink.

DEM alteration to allow delineation of inland catchments

The Fill function should be run before the flow direction grid is created because artificial pits cansignificantly alter the flow direction. Similarly, if inland catchments are present, a program should

be run to handle them before the Fill function is run or the flow direction grid is created.

With a flow accumulation grid, streams may be defined through the use of a threshold flow

accumulation value. For example, if a value of 5 were set as the threshold, than any cell with flowaccumulation greater than 5 would be considered a stream. Cells with flow accumulations greaterthan or equal to the threshold are given a value of 1 in a newly created Stream Grid, with all other

cells containing the value 0.

Stream Definition from the Flow Accumulation Grid and a threshold value A) Grid cells with accumulation greater

than or equal to 5 are considered stream cells (red); B) Streams identified on the flow network (red)

The next step is to divide the stream network into distinct stream segments this is useful if thepurpose of the delineation is to determine the individual Edge Catchments. If only the overall


13/33


watershed is desired, the delineation function could be used on the established grids as long as the

outlet cell is defined. For this discussion, Edge Catchments will be delineated.

Stream Links defined A) Stream Grid representation, B) Stream Links (numbers) defined, link outlets (blue),

watershed outlet (red)

To segment the stream network, the Streamlinkfunction in ArcInfo is used on the Stream Grid, as in

the figure above. This function determines stream links within the network, and assigns each link aunique number, or GridCode. Each cell within a link is assigned the same number. The mostdownstream cell in each link is the link outlet cell. An outlet grid, with the individual outlets cells

(in B), the blue or red cells) containing GridCodes and all other cells containing NODATA is thenproduced from the streamlink grid.

At this point additional outlets may be added to the outlet grid, and the Stream Link Grid is thenadjusted to add the newly created links. Such extra outlets, represented asHydro Points in the

ArcGIS Hydro Data Model, may represent water supply locations, monitoring stations, etc. Thefollowing screen captures from ArcView demonstrate these actions.

Adding outlets on a stream grid A) an outlet (red dot) is added on a stream (black); B) A new stream link (red) is

added to the stream link grid, and the original link (black) is adjusted.

The watersheds may now be delineated through the use of the Watershedfunction in ArcInfo. Thisfunction uses the flow direction and outlet grids to determine all of the cells that drain to each

outlet. It assigns each of these cells the value of the of the outlet cell (which also corresponds to thevalues for the cells in the stream link grid). The results of the delineation are stored in a watershedgrid.


14/33


Delineated Edge Catchments

To this point, all of the data has been in raster format. For further processing, the streams andwatersheds are vectorized based on the grid value fields. The grid value is transferred to the

gridcode of the corresponding vector object. Similarly, the vector representation of a delineatedcatchment has a gridcode attribute equal to the value of the catchment grid cells.

Grid streams and vector streams linked by value and gridcode

A similar process is carried out for vectorizing watersheds, however a complication arises throughthe creation of spurious polygons. Due to the regular, gridded form of the raster data, irregularlyshaped watershed borders often consist of grid cells connected through the cell corners. When

vectorized, these cells may become isolated from the main watershed, forming spurious polygons.

Spurious polygon created during vectorization

These polygons will be vectorized as distinct drainages, thereby yielding an artificially high numberof drainages in an area. An automated procedure can be used to detect the existence of such


15/33


polygons, and eliminate them by dissolving them into the surrounding polygon. This dissolving

process can be carried out with the GeoProcessing Wizardfunctions in ArcView.

Exam ple Raster Ana lys is o f Dra inage Areas

The accuracy of drainage delineation from DEMs depends on the resolution of the grid, namely thegrid cell size. Smaller cell sizes allow for a more accurate representation of the topography,especially in mountainous areas. However, higher resolution grids require more cells to cover thesame area as a lower resolution grid. Increasing the number of cells stresses the processing

capability of todays computers. It has been only with the recent advances in computer technologythat even 30m DEM grids have become functionally usable on a watershed scale. These grids have

been made available by the United States Geological Survey, and are referred to as the NationalElevation Dataset (NED).

The NED is officially described as seamless raster format elevation grids from the conterminous

United States at 1: 24,000 scale5. The elevations are given in decimal meters, using NAD83 as the

horizontal datum, and are available in geographic projection in 1x1 tiles with 1 arc-second cells.

The U.S. Geological Survey, along with maintaining the NED and periodically updating the dataset,is currently involved in using the NED to re-evaluate and refine the presently recognized HUC

boundaries. These boundaries were created with DEMs of lower resolution than the NED, and assuch the accuracy of the boundaries may now be improved. Efforts are also underway to use therecently developed National Hydrography Dataset (NHD) in order to develop a hydrologically

accurate NED. This program is responsible for development of the National Elevation Dataset Hydrologic Derivatives (NED-H).

One goal of the NED-H development is to obtain the ability to determine drainage basin boundariesupstream of any point in the conterminous United States6. All locations downstream from a givenpoint will also be easily discernible. This goal corresponds well with part of the impetus behind

development of the ArcGIS Hydro Data Model, which is able to use the network tracingfunctionality of ArcInfo 8 to trace upstream and downstream from user defined points in the Hydro

Network. This process is demonstrated with the Upper Washita watershed in Southern Oklahoma.The data and procedures presented are available for review at the NED-H website7.

3-D Representation of the Upper Washita Watershed and surrounding area7


16/33


Raster Analysis of the Upper Washita Watershed

In studying the current GIS hydrography of the Upper Washita Watershed, it is realized that the oldHUC boundary is no longer accurate. In at least one location, rivers flowing through the watershed

(as obtained from the National Hydrography Dataset) also traverse the watershed boundary.

Upper Washita Watershed Cataloging Unit Boundary and NHD reaches, waterbodies for the area (center); Watershed

location in Oklahoma (right); NHD reaches crossing the boundary, indicating a boundary inaccuracy (left)

To eliminate this and other discrepancies, the area drainages were re-delineated using the 30m NEDwith the delineation procedure described previously. An upstream drainage area threshold ofapproximately 4.5 km2 was used in defining streams.


17/33


Re-delineation Results delineated streams show reasonable agreement with NHD, watershed boundaries also similar

In general, the two boundaries showed reasonable agreement, although the NED boundary includedthe NHD reach that was previously identified as a problem. The NED defined reaches matched wellwith the NHD reaches, except for the reaches of the highest order, and those reaches flowing

through the flatter areas.

To ameliorate this problem, the DEM was altered on a cell-by-cell basis until the delineated reaches

matched the NHD reaches to a high degree of accuracy. In this process a new NED was created,

called aHydrologically Conditioned DEM (HC-DEM). It was from this DEM that new flowdirection, flow accumulation, stream, stream link, outlet, and watershed grids were created. The

resulting watersheds and subwatersheds, corresponding to the potential additional WatershedBoundary Dataset levels, were determined.


18/33


Delineation Results - Watersheds (Red outline) with outlets (black dots); Subwatersheds (colored polygons)

It should be noted at this Subwatershed delineation level, there are not an equal number ofsubwatersheds and Flow Edges. In order to obtain the Edge Catchments associated with each FlowEdge in this Hydro Network, the watershed delineation process was carried out with the 30m HC-

DEM provided from the NED-H effort. The catchments derived from this effort show a one-to-onerelation between Flow Edges and Edge Catchments. These catchments were then incorporated into

the ArcGIS Hydro Data Model.


19/33


Upper Washita Edge Catchments A) individually colored Edge Catchments, B) Edge Catchments with Flow Edges

(from NHD), Watersheds, and Watershed Outlets (Hydro Points) .

As previously stated, one of the goals of the NED-H program is to obtain the ability toautomatically determine the upstream and downstream elements from a given point in a hydro

network. This goal is currently obtainable with the ArcGIS Hydro Data Model in combination withArcInfo 8, as demonstrated in the following discussion using the Upper Washita Hydro Network.

The first step in this process is to load the appropriate data sets into ArcCatalog and the ArcGISHydro Data Model.

To carry out a network trace, it is only necessary to have a Hydro Network consisting of Hydro

Edges. In terms of the Upper Washita Watershed, only the Flow Edges are needed. In thisdiscussion, the Edge Catchments and Hydro Points from B) will also be included. The Edge

Catchments themselves do not form part of the Hydro Network created in ArcInfo 8, but they areassociated with the network through relationships defined in the ArcGIS Hydro Data Model.


20/33


The first step in this process is to load the existing shapefiles (the Edge Catchments, Flow Edges,

and Hydro Points) into the ArcGIS Hydro Data Model through the import function in ArcCatalog.ArcCatalog then transfers this data into a newpersonal geodatabase in Microsoft Access, carrying

the .mdb filename extension. Once the geodatabase is created, a model schema must be createdfrom the UML database containing the ArcGIS Hydro Data Model. This step is carried out byretrieving the UML model from its repository and then using this connection to establish the feature

classes to be used to represent the Upper Washita Watershed. In this case, userpoints,Hydro Edges,andEdge Catchments were instantiated in order to develop the desired feature classes. At this point,

a geometric network is created from the data in ArcCatalog. This network is then transferred intoArcMap in order to carry out the desired network operations.

Network Representation of the Washita.mdb geodatabase in ArcMap

Shown is shows the network representation of the Washita.mdb geodatabase in ArcMap. The

section on the right lists the data layers that were instantiated from the UML repository. Note thateight layers are listed. This is a result of the relationships between features as required by the

ArcGIS Hydro Data Model UML coding. For example, Edge Catchments are related to Watershedsin the data model, and in instantiating the Edge Catchment feature class, the Watershed class is


21/33


automatically instantiated as well. Similar relationships between feature classes caused the

instantiation of four extra feature classes, yielding seven total data layers. Of these seven, only thethree layers imported into the geodatabase contain spatial data; each of the other layers is empty.

The final data layer was created while constructing the new geometric network. This layer, shownas WashitaNet_Junctions , is similar to the outlets file created in watershed delineation. Thesenetwork junctions are shown as blue diamonds in the ArcMap view.

Before a network trace may be carried out, the direction of flow through the network must bedetermined. This is accomplished by assigning to the downstream outlet(s) the ancillary role of

sink. This directs ArcMap to direct all flow on the network toward the designated outlets. Once thesinks are defined, the display may be set to show the flow direction on the network.

Flow Direction displayed on the Hydro Network in ArcMap Triangles point downstream

These directions are depicted by the black triangles on the network, with the triangles pointing inthe downstream direction. With the flow direction established, the Trace Task functions becomeactive, allowing for a variety of traces to be conducted.


22/33


To conduct either downstream or upstream traces, a flag must first be set on the network. The

location of this flag represents starting point of the trace. Such flags and their trace results areshown below for the Washita network.

Washita Network Traces A) Downstream trace (red) from flag (green square); B) Upstream trace

Although the trace is conducted on the Hydro Edge network, the Edge Catchments associated witheach Flow Edge selected in the trace may also be selected through the common GridCode attribute.

This allows the user to determine the land areas upstream of a given point on the Hydro Network,and to determine the land areas that border the flow downstream from that point.

Def in ing Edge Catc hment s us ing the Hydro Netw ork

The previous section has described a raster-based method of drainage delineation where syntheticstreams and their corresponding catchments are created from the raster grid. An alternativeapproach is to take the vector Hydro Network and build the Edge Catchments around its Hydro

Edges without creating any synthetic streams. This could be called the network-based drainagedelineation procedure.

Drainage delineation with a Hydro Network uses many of the same techniques as the solelyelevation-based method, but rather than locating streams by a certain flow accumulation, the streamlocations are fixed by the Hydro Network. This process, as described below, reduces delineation

error by assuming great accuracy of the Hydro Network. This assumption is often justified,especially if the Hydro Network is derived from aerial photogrammetry.


23/33


Delineation with a given Hydro Network involves altering the DEM data in order to obtain known

hydrologic conditions at certain areas. The case of known stream locations is discussed first.

The process of using known stream locations in the delineation procedure is referred to as burning

in the streams. Through this procedure, the DEM grid is modified to force waterflow into theHydro Network. The key to understanding this method is the realization that individual elevationvalues in the DEM may be uniformly changed without altering the resulting delineation.

DEM and Flow Direction grids with stream cells highlighted (red)

The DEM and Flow Direction grids shown here are identical to those used in the previous section,

except certain cells are highlighted in red. These cells correspond to the stream cells determinedfrom the flow accumulation grid with a 5-cell stream threshold. The same grids are now shownagain, except that the elevation of each of the non-stream cells is increased by 100.

Burned DEM and Flow Direction grids with stream cells highlighted (red)

In this example, all but the known stream cells were raised by an arbitrary amount. The result is thatthe stream cells form a trench in the DEM. Water is made to flow into this trench and down the

stream network to the outlet. The phrase burning in the streams is appropriate because the streamshave been forced, or burned into the topography described by the DEM. This new DEM is

referred to as aBurned DEM.

As shown, the flow direction grid derived from the burned DEM is identical to that from theoriginal DEM. This is because the flow direction is partly determined by the relative difference in

cell elevations, which is unaltered by the uniform elevation increase. Only the slopes calculated forthose cells adjacent to the non-raised stream cells will be altered, and these new slopes are

sufficiently large to assure flow into the stream cells.

.


24/33


Burning in the Streams Adjusting the DEM to form a trench (black) surrounding the stream network (blue)

In order to use an established Hydro Network to create burned DEMs, the vector network isconverted to a grid, and this resulting network grid is intersected with the DEM. Cells in the DEM

that correspond to cells in the network grid remain unaltered while all of the other cells in the DEMare raised. The actual increase in elevation is arbitrary, except that it must be greater than thehighest elevation of the unaltered DEM cells. This assures that flow in the trenches remains in the

trenches until the furthest downstream outlet is reached. Once the network is burned into theDEM, the delineation process continues as previously described.

The results of the DEM-burning process are shown below for the Guadalupe Basin in South CentralTexas. The stream network was extracted from the Reach File 1 dataset, and the DEM cells are500m in length. The burned DEM, shown in B), was created from the original DEM and stream

network in A) by raising the elevation by 5000 units. The white lines in B) are formed from theburned DEM cells corresponding to the original stream network. It should be noted that the burned

DEM does not appear as detailed as the original DEM. This is a result of the burning process, whichincreases the range of elevation values to be displayed, and thereby increases the range of elevationsdescribed by each color in the display.

Burning in the Guadalupe Basin Hydro Network Basin boundary (red) displayed as a reference.

Further examples of the use of the network burning process are given in Chapters 8-10.


25/33


A similar process of DEM alteration can be employed if known drainage boundaries are to be

incorporated into the DEM. Such a situation may arise if the DEM is of too low resolution todescribe local drainage barriers such as berms or elevated highways. This process, referred to as

Building Walls, involves raising the elevation of a connected set of DEM cells in order to preventflow into the cells. This essentially the opposite of the network burning process, for here the cellscorresponding to the known boundary are elevated while all of the other DEM cells retain their

original values. An example of this process as applied to the City of Austin is given in Chapter 9.The figure depicts the building of walls about the border of the Guadalupe Basin discussed

previously. The boundary walls, shown in B), were created from the original DEM and basinboundary in A) by raising the boundary elevations by 5000 units

Building walls around the Guadalupe Basin Hydro Network (blue) displayed as a reference

A variation to the previously discussed delineation techniques has been developed in order todelineate Shoreline Catchments. One complication associated with this Edge Catchment type is thatunlike Flow Edges, Shoreline Edges do not have an outlet at one end. In a physical sense, flow is

transferred from the catchment across the Shoreline Edge to the waterbody, instead of beingtransferred along the length of the Shoreline Edge. Also, flow is not to be transferred from the

waterbody to the Shoreline Edge. This distinquishes Shoreline Catchments from the other HydroEdges, which may receive flow from either side. The previously discussed delineation method doesnot allow for these constraints, unless modifications are made to the grids involved.

The key modification is to consider the entire Shoreline Edge as an outlet through which water fromthe Shoreline Catchment drains to the waterbody. In raster format, each of the grid cells making up

the Shoreline Edge is considered as a separate outlet cell. These cells are connected through acommon GridCode attribute, which identifies the cell as part of the Shoreline Edge. The drainageareas to each outlet cell are determined by the previously established methods, and the grid cells in

each of these drainage areas contain the common GridCode attribute. The Shoreline Catchment isthen determined from the aggregation of all cells containing the GridCode of the Shoreline Edge.


26/33


Representation of Shoreline Catchment Delineation A) Shoreline Edge in raster format, each cell as an outlet. B)

Aggregated drainages to outlets form Shoreline Catchment.

The DEM used in the delineation process assigns elevation values to cells corresponding to thewaterbody, and as such it incorporates these cells in the delineation process. In order to eliminate

these cells from the process, it is necessary to change their elevation values to NODATA. This isaccomplished with theIsNull function in ArcInfo. The DEM cells corresponding to the waterbody

are identified by rasterizing the vector waterbody and then intersecting the waterbody grid with theDEM. This process is described fully in Chapter 10.

The delineation process just described for Shoreline Catchments is equally valid for determining

Flow Edge Catchments. The only difference between the two processes is the conversion ofwaterbody cells to NODATA cells in the DEM for Shoreline Catchment delineation. This process is

not applicable to the delineation of Virtual Catchments. This is due to the fact that VirtualCatchments are not defined by flow directions, but rather by waterbody and Hydro Network

geometry.

Virtual Catchments, as defined above, are the catchments associated with Virtual Flow Edges.These Hydro Edges serve to connect Flow Edges that drain into or flow out of the waterbody, and

are used to maintain edge-to-edge downstream connectivity through the Hydro Network. As shownbelow, the use of Virtual Flow Edges eliminates the uncertainty in flow along the Shoreline Edge.


27/33


Flow Routing through a Waterbody A) Flow travels along the Shoreline Edges (various colors); B) Flow travels

through the waterbody on Virtual Flow Edges (black). Arrows depict flow direction

As in A), in order to route flow through the Hydro Network, a flow direction must be assignedalong the Shoreline Edges. The appropriate direction of flow is indeterminate because the Shoreline

Edges form a loop in the Hydro Network. The resulting ambiguity in flow direction is demonstratedby the arrows on the Shoreline Edges. The situation in A) also suggests that flow travels around the

waterbody rather than through the waterbody. This model does not represent the actual physicalmovement of water in the environment.

The inclusion of Virtual Flow Edges in B) eliminates any ambiguity in flow direction, and water is

routed through the waterbody. The location of the Virtual Flow Edges is essentially arbitrary,although often they are drawn over the waterbody thalwag. The location is not derived from the

DEM, and therefore determining the Virtual Catchment from a DEM is not possible. If the HydroNetwork is derived from the National Hydrography Dataset or another such dataset, the VirtualFlow Edges may already exist in the network. If they do not exist, they must be manually added for

use in the ArcGIS Hydro Data Model, or created using the ArcInfo Grid function Thin.

The delineation of Virtual Catchments is determined by selecting those sections of the waterbody

that are closer to a given Virtual Flow Edge than to all other Virtual Flow Edges. In determining theclosest Virtual Flow Edge, the ArcInfo CostAllocation function is run on a raster representation ofthe waterbody. This function assigns to each waterbody cell the gridcode of the nearest Virtual

Flow Edge. The figure below displays the Virtual Catchments for the example given in the previous

picture as determined by this technique.


28/33


Virtual Catchments (Colored Areas) Defined with the CostAllocation Function

Hydro Response Uni t s

A Hydro Response Unit is an area of the land surface that has homogeneous precipitation and/or

land surface characteristics. It is used in hydrologic modeling simulations in order to accuratelydescribe and predict how water will move through the environment. By intersecting the Hydro

Response Unit coverage with the Edge Catchment or Watershed coverage, the contributing area ofeach Hydro Response Unit within a given Edge Catchment or Watershed can be determined. In the

figure, the Hydro Response Unit cells may describe areas that receive equal precipitation. Thisrepresentation of precipitation is common in weather forecasting and reporting, as shown in Chapter8for the Lower Colorado River Basin.

Hydro Response Unit Cells (Dashed Boxes) overlain on Drainage Areas

As mentioned in Chapter 2, Hydro Response Units account for the vertical exchange of waterthrough the hydrologic cycle. The example above considered precipitation, which is a method of

transferring water from the atmosphere to the land surface. Other Hydro Response Units considercharacteristics such as hydraulic conductivity, which is involved in the transfer of surface water to


29/33


groundwater systems. Still other units are used in describing how precipitation is transformed to

runoff on the land surface, and how this runoff migrates toward the ocean. Three general classes ofHydro Response Units can therefore be defined:

1. Units linking Atmospheric and Surface processes2. Units linking Surface Processes and Subsurface processes, and

3. Units linking Surface Processes to other Surface processes

Each of these three general classes is discussed below. It is necessary to understand that HydroResponse Units are not limited to one of these three broad categories, and the combinations ofhydrologic characteristics can yield complicated Hydro Response Units. The type of Hydro

Response Unit used will likely depend on the data available and on the purpose of the modelingstudy.

Due to the spatial variability of land surface attributes, it is extremely unlikely that most actualHydro Response Units will assume a uniform grid-like shape as shown above. The following

examples were selected to demonstrate the affect of varying characteristics on the vertical water

balance of an area, and they demonstrate the spatial variability of their defining characteristics. Oneexample is the complex surface characteristics of Chippewa Lake in Wisconsin.

Chippewa Lake A Hydro Response Unit with land areas and water

Representing every island and every lake within an island in the Hydro Network for Chippewa Lakewould unnecessarily complicate the network. Instead, the land and water surfaces may be identifiedas separate Hydro Response Units within the waterbody, which allows them to have different

hydrologic properties and therefore different responses to hydrologic models.

In considering Hydro Response Units for determining surface runoff from storm events, terrestrial

characteristics such as land use and soil type are important. A standard runoff-prediction method,the Soil Conservation Service Curve Number method, considers both of these characteristics. Inareas with high porosity soils, infiltration is generally increased, thereby reducing the overall

amount of surface runoff. Similarly, land use affects runoff by altering the permeability of the soil.Urban areas with many buildings and much pavement have highly impervious surfaces, and often

experience significant runoff. Conversely, agricultural areas, or areas of greater open space, tend to


30/33


have less runoff because the land surface is more receptive to infiltration. Much spatial data is

available describing both of these features, and may be incorporated into Hydro Response Units.

Surface Characteristics A) Land Use Classifications; B) Statsgo Soil Classifications

The land use and soils data for locations in Texas are shown. In A), rangeland (brown) and

forestland (green) are seen in proximity to a waterbody. This land use coverage, obtainable throughthe U.S. EPA BASINS program8, clearly displays the non-uniformity in land use over relatively

small areas. The soils data shown in B) also displays great heterogeneity. The area shown is from anurbanized area in South Austin, consisting of parkland (green) about the river with the surrounding

areas significantly covered with pavement. This data was obtained from the Statsgo9 database forcentral Texas.

The two main processes linking the atmosphere and the land surface with regard to water transfer

are precipitation and evaporation. Each of these processes is dependent on the geographic locationof the land area, and evaporation is also highly determined by the land use and season. Both

processes are commonly recorded at specific points on the land surface, and method have beendetermined to interpolate the respective values for the entire area based on only a few pointmeasurements. One standard method is the use of Thiessen polygons. By this method, the data

measured at one point is assumed representative of all the area closer to that point than to any otherpoint where measurements were made. This is similar to the method used to delineate Virtual

Catchments as described previously. By this method, the land surface is sectioned into polygons.

Thiessen Polygon Method for spatially distributing point measurements


31/33


The division of the land area occurs by intersecting the perpendicular bisectors of the lines

connecting each of the measuring points. The outer boundary is often set at a certain distance fromthe points or at another known physical location. This process is easily carried out manually or with

the Thiessen function in ArcInfo.

Another common method of representing precipitation onto the land surface is through the use ofNexrad Cells. These cells describe atmospheric moisture through the use of remote sensing

technology. This data is available in grid format, as shown in the figure below. This grid isintersected with a surface grid in order to determine the quantity of precipitation that has fallen onto

each land area.

Nexrad Data September 5, 1996 Hurricane Fran10

An example Hydro Response Unit application using Nexrad data is discussed in Chapter 8.

The third general type of Hydro Response Unit, that involved in subsurface to surface waterexchanges, is often characterized by heterogeneous grids. These grids store numerous values for use

in groundwater flow models, such as hydraulic conductivity, porosity, depth, and initial saturation.These properties are rarely homogeneous and isotropic over large subsurface areas, and as such thehydrologic response units derived for these areas will consist of irregularly shaped grids. A popular

groundwater flow model, MODFLOW, allows for the use of multi-layered, irregularly spaced gridsin its operation, as shown below.

Groundwater Modeling Cells overlain by rivers and terrain boundaries

Overlain on this grid are the geographical features of the study site, namely the rivers and property

boundaries (green). It is obvious that these boundaries do not coincide in any way with the grid


32/33


cells, thereby highlighting the difference between the two datasets. This spatial difference in the

Hydro Response Units and the geographical data becomes important when the two datasets arecombined.

Both for display and for analysis purposes, it is often necessary to combine Hydro Response Unitdata with geographic data from a Hydro Network. For example, comparing storm locations withcatchment and watershed locations my help determine the threat of flooding to a town or

establishment. This intersection is easily accomplished in ArcInfo and the results may beincorporated into the ArcGIS Hydro Data Model.

Thiessen Polygons overlain on the Upper Washita Edge Catchments

In this figure, a set of Thiessen polygons is overlain on the Upper Washita Edge Catchments. Theindividual polygons cover many individual Edge Catchments, and the polygon edges always divide

individual catchments. This division is shown in the left figure, which is a close-up view of theouter edge of the polygon network. For those catchments completely contained in one of thepolygons, the received rainfall is assumed equal to that measured at the measuring point. For those

catchments spanned by multiple polygons or on the polygon outer boundary, the received rainfallvaries depending on the location in the catchment. If this scenario were used in a modeling study, it

may be necessary to divide these catchments so that each catchment may have only one value forreceived rainfall.


33/33

ArcGIS Hydro Data Model Objec t Diagram for Net w ork Representa t ion

References1

http://wy-water.usgs.gov/projects/watershed/whatrhucs.htm (6/4/00)2http://edcdaac.usgs.gov/gtopo30/hydro/P311.html (6/9/00)3http://edcdaac.usgs.gov/gtopo30/hydro/na_basins.html (6/4/00)4http://edcwww.cr.usgs.gov/nsdi/gendem.htm (6/6/00)5http://edcnts12.cr.usgs.gov/ned/ (6/6/00)6http://edcnts12.cr.usgs.gov/ned-h/index.html (6/6/00)7http://edcnts12.cr.usgs.gov/ned-h/washita.html (6/6/00)8http://www.epa.gov/OST/BASINS/ (6/15/00)9http://www.gis.uiuc.edu/nrcs/statsgo_inf.html (6/15/00)10http://www4.ncdc.noaa.gov/cgi-win/wwcgi.dll?WWNEXRAD~Images2 (6/15/00)

catchments and watersheds

Documents