University of Texas at Austin

GIS Based

Flood Damage Reduction Analysis

Of

Mill Creek, Ohio

By Peter B. Andrysiak Jr.













       TABLE OF CONTENTS
            OBJECTIVES
            WATERSHED DELINEATION
            SOILS DATA
            LAND USE AND LAND COVER DATA
            CURVE NUMBER DEVELOPMENT AND SOIL STORAGE
            DEVELOPING A HYDROLOGIC MODEL
            HYDROLOGIC MODELING SYSTEM (HEC-HMS)
            CONCLUSIONS
            FUTURE WORK
            DATA DICTIONARY
            REFERENCES



  INTRODUCTION

The work in this project is a rather lengthly compilaton of the procedures used in ArcView GIS and HEC-HMS.  For the purposes of the GIS final, most of what is found in this documentation was learned and applied  throughout the semester.  As a result, only certain portions should be read along with the conclusions and future work in order to understand the purpose and results of my research.  The major deviations from in class prcedures will be found in the soil storage portion of CURVE NUMBER DEVELOPMENT AND SOIL STORAGE and the complete HYDROLOGIC MODELING SYSTEM (HEC-HMS) portion.

This docment is detailed and covers a step by step procedure for the major topics found in the table of contents.  This work is being done to aid the Louisville Corps of Engineer District in understanding how GIS can be a useful tool in flood studies.  Currently the district is conducting a feasibility study on Mill Creek, Ohio after an economic re-evaluation in 1997 showed that a project that was only partially completed in 1992 will result in significant damage for a 50% event. Due to its history of flooding problems, Mill Creek had two Local Cooperation Agreements by 1976. This Cooperation was 50/50 cost shared activity with local and federal agencies. Construction began in 1981 and was stopped in 1992 at 50% completion. Some of the reasons that contributed to the projects termination were: real estate issues, costs rising 126%, hazardous wastes problems from leaching landfills, and maintenance problems with the completed portions. Mill Creek flows from the southeastern part of Butler County in a southerly direction across Hamilton County and through the City of Cincinnati to its confluence with the Ohio River at approximately river mile 472.5. The total fall in elevation of the thalweg from the headwaters of Mill Creek to the mouth, over a distance of approximately 28 stream miles, is about 250 feet, with an average gradient of 8.9 feet per mile.

The Mill Creek watershed is approximately 165 square miles in area. In the upper half of the basin the valley bottom is wide, averaging 1-1/2 miles, but it narrows in the downstream reaches, averaging only ½ mile through the City of Cincinnati. In the lower portions of the basin, valley walls are steep, rising 200 to 300 feet above the valley floor, but are much less steep in the upstream areas. In 1996, the environmental interest group, American Rivers, designated Mill Creek as the most threatened urban stream in North America.
 

TOP



OBJECTIVES
  TOP

WATERSHED DELINEATION

ArcView GIS and its Spatial Analyst extension provide tools necessary for the Watershed delineation of Mill Creek. To compliment these tools, the Center for Research in Water Resources at the University of Texas at Austin and other organizations have developed a watershed characterization system currently named CRWR-PrePro (PreProcessor). CRWR-Prepro used in conjunction with ArcView GIS and the Spatial Analyst extension can process a digital elevation model of any basin region and delineate its watersheds and stream networks. Below is the step-by-step procedure used to process the watershed and prepare it for use in the Hydrologic Engineering Center’s Hydrologic Modeling System (HMS).
 
 

1. Obtaining The Data.
 
 

Data for use in GIS is readily available on the internet at no cost to users. The data can be downloaded from USGS and EPA websites found below. The key to finding the required data is the knowing the required hydrologic unit.

The United States is divided and sub-divided into successively smaller hydrologic units which are classified into four levels: regions, sub-regions, accounting units, and cataloging units. The hydrologic units are arranged within each other, from the smallest (cataloging units) to the largest (regions). Each hydrologic unit is identified by a unique hydrologic unit code (HUC) consisting of two to eight digits based on the four levels of classification in the hydrologic unit system.

The first level of classification divides the Nation into 21 major geographic areas, or regions. These geographic areas contain either the drainage area of a major river, such as the Ohio region, or the combined drainage areas of a series of rivers. Eighteen of the regions occupy the land area of the conterminous United States. Alaska is region 19, the Hawaii Islands constitute region 20, and Puerto Rico and other outlying Caribbean areas are region 21.

The second level of classification divides the 21 regions into 222 subregions. A subregion includes the area drained by a river system, a reach of a river and its tributaries in that reach, a closed basin(s), or a group of streams forming a coastal drainage area.

The third level of classification subdivides many of the subregions into accounting units. These 352 hydrologic accounting units nest within, or are equivalent to, the subregions.

The fourth level of classification is the cataloging unit, the smallest element in the hierarchy of hydrologic units. A cataloging unit is a geographic area representing part of all of a surface drainage basin, a combination of drainage basins, or a distinct hydrologic feature. These units subdivide the subregions and accounting units into smaller areas. There are 2150 Cataloging Units in the Nation. More detals on hydrologic can be found at the followoing link http://water.usgs.gov/GIS/huc.html. Mill Creek falls within the following HUC, 05090203.

Before beginning a download, two folders must be created. The two folders are Data and Millcreek. The Data folder is used to store all downloaded files and .zip files. This folder will ensure that a clean copy of the data remains preserved in case it is needed later. The Millcreek folder is the project working directory. All necessary project files will be copied or extracted to this folder from the Data folder. Within the Millcreek folder a tmp folder must also be created as required by ArcView.

Better Assessment Science Integrating Point and Nonpoint Sources (BASINS). http://www.epa.gov/OST/BASINS/ This EPA website is mainly for using GIS to assess point and nonpoint sources. The site is a great source of GIS data for download. Once on the site go to Download/Basins GIS Data. From the United States map, Ohio was selected. The next page displayed the HUCs of Ohio were the general location of Mill Creek was then selected. A page titled Middle Ohio-Laughery HUC 05090203 revealed three data download links. The links are titled BASINS Core Data (05090203_core), Digital Elevation Model (DEM)( 05090203_dem), and Reach File Version 3 (RF3)(05090203_rf3). The Rf3 files for neighboring HUCs 05090202 and 05080002 were also necessary for the procedure as well. The following is a more detailed description of the files that are found within the three downloads.

Spatially Distributed Data

Environmental Monitoring Data Point Source Data The Center for Research in Water Resources at the University of Texas at Austin produces on an annual basis a collection of exercises, data, reports and information designed to support the use of Geographic Information Systems in Hydrology and Water Resources. It is produced in CD form and can be found at the following link. http://internetcity.crwr.utexas.edu/gis/gishydro99/GisHyd99.htm On the main page under Watershed Characterization a number resources can be found. Essential for watershed delineation is Prepro04.apr, which is found on this page. Prepro04.apr is an ArcView project file that contains the Scripts, Menus and Buttons for running CRWR-Prepro. This project file must be downloaded and placed in the data file.

The Louisville Corps of Engineer District provided the Digital Elevation Model (DEM) for this project. The DEM is a 10 by 10 seamless version made up of 30 meter by 30 meter cells. This is the highest quality DEM available for this analysis and is better then the version found on the BASINS site.

After all necessary download are complete, the three basins .zip files were extracted from Data to Millcreek. In addition, PrePro04.apr was copied and placed in Millcreek.

TOP

2.  Setting DEM Parameters.

A new DEM comes projected using the Geographical Coordinate System. Elevations were specified in meters. Chapter 6, Cartographic Guidelines, of the Ohio Geographic Information Management Systems (GIMS) Program website was accessed at the following link: http://www.dnr.state.oh.us/odnr/occ/gims/hb_chpt6.as In this chapter, there are recommended Projection/Coordinate Systems that are in accordance with surveying law (ORC 157). The Ohio State Plane Coordinate System (OSPC) was chosen for this project. The OSPC System is a planer Cartesian coordinate system based on a Lambert Conformal Conic projection. Ohio has been divided into two zones, North and South, to restrict distortion to no greater than 1 in 10,000, or roughly one foot for every two miles. The input and output parameters were specified in ArcInfo by starting the program and specifying the working directory where the DEM is stored. The project command was then used, which identifies the input projection found below:
INPUT

PROJECTION GEOGRAPHIC

/*DATUM NAD83

ZUNITS METERS

UNITS DD

SPHEROID GRS1980

XSHIFT 0.0

YSHIFT 0.0

PARAMETERS

-85.000

-84.000

39.000

40.000

Then the output projection was specified by inputting the following text into ArcInfo: project cover <DEM file name>

At this point, ArcInfo identified the projection above and prompted for input of the output parameters for the new file. The following text was then input:

OUTPUT

PROJECTION LAMBERT

/*DATUM NAD83

ZUNITS NO

UNITS METERS

SPHEROID GRS1980

XSHIFT 0.0

YSHIFT 0.0

PARAMETERS

38.733333

40.033333

-82.5

38.0

600000

END

This completed the projection of the DEM into the Ohio State Plane Coordinate System. ArcInfo is not needed to project the other shapefiles that will be loaded into ArcView. ArcView has a projector extension that performs this function very easily.

3. Starting ArcView

With ArcView operating, the project file prepro04.apr was opened. Two new dropdown main menus, CRWR-Prepro and CRWR-Utility will now be visible. In addition, some new buttons will also be visible, these new functions are implemented by Avenue Scripts. Clicking on the Scripts icon in the Project window, will reveal the customized scripts used in this project. From the File/Set Working Directory, the working directory was set to /Millcreek. Also under File/Extensions the following were selected: 3D Analyst, Spatial Analyst, CRWR Vector, CRWR Raster, Projector!, and Geoprocessing.
 
 

4. Viewing the DEM

In View1, using the Add Theme Button , Demgridp and Huc.shp were added from the Millcreek directory. For Demgrip, the Data Source Type in the Add Theme Dialog box must be set to Grid Data Source and for Huc.shp it must be set to Feature Source Data.


From the view above, it is apparent that the HUC only covers a small portion of this DEM. Mill Creek is found in the upper right of the HUC so no other DEMs are necessary. It is recommended that DEMs used in CRWR-PrePro not exceed 1,000,000 cells due to the fact that certain procedures can take hours to process. The DEM above contains about 12,960,000 cells and must be reduced in size before moving on. Using CRWR-Raster and the Clip Grid by Polygon function, the necessary portion of the DEM that matched the HUC can be clipped out. With Demgridp the active theme, the Clip by Polygon function uses HUC.shp and performs the function. The new DEM will resemble the intersection between the DEM and polygon.

5. Set Analysis Extent

Before processing the DEM, the Analysis Extent must be set to tell ArcView what part of the view is to be analyzed. From Analysis/Properties, the Analysis Extent at the top of the window to Same as Demgrid, and the Analysis Cell Size in the middle of the window to Same as Demgrid. Setting the analysis extent and keeping it consistent ensures that the various grids generated in the analysis are locationally consistent with one another and their cells are the same size.

6. Preparing the River Reach Files

The River Reach Files must be refined before they can be used in CRWR Pre-Pro. The RF3 files from the Mill Creek HUC and two northern most HUCs must be added to the view. The purpose of adding the additional RF3 files is to ensure the watershed area for the Mill Creek Basin is reduced as much as possible to the actual area. The reaches from the neighboring HUCs must be burned in to the DEM to ensure that any runoff contributing to those reaches runs away from the Mill Creek Basin. In the view below, HUCs 05090203 (blue) and 05080002 (green) are visible. In the view, green stream segments can be seen running away and to Mill Creek. If the segments running away were not considered, the area that contributes runoff to those segments may actually be included in the Mill Creek Watershed. Though the contribution may seem small, the aggregation across the watershed may be significant.

With the RF3 images in view, the Mill Creek segments must be inspected for larger bodies of water that are represented by their banks and not centerlines. The network should also be proofed for breaks in the streams. Shown below is an example of a lake in the upper portion of the basin. If the "Burn Streams" function were to be applied, it would burn in the banks and create two streams in the terrain. In CRWR Pre-Pro, the center-line must be drawn in lieu of the banks.

With the 05090203.shp theme highlighted, the query tool is accessed and the following entered. This highlights the "R" and "S" reaches as shown below so that they can be saved as another shapefile for editing.

Under Theme/Convert to Shapefile a new RF3 theme was created called Cleanrf.shp. This theme is shown below in red on top of the original RF3 file. This is the view that was used to create the centerline of the lake. The original RF3 file was used as a guide to aid in drawing the centerline


With the Cleanrf.shp file active, Theme/Properties was accessed and the Edit function was then accessed. General and interactive snapping were selected and set to .005 miles. Then selecting Theme/Start Editing the Cleanrf.shp file is ready for the stream centerlines to be added in those areas that require it. Using the old RF3 file as a guide the centerline was added as shown below. Once all edits are complete, it is saved by choosing Theme/Stop Edits.
With the Mill Creek RF3 files clean and prepared for Stream Burning, all of the RF3 files must first be merged into one theme. After loading the Geoprocessing Wizard extension, Edit/Geoprocessing was chosen and the three files were merged.

TOP

7. Burn in the Streams

The next step is to perform what is called "Burning Streams." It entails raising the land surface cells that are off the streams by an arbitrary elevation amount so that the streams delineated from the DEM exactly match those in RF3.

With both Cleanrf.shp and Demgrid themes active, CRWR-PrePro/Burn Streams was selected. In the Elevation Rise dialog box, 1000 was chosen as the arbitrary elevation rise. Since this is the temporary grid, it was saved in the Millcreek\tmp directory. This allows one to rebuild the project easily should ArcView crash. Under Theme/Save Data Set, the grid was saved as millburn. Highlighting the Burned_Dem theme and using the Identify tool near the streams, it is evident that the stream cell elevations remain as they were on the original DEM surface but the land surface elevations have been raised 1000m higher.

8. Fill the sinks of DEM

Most of the DEM data are accurate, however, aberrations do occur in the DEM which cause pits to form in the terrain. These pits need to be filled, otherwise they will cause the wrong flow direction. The Fill Sinks function raises pit cell elevations so as to level of the lowest neighboring cell. Only tiny sinks will be filled, since large sinks, such as lakes, are real sinks that we do not want to remove from the DEM.

Under CRWR-Prepro/Fill Sinks, Burned_DEM is chosen from the prompted dialog box as the Input Theme. Burned_Dem is automatically populated in the Input Theme 1 field and millfill is chosen as the Output Theme 1. Once OK is selected, a blue bar ca be seen running across the bottom of the View1 window to indicate that processing is occurring. When it is completed, the new grid millfill will be added to the View window. This process is the most time consuming of all the functions and may take some time to execute on a slower computer. Once the millfill theme is added, the project is saved to preserve the project in case ArcView crashes. This periodic saving should be common practice.

9. Compute the Flow Direction Grid

With the DEM grid filled, the flow direction grid can be calculated using CRWR Pre-Pro/Flow Direction. The Input Theme1 is automatically populated with the FilledDem (millfill). This is because the theme guafill has a theme tag of FilledDem. The theme tag helps ArcView to recognize the intrinsic properties of the grid, no matter what the given name of the theme. All of the current theme tags by choosing menu CRWR-Utility /Display Theme Tags.

For Output Theme 1, millfdr was chosen and given and the Flow Direction Grid was calculated. After a short period, a flow direction grid was added to the View. For a better view of the flow direction Grid, the legend of millfdr was opened and in the Legend Editor, a file titled fdr.avl was applied using the Load button. Shown below, is the new flow direction grid with a 3-D appearance.


10. Compute the Flow Accumulation Grid

From CRWR-Prepro /Flow Accumulation, The Input Theme1 is automatically populated with the flow direction grid, in this case, MCfdr. The Output Theme 1 is entered as Mcfag (Mill Creek Flow Accumulation Grid). After a short period, a flow accumulation grid will be added to the View. Some faint streams can be visible running from the upper right to lower left. The change in color represents the accumulation of grid cells , where the darker the color the more grid cells drain into that particular cell.

Using the Identify tool, the concept of flow accumulation can be seen. By clicking on the most down stream portion and working upstream, the number of cells will decrease. This is directly associated with the change in drainage area that is associated with the number of cells. Another location to look at is the confluence of tributaries. By looking at the upstream portion then the down stream portion, one can see the contribution by the incoming tributary or branch.

11. Define the Basic Stream Network

Before constructing the stream network, the cell threshold or minimum stream drainage area must be designated. From CRWR-Prepro /Stream Definition (Threshold), the Input Theme1 is automatically populated with the flow accumulation grid (Mcfag) and MCstr (Mill Creek stream Grid) was added as the Output Theme 1.

In the prompt dialog box, the stream threshold was changed from the default 10000 to 1000. After a short period, the stream grid MCstr was added to the View. The stream grid has a value of 1 in each cell with a flow accumulation value larger than 1000, and NODATA on all other cells.

When viewing the Merge5.shp theme (red) just below MCstr (blue) as shown below, some of the streams of Merge5.shp are not covered. This is due to the fact that the 1000 cell threshold is too large to identify some of the smaller streams represented by Merge5.shp. Using this threshold, eliminates some of the more insignificant sub-watersheds.


12. Add a stream to the Stream Network

Although we do not want to delineate all of the extra streams described by a Rf3 file, there will be a few streams that need to be kept for watershed delineation. In the view below there is an example of a stream that must be added to MCstr. Segment #2 is a segment identified as significant to the watershed by the Ohio EPA. If the stream grid (MCstr) does not define these streams, you can add the streams to the MCstr grid using the Edit Stream tool.

In order to include these segments, the Edit Stream Tool is clicked then each of the segments on Merge5.shp that is not defined by stream grid MCstr is clicked. This adds a new theme named Addlines.shp with the stream traced from where you clicked is added to the view.


The new stream Addlines.shp needs to be added into the existing grid stream network. Using CRWR-Prepro/Add Streams a prompt dialog box will show with the three Input Themes already populated with the proper grids. The Output Theme name was named MCmodstr (modified stream grid).

Shortly after execution, a Yes/No dialog box appears. Yes is chosen to use traced and threshold streams. The modified stream grid MCmodstr includes both the original stream with a cell threshold 1000 and the traced stream added.

13. Segment Streams into Stream Links

The Stream Link function gives each stream segment a unique ID. From CRWR-Prepro/Stream Segmentation (Links), a pop up dialog box shows and ModifiedStreamGrid is chosen. The two Input Themes are automatically populated with the flow direction grid (MCfdr) and the modified stream grid (MCmodstr). The output grid name as MClnk. Shortly after execution, the stream link grid MClnk will be added in the view. To better view the stream link grid, the Legend Editor is accessed, Unique Value is chosen for Legend Type and Value for Values Field. For the color scheme, Fruit & Vegetables is chosen and applied. This will show the stream grid segmented to stream links, with each link having its own unique color and value as shown in the legend bar.


14. Find Link Outlets

An outlet cell is the cell in each link that has the largest flow accumulation value. All of the cells upstream of the outlet cell flow into the outlet cell. From CRWR-Prepro/Outlets from Links LinkGrid is chosen to create the outlet grids. The two Input Themes are automatically populated with the flow accumulation grid (MCfag) and stream links grid (MClnk) and the output grid name is designated as MCout. After execution, the resulting grid is a scattered set of single cells, each the farthest downstream cell of a stream link.

For a better view of the outlets grid, the Legend Editor is accessed. The first color square is double-clicked and the color black is chosen in the Color Palette. Then on the Legend Editor the last color square above the No Data square is double-clicked and the color black is once again chosen in the Color Palette. Then the Color Ramp button is clicked resulting in all of the symbols becoming black. A view of what the outlets look like will be shown below in the next section.

15. Creating a Point Coverage of Flow Gaging Locations.

Gage locations are necessary in HEC-HMS to calibrate the flows from design storms. In the Basins Core Data, there is a point coverage of gage stations with attributes. This theme should be loaded, viewed, and reviewed. Using the gage numbers, the USGS website was accessed to verify if the gages were still active or if any others have been added or moved. It was found that many of the gages were not active or moved while the rest did not pertain to the Mill Creek Watershed. All the current gages were downloaded and the required data was entered into an Excel Spreadsheet and saved as gages.dbf file. In the spreadsheet, several fields were entered: First were Shape, then No_, and Name. . Followed by the fields Lo_d, Lo_m and Lo_s that correspond to the longitude degrees, minutes and seconds, while fields La_d, La_m and La_s correspond to the latitude degrees, minutes and seconds. Next were Longitude and Latitude in decimal degrees. Then Meanannual was entered for the mean annual flow at the station in cfs. The last entry was Station_ which contained the station number. Note that Longitude = - Lo_d - Lo_m/60 - Lo_s/3600 (negative signs because is West longitude) and Latitude = La_d + La_m/60 + La_s/3600.

With the file created and placed in the working directory, it was added using the Tables icon of the Project window. Once opened, the table will look as shown below.

In order to create a point coverage, a new View window must be opened. Then by clicking View/Add Event Theme, Gages.dbf was selected in the Table slot, Longitude in the X Field slot and Latitude in the Y Field slot. After clicking OK, a point theme was obtained called Gages.dbf. Before adding this to the working project, it theme must be put into the correct projection. First the view map units must be defined. Define the map units of View2 by clicking on View/Properties and selecting decimal degrees in the Map Units slot of the View Properties window in the View2.

With the map units set, the theme needs to be added to the view and Projected the from Geographic into Ohio State Plane Projection by clicking on CRWR-Vector/Project. Meters was then selected in the Output Units window. Next State Plane was selected in the Category slot, and Ohio South in the Type slot of the Projection Properties window. When prompted for the name of the projected coverage, it was called GagesPro.shp and added the View1 as shown below in purple. Also shown here in red are the outlets defined by CRWR-PrePro.


In order to add these gages as outlets, zoom into the Mcmodstr theme until you can see the individual cells. Using the add outlets tool, click on the stream segment nearest the gage location. Ensure that when selecting a location, it is not one of the defined outlets not more than 2 cells from an outlet. Once an outlet is added, a new point theme called Addasoutlets.shp added to the view as shown in green.

16.  Manually Defining Watershed Outlets.

Watershed delineation can be executed with the data compiled up to this point. PrePro will use all of the outlets found in MCout and generate one watershed per outlet. In order to keep the project consistent with the Louisville District, outlets must be manually defined in order to generate the same watersheds. To do this, the outlet locations must be given. Below is a digitized version of the watershed that were delineated by hand at the Louisville Corps of Engineer District.
Using this as a template, outlets were manually added in an effort to replicate the watersheds. The outlets that were selected were added to addasoutlets.shp with the gage locations. The following is a depiction of the outlets that were added on to the Mcmodstr theme. The breaks in the stream really do not exist, they are caused when adding the photo to the document.


When all the proposed outlets are added, CRWR-Prepro/Add Outlets was selected. ModifiedStreamGrid was chosen for outlets processing. In the prompted dialog box, five Input Themes are automatically populated. MCmod_out was designated as the modifiedoutletsgrid, and MCmod_lnk for the modifiedlinksgrid.

A Yes/No dialog box appears, and No was chosen to use the manually selected outlets and also outlets. The modified outlets grid MCmod_out and stream links grid MCmod_lnk are added to the view.
 
 

17. Delineate the watersheds

With the links and outlets finalized, the watersheds can be delineated.

From menu CRWR-Prepro / Sub-Watershed Delineation, . ModifiedOutletsGrid is chosen from the prompted dialog box. In the next dialog box, the two Input Themes are automatically populated with flow direction grid MCfdr and modified outlets grid MC_out. MCwshd is the name given to the output watershed grid.

Once clicking OK, the watershed grid MCwshd is added to the view. A sub-watershed is a zone of cells with the same cell value as the first outlet cell they drain through.

With Mcmodstr added to the top layer in the legend, the correlation of the streams to the watershed can be seen. In order to view the watershed in a more contrasting view, the Legend Editor is accessed and Unique Value is chosen for Legend Type, Value for Values Field. The default Color Schemes Bountiful Harvest. is applied. Notice that each of the stream segment has a watershed associated with it. Drag the MC_out theme to the top of the legend bar. You'll see an outlet at the end of each stream segment. Using this process, oulets had to be manually adjusted by editing addasoutlets.shp until the desired subwaterheds were achieved as found in the figure above submitted by the Louisville District.













                TOP

18 Vectorize the Stream and Watershed Grids
 
 

To this point, grids have been the primary data worked with. Grids are excellent for cell based analysis, however, vector data are easier to use and store. From the menu CRWR-Prepro /VectorizeStreams and Watersheds is chosen. In the prompted dialog box, the Input theme has automatically been populated with the watershed grid MCwshd and MCwshply is the name given as the output theme name.

In the Vectorize Streams dialog box, ModifiedlinksGrid is selected and the two input themes will have already been populated. MCrvr is selected as the Output Theme name. A Yes/No dialog box for backing up the MCwshply.shp will appear and no should be chosen. Then a prompt indicating that a certain number of dangling polygons have been merged. Due to the fact that a vector polygon does not necessarily have a square shaped border like the grid, the conversion of the grid to a polygon will create a dangling polygon may on the edge of an existing polygon. This dangling polygon is a tiny watershed that does not exist, and will be dissolved into the parent watershed polygon to which it belongs. The blue running status bar will stay at 100% for quite some time before the process is complete. This should not be taken as a problem with the program. Once complete, the watershed polygon MCwshply.shp and the river line MCrvr.shp are added to the view. Shown Below is the vectorized watershed with its stream segments.

Once the outlets are refined and appear to be the same as those used by the Louisville District then the process can continue. After a series of delineations and outlet adjustments the following was the result. On the left is the Louisville watershed and on the right is the final watershed developed in ArcView. Using Field/Statistics, the difference in area for the two watershed was less than 2 square miles.
                TOP


SOILS DATA

                    19.  Working with STATSGO data

In a new View and the theme statsgo.shp was added which is in the core data folder in the working directory. The view reveals a large number of Statsgo polygons which are needed to describe the soils of the HUC. Highlighting statsgo.shp and using the legend editor, the field Muid is used to label the mapunits. Muid is the mapunit identification number for each unique combination soil components depicted in Statsgo.

From the Tables icon in the Project window, the mapunit.dbf and comp.dbf tables are added from the working directory. Each of these tables has a number of fields that are not necessary for the project. With the mapunit.dbf table highlighted, Table/Properties is used to click off all of the check boxes except for Muid and Muname In the Comp.dbf table, all fields were clicked off except for Muid, Seqnum, Compname, Comppct, Slopel, Slopeh, Surftex, and Hydgrp. The percentage of each hydrologic soil group in each map unit is necessary for the calculation of the curve number theme. Using CRWR PrePro/Soil Group Percentages, along with the required responses for the map unit (mapunit.dbf) and component tables (comp.dbf) when asked, a table called muidjoin.dbf will be created.

Using the Geoprocessing Wizard extension found in View, the statsgo.shp file was clipped as found below using the Mcwshply.

                TOP



 

LAND USE AND LAND COVER DATA

20. Working with Land Cover and Land Use Data

Land use/land cover files use the Anderson Land Use Code classification system, in which major land use types are broken out into 9 categories as found below:

1 = urban

2 = agriculture

3 = rangeland

4 = forest

5 = water

6 = wetlands

7 = barren land

8 = tundra

9 = ice and snow.

The second describes subcategories of these principal categories, e.g.

11 = urban residential

12 = urban commercial

13 = urban industrial, etc.

This land use classification of the United States was developed in the late 1970's and land use has changed since then particularly in and around cities. Regardless, the LU/LC files are still the standard land use classification of the United States.

With the theme lulc.shp added to the view, the individual land uses are seen by accessing the Legend Editor and changing the Legend Type to Graduated Color and the Classification Field to lulc_code. On the Classify button the number of classes was changed to 8. The Value buttons were used to change the values of each category and the Label buttons were used to change the labels of each category. The Fill Palette was then used to color-code each of the symbols into land use categories as seen below. The Geoprocessing Wizard was used to clip the theme to the Mill Creek Basin as seen below.

Land Use and Land Cover

                TOP

CURVE NUMBER DEVELOPMENT AND SOIL STORAGE
 

21. Curve Number Calculation

By combining the land use and soils data, a curve number grid can be generated. Soil Conservation Service (SCS) curve numbers are parameters for calculating abstractions. With the soil theme and land use theme active, CRWR PrePro/Curve Number Grid is selected. When prompted for the name of the lookup table, rcn.txt is selected. It may be necessary to add this table to the working directory. It can be found on the GIS Hydro ’99 website under Watershed Characterization. When prompted for the name of table with soil group percentages, muidjoin.dbf is selected. The name given to the theme is Curvenumber. Below is a depiction of the Curve Number Grid that is calculated after various color schemes are selected.



 
 
 
 
 
 
 
 
 
 
 
 
 

22. Soil Water Storage

Using the Curve Number coverage found above, the SCS equation for Soil Water Storage (S) is applied. The equation is found below. Once the soil water storage coverage is found, the SCS equation for initial abstractions is applied and the theme statistics accessed to find the mean across the watershed.

SCS Soil Storage (in)


The mean was converted to (mm) which is the required units for HEC-HMS. A value of 6 mm was found to be the mean across the basin. This differs from the 12.7 mm used by the Louisville District.
TOP

DEVELOPING A HYDROLOGIC MODEL

23. Calculating Hydraulic Attributes

To calculate the attributes, CRWR-PrePro/DEM Based Parameters is used. For the choose an Outlets Grid window, ModifiedOutletsGrid is selected and in the DEM Pre-processing: Calc. Watershed and Stream Parameters window, the blanks are filled with the following theme names: Demgridp, MCfdr, MCout, MCWsh, MCwshply.shp, MCrvr.shp and LFP. In the first Calculation Method window, SCS is selected for the abstractions method, and in the second Calculation Method window, SCS is selected for the lag-time method. Yes is selected in both Input Warning windows. The Curve Number theme is selected for the Curve Number Grid.

In the Time-Step window, 1-30 minutes is selected and in the Time-Step [minutes] window, 30 is selected. For Select Stream Input File (Grid_code, ... window, StreamP.txt is selected from the working directory. This table can be found on the GIS Hydro ’99 Website under Watershed Characterization. This table will require modification for each project. This particular one was developed for the Guadalupe basin in Texas. The main field that must be adjusted is Grid-code. Using the vectorized stream theme for the water shed, accessing the attributes table will allow one to view the Grid-code numbers assigned. These numbers must be the same ones used for the StreamP.txt. A sample table is shown below with the stream parameters: flow velocity = 1 m/s and Muskingum X = 0.2.

This process has added fields to the attribute tables of MCwshply and MCrvr and populated them with values that represent the hydrologic parameters of the feature elements.

The fields that are added to the MCwshply table are: LngFlwPth (length of longest flow-path), Slope (slope of longest flow-path), Baseflow (baseflow method), Transform (unit hydrograph model), LossRate (loss rate method), CurveNum (average curve number), InitLoss (initial loss for the initial plus constant rate loss method), CLossRate (constant rate loss for the initial plus constant rate loss method), WVel (average velocity of longest flow-path), and LagTime (lag time for the SCS unit hydrograph). The fields that are added to the MCrvr table are: StreamVel (flow velocity), MuskX (Muskingum X), Route (flow routing method), MuskK (Muskingum K), NumReachN (number of subreaches for Muskingum flow routing method), LagTime (lag time for pure lag flow routing method).
 
 

24. Determining the schematic and writing a basin file for HMS

The HMS schematic is a conceptual model that captures the connectivity between the different elements of the hydrologic system. The HMS basin file is an ASCII file, readable by HMS, that includes all the information stored in the schematic.

To transfer the hydrologic attributes calculated in Step 20 to the schematic and basin file, it is necessary to add to the project the following tables from the working directory: hecsub.dbf, hecjunct.dbf, hecreach.dbf, hecres.dbf, hecsink.dbf, hecsource.dbf and hecdiv.dbf. This is done in the project window under Tables/Add.

To determine the schematic and write the basin file for HMS, the AddAsOutlets.shp, MCwshply.shp and MCrvr.shp themes must be active prior to using CRWR-PrePro/HMS Schematic. In the HECPREPRO window fill the six blanks with the following strings or values: yes, default, 23, 2, MCBasinSCSv1 and Mill Creek BasinSCSv1. The Hydrol#.shp and Hydrop#.shp, as well as Syml#.shp and Symp#.shp store all the relevant information of the hydrologic system necessary for HMS. Additionally, Syml#.shp and Symp#.shp are the stick diagrams of the system. A text file called MCbasinSCSv1.basin has also been created in the process, which is the basin file for HMS. Below is the view that will be generated in the process.

                TOP


 

HYDROLOGIC MODELING SYSTEM (HEC-HMS)

25. Creating the HMS components

When HMS 1.1 is opened a HMS Project Definition window will appear. A new HMS project is created using File/New Project. In the HMS * NEW PROJECT window that appears a name is entered in the Project slot such as Mill Creek Basin.

To import the MCbasinSCSv1.basin file created in GIS, Edit/Basin Model/Import is accessed. In the HMS Basin Model * IMPORT BASIN MODEL window, basin file is selected from the working directory under tmp. The file description Mill Creek BasinSCSv1 will appear and not the file name.

With the basin Model opened, some editing is required before moving on. Earlier in the project lakes were taken out of the river reach file and replaced by centerlines. These lakes were manmade and created by Dams. In HMS dams are replicated by reservoirs. This is made easy in HMS by simply adding the new element. In this case, outlets were used where the dams were located earlier in the project so they only need to be replaced. Outlets become junctions in HMS. By finding the junction that corresponds to the dam location, it is first replaced using Edit/Delete Element(s). Then using Edit/New Element/Reservoir one is added to the view. The new element must be moved into place then connected to its respective neighboring elements. To do this, the up stream basin is connected by right-clicking and selecting Connect Down Stream. Using the cross hair, the reservoir is selected and thus connected. Then the same process is used to connect the reservoir to the downstream reach. This process is done for West Fork Mill Creek Lake (Reservoir-1) and Shannon Lake (Reservoir-2 as shown below. By opening the Reservoir icon, parameters must be entered for its use. In both reservoirs, a storage/discharge relationship was used with a given initial discharge. The values used in the model were provided by the Louisville District. The other parameter that must be added to the basin file is the initial losses for each basin. By double-clicking on each basin the initial losses are manually entered. The value of 6mm was used on each basin and it was calculated using the soil storage grid developed above.

26. Creating a Precipitation Model

To create a precipitation model, click on Edit/Precipitation Model/New, and then write Thiessen# in the Precipitation Model and in the Description slots and click OK. In the HMS PRECIPITATION MODEL - METHOD SELECT window User Specified Gage Weighting is selected.

With this option, you specify weighting factors (Thiessen-type) to be applied to gaged precipitation to calculate spatially-averaged precipitation for subbasins. The USER-SPECIFIED GAGE WEIGHTS screen contains a "notebook" with three sections as shown below. The first, labeled Gages, provides for specification of a gage ID, gage type, total storm depth, and index precipitation for each precipitation gage (both recording and non-recording). The second section, Subbasins, provides for the addition of subbasins to the precipitation model, and allows for specification of index precipitation for each subbasin. The optional specification of index precipitation for subbasins and precipitation gages enables adjustment for bias in gage-precipitation values. The third section, Weights, specifies both the total-storm weight and temporal-distribution weight for each gage.



 
 
 
 
 
 
 
 
 
 
 

The first tab is Gages. At this tab you specify the gages that are to be used for the design storm. Two gage types can be used with the method which are recording and non-recording gages.

Gages for total storm precipitation (i.e., non-recording gages) can be entered in the Gages section by using Edit/Gage Data/Precipitation and selecting the Add Total-Storm (NR) Gage button and entering a gage name and appropriate storm depth. Non-Recording gages entered in this way can only be accessed by the current precipitation model. Recording gage ID’s are obtained by selecting them from the GAGE SELECTION LIST, which is accessed by clicking the Gage Select button. A total-storm depth is not normally entered for recording gages, but can optionally be added. When total-storm depth is specified for recording gages, the individual precipitation values will be scaled so that the total-storm rainfall will equal the specified amount. Recording precipitation gage data must have been preciously entered before it can be referenced from the Gage Select button. The gage type for both recording and non-recording gages is entered automatically. Using Edit/Gage Data/Precipitation recording gage data for an actual storm can be entered. The storm used for this project can be found in the Control Specifications section. The gage data for this design storm were found on a web site for The Ohio Advanced Warning Flood System. Using the archive, the incremental depths of rainfall were found for the design storm. Once the values are entered, the gages are stored for use in other projects as well. They are simply added to the precipitation model as discussed above. Below are the incremental hyetographs for the storm that are produced in HMS.













In the Subbasins section the subbasin names are entered and the number of gages associated with each subbasin is shown. Subbasins are added from the basin model by clicking the Add button. The addition of subbasins from a basin model must be done before the subbasins can be referenced in the Weights section. Index precipitation (e.g., normal annual) amounts can optionally be entered for both precipitation gages and subbasins. If such data is entered, it will be used to apply a bias adjustment to gage precipitation.

The Weights section displays information for a single subbasin at a time. A droplist allows selection of a subbasin that has previously been specified in the Subbasins section. Gage ID, type and associated weight are entered in the table. When a gage ID field is selected, a droplist appears that shows the gages previously entered in the Gages section. The gage type is either "R" for recording, or "NR" for non-recording, and is entered automatically. The Total Storm Gage Weight is applicable to both recording and non-recording gages, whereas the Temporal Distribution Gage Weight applies only to recording gages. The weights of each type are normalized to sum to 1 if the entered values do not do so.

Weights were assigned using the Thiessen diagram below. It was created in GIS using CRWR PrePro/Precipitation Gage Weights. Before this could be done a point coverage had to be created similar to the section on flow gages. For Mill Creek the two Precipitation gages above were added to the view. Then using the script, a Thiessen Diagram found below was created. This diagram simply shows the sub basins with the two gage locations and the mid point between the two points. This line is drawn perpendicular to a line that would connect the two points. This allows for weighting given to the sub basins for each gage. For the most part, most basins were weighted with a 1 while the three basins that contain a portion of the line are weighted according to the percentage of the basin that is bisected by the line. These values are entered in the weights section.

                TOP

27. Create Control Specifications.

To create the control specifications, the third component of HMS, click on Edit/Control Specifications/New in the HMS * PROJECT DEFINITION window. Write Control in the Control Specs ID and in the Description slots of the HMS * NEW CONTROL SPECIFICATIONS window and click OK. In the HMS CONTROL SPECIFICATIONS *SETUP window fill the blanks as shown in the figure below.

28. Running a Design Storm for Calibration

In order to calibrate the derived model, a USGS gage is necessary. For Mill Creek, a gage at Carthage (junction 35) was used. The time series data for the gage provided by Tim Raines. This data was entered using Edit/Gage Data/Discharge. In the NEW GAGE RECORD screen, the name Carthage was given for the gage. A description can be entered for the gage, but is optional. Selection of the data type and units is required when the data will be entered manually. Latitude and longitude data is not required. Select whether the data will be referenced in an external DSS file, or entered manually. Once the data is entered for the gage, the graph can be attached to its correct location in the basin file. At junction 35, the right-click was used and observed flow was selected. In the next window, the Carthage Gage is selected. When HMS is run and the results at Junction are viewed, the observed hydrograph will be present for comparison.

To run the design storm for model calibration the basin model is accessed from the HMS - SCHEMATIC window. First Simulate/Run Configuration must be selected. Then the components created above must be highlighted and the Add button selected. In the HMS-SCHEMATIC window, clicking on Simulate/ Compute <Run 1> will begin the simulation. Once the run is complete the HMS Compute window must be closed unless there is an error in which case you need to view the run log. To see the hydrograph at any junction, watershed or stream, click on the arrow located on the upper-left corner of the HMS-Schematic window and right-click on the element you are interested in. You will get a pop up menu from which you should choose View Results/Graph. The following figure is the hydrograph for the Carthage Gage.

In order to calibrate this model, parameters must be identified and isolated. From the initial basin model, the StreamP.txt file designated initial estimated values for the velocity and Muskingum X which were 1 m/s and 0.2 respectively. In looking at the gage hydrograph it is evident from the curves that the observed (red) flow and the model (blue) flow are not similar. Both curves indicate a steep climb to the peak on the approach of the flood wave but the peaks and time of peak are off a great deal.

In looking at the parameters that could be adjusted in calibration, velocity and Muskingum X were chosen first due to the fact that they were arbitrarily chosen. Muskingum X is a storage constant and has a range of 0-0.5. It is well known that changing this value slightly has little to no effect on the flows. The value of 0.2 is pretty common in natural streams. The only other parameter is the velocity. Using the StreamP.txt file and manually changing the velocities to 0.5 m/s and 1.5 m/s two more basin files were created using CRWR PrePro. It is evident from the hydrograph below that the desired endstate is not achieved. Here at V = 0.5 m/s, the peak flow drops just below the desired level but the time of peak moves to the right which in contrary to the desired effect. It is also apparent that the shape of the curve is not consistent with the observed data. The rising portion is characteristic, but the peak and falling portion are not characteristic.

In the second hydrograph, once again the desired endstate is not achieved as well. Here using V = 1.5 m/s the peak is almost 150 cms higher than the observed but the time of peak has moved to the left in the desired direction.

                TOP


CONCLUSIONS

GIS has proven to be a great tool in watershed characterization. CRWR PrePro worked well to create a basin file using a myriad of data to do so. The basin file that is created works well in HEC-HMS. Issues do arise when making generalizations about parameters within the watershed. Problems in the generalizations can be seen in the routing method used by PrePro. For example, when the Longest Flow path is divided by the stream velocity two routing methods are possible. These routing methods determine the time of travel through the reach. If the resulting value is less than the time step used (30 min) the pure lag equation (tlag) is used. This equation is typically used in shorter reaches and the units for the lag time is minutes.

When the resulting value is greater than the time step, the Muskingum method (K) is used. The units of this equation are in hours. In looking at the two equations it is readily apparent what variable has the greatest effect on the outcome. The longest flow path was calculated in PrePro and does not change. The velocity, which is in the denominator, will vary and as result have major impacts on routing times.

The Mill Creek Basin is contains a number of oddities that make it difficult to model. Below are some of the characteristics that contribute to its complexity.

Combined Sewer Overflow Outfall

                     (L)Concrete Channel with concrete sides on the left and rip rap on the right.     (R) Condcrete channel preceded by an improved natural channel.
 
 


Concrete Channel with Deposition

The green triangles represent approximately 5 Combined Sewer (CSO) outfalls per location. The effects of these are difficult to understand. The question one must ask is what is the drainage area of each outfall? It is possible that the drainage areas actually encompass an area larger than what the model would depict due to the fact that water was being pulled via sewer system from another watershed. Another question that it creates is what happens to the basin lag time as a result of the sewer system? Does the lag time decrease as a result of runoff moving through pipes or does it actually slow down due to an intricate sewer system that backs up at time due to large flows. In either case, the dynamics created by the CSOs alone lend to uncertainty in discharges and basin lag times. Another variable in Mill Creek is the work that was completed some years back when the project was originally begun in the 1970’s. Shown in purple bars, are locations where channel improvements were conducted. In some cases, these improvements consisted of concrete lined channels with concrete or rip-rap walls. In some locations, a natural channel was improved. In either case, these improvements have major impacts on stream velocities in these areas.

Below is an example of the effects of channelization on velocity. In the table, arbitrary values were selected for the slope and channel dimensions. When using Manning"s equation, the Roughness Coefficient (n) varies by channel texture. Typical values for the roughness coefficient can been seen in the table’s left column for different textures described in the right column. Applying the equation below, typical velocities are found as in the center column.

Velocities vary from 1.4 f/s to 6.0 f/s depending on the channel type. This change in velocities is indicative of the range within the Mill Creek Basin as the flood wave passes through the different sections that are natural or altered. Below the table is a graph that depicts how velocity is affected by the different channel properties.

Going back to the equations used to determine routing method and travel times, the actual variable velocities will adjust these times significantly in many cases. This should allow for a more accurate model and bring the observed flows much closer to the model flows. This will require a closer look at each reach and applying Manning’s equation with data provided by the Louisville District. Mush of this data exists already in the form of HEC-RAS cross sections already provided.

On other aspect to look at is the curve numbers for the basin. The land use and land cover data is old and does no reflect the changes that have occurred as a result of urban sprawl. In reviewing the LU/LC coverage below, the obvious presence of the urban environment in red will not change much in time. The agricultural area in yellow is Butler County that have undergone a great deal of urban growth over the years. As a result the curve number will actually be higher than calculated in this report. That has an affect on the basin lag time for which the equation is shown below. An increase in the CN will cause the basin lag time to drop thus having an affect on basin time of peak discharges.

In conclusion, all of these factors have an impact on the modeling of the Mill Creek Basin. GIS and the CRWR PrePro scripts do an outstanding job of characterizing the watershed and writing a basin file for HEC-HMS. This research reveals how the human dimension of modeling is necessary to bring everything together and make the model useful.
TOP

FUTURE WORK

Work is ongoing to calibrate the Mill Creek model in HMS. This will be accomplished by adjusting reach velocities by accounting for channel properties using Manning’s equation. This work should refine the routing method times and bring the model within an acceptable range of the observed discharge and time of peak at the Carthage gaging station. If the desired results are no satisfactory, updated LU/LC will be used to adjust the curve number and basin lag times for Butler County. In conjunction with the above refinements, the current data will be imported into HEC-RAS for evaluation and then using AVRAS and DOQQs achieve a visual representation of the flood damage. The ultimate objective is to establish relationship between watershed storage and how its extent can be used to deter mine types of control measure for flood control.

TOP


DATA DICTIONARY
 
 








TOP


REFERENCES Chow, V.T., Maidment, David R. and Mays, L.W. / Applied Hydrology. / New York, 1998.

Hydrologic Engineering Center. / HEC-HMS Hydrologic Modeling System, Program User's Manual. / U.S. Army Corps of Engineers, Davis, California, 1998.

Maidment, David R. / Handbook of hydrology. / New York, 1993.

Olivera, F., Reed, S. and Maidment, D. / HEC-PrePro v. 2.0: An ArcView PreProcessor for HEC's Hydrologic Modeling System. / Center for Research in Water Resources, Austin, Texas, 1998.

Azagra, Esteban./ Rainfall Runoff in the Guadalupe River Basin / Center for Research in Water Resources, Austin, Texas, 1998.

Azagra, Esteban / Regional Hydraulic Model for the City of Austin / Center for Research in Water Resources, Austin, Texas, 1999.

Maidment, David, R. and Ahrens, Seth / Introduction to HEC-HMS / Center for Research in Water Resources, Austin, Texas, 1999.

Olivera, F and Maidment, D / Developing a Hydrologic Model of the Guadalupe Basin / Center for Research in Water Resources, Austin, Texas, 1999.

TOP