Erosion Patterns in Austin Streams

Cindy Lancaster

GIS Term Project -- Fall 2002

Introduction

In setting out to do a project on Erosion Patterns in Austin Streams, my original intent was to attempt to establish a quantitative link between changes in land use and/or stream flows and the erosion that has occurred in Austin streams.    I had various ideas on how to do that, but for each idea it turned out that either the data was insufficient and/or my approach was inappropriate and/or my knowledge of statistics was inadequate.  As background, I reviewed some of the current literature regarding erosion in urban streams, with special attention to efforts that have been made to establish mathematical relationships between channel geometry and land use changes.  I pulled gage data for streams in Austin and made some attempts to look for trends in streamflow patterns.  In pursuing information on erosion in Austin, I discovered that a very extensive assessment of erosion in streams had been done for the City in 1997.  Although other urban areas have done such assessments for particular watersheds, this may have been the first city-wide study of this type.  Termed a "Watershed Erosion Assessment", the study was commissioned in 1997 and covers 17 watersheds lying at least partly within the Austin city limits.

The need for such a study was evident from the serious property damage caused by streambank erosion and the large expenditures required to address erosion problems in the city's streams.  Another concern is the water quality degradation that results from the sediment loads contributed by streambank erosion.  There has long been a recognition that urban development greatly accelerates stream channel erosion, but few have attempted to quantify this relationship.  In the erosion assessment done for Austin, graphical curves specific to the City's watersheds were developed for relating channel enlargement to percent impervious cover.   Additionally, these curves were used to predict future channel enlargement based on projected development in the year 2040.

Evidence of Erosion Problems in Austin Streams

                            Streambank erosion endangering a home                                                                        Some failed attempts to control streambank erosion

The Austin Erosion Assessment produced a massive amount of information, with a separate report produced for each of the 17 watersheds.  I deemed the study important enough that I eventually decided to re-develop my project objectives around it.  My revised project objectives are as follows:

• examine the study to understand how the quantitative relationships were developed

• input the key results for each stream reach into a geodatabase

• draw conclusions as to the advantages/disadvantages/pitfalls of using a geodatabase for a study such as this

Background

There are some key concepts necessary for understanding erosion processes in streams and the data produced by the Austin erosion assessment.  The special area of study of how the physical characteristics of a river or stream interact with the climate, geology, topography, vegetation and land use of the watershed is called fluvial geomorphology.  The most important process in defining channel form is the bankfull discharge, occurring roughly every 1.5 years.  This level of flow transports the majority of a stream's sediment load over time and thus defines the channel geometry.  While much of the effort in management of urban streams has been directed towards reducing peak flows for storms with a recurrence interval of 10 to 100 years, the minor system flows with return periods of 1 to 5 years are of more importance for understanding and addressing erosion in streams. 

A stable stream maintains a dynamic equilibrium with its flow regime, the pattern of streamflows it experiences over a long period of time.  Land use changes such as urbanization or logging change the rainfall/runoff relationships in the watershed.   In fact, research indicates the impact of urbanization on the flow regime is greatest for the minor system flows -- the flows most critical in defining channel form.  These changes in the flow regime cause the stream to become destabilized.  At that point, the stream's channel geometry will begin to adjust to accommodate the new pattern of streamflows, typically by enlargement of the cross-sectional area through downcutting (erosion of the bed) or widening (streambank erosion) or a combination of both. 

In order to study erosion processes in streams, it is first necessary to classify the streams based on their morphology.  In recent years, the Rosgen stream classification system has become popular because of its usefulness in stream restoration (A Classification of Natural Rivers, 1994).  Using such a system allows one to develop specific quantitative relationships for a given stream type.   In North Carolina, for example, the Rosgen classification system was applied to streams in each of three topographic regions and, for most classes of streams in each region, quantitative equations were developed for relating channel geometry to watershed area.  These equations are then used as tools in applying natural channel design in developed watersheds.

City of Austin Erosion Assessment

In the City of Austin study, a more simplified method of stream classification was used.  Streams were classified as either alluvial, rockbed or rock-controlled.  Alluvial streams have a bottom and banks that are alluvial.  Rockbed streams have a rock bottom and possibly one side rock.   Rock-controlled streams are rock on the bottom and both sides.  All stream reaches included in the City study were classified as one of these stream classes or some combination of two classes.  Besides classifiying the stream type, stream reaches were also examined as to the stability of the stream and were assigned a stability index.  A range of 0 to 0.2 indicates the stream is stable, 0.2 to 0.4 indicates the stream is in transition (i.e. stressed but not yet destabilized) and greater than 0.4 indicates a stream that is in adjustment.

For some stream reaches where the stream had restabilized after urbanization, a ratio of enlargement was computed by comparing the existing cross-section to the historical cross-section (available from old sewer-line plans).  For each of the 3 stream classes, curves were developed for ratio of enlargement as a function of the percent impervious area in the watershed.  From these 3 curves, a current and future ratio of enlargement was estimated for each stream reach.

                                                  Example of Past Channel Enlargement

                                                                        Source: Raymond Chan & Associates, 1997

From the City's point of view, an important aspect of the erosion assessment was to rank existing erosion problems in order to prioritize capital expenditures.  Type 1 sites are those where a home, business or road is currently threatened by channel bank erosion.  Type 2 sites are those where other resources, such as walls, fences and utility lines, are threatened by erosion.  Type 3 sites are those where there is no current threat, but there is likely to be one in the future.  A weighted average of problems was developed for each reach.  The Phase 1 Watersheds Rank Score was included in the data I input into the GIS database.

A summary of the erosion assessment study is included in the City of Austin Watershed Protection Master Plan. 

GIS Data Input

It seemed that the easiest way to incorporate the results of the Erosion Assessment into a geodatabase is to use a Hydro Line Event table.  This turned out to be a great tool for this type of effort.  However, there were some hurdles and pitfalls.

First of all, Route Events have to be referenced to some set of drainage network reaches that have unique identifiers.  The most obvious approach seemed to be using EPA river reach codes since this is a nationally standardized way of identifying reaches of a drainage network and is a part of the National Hydrology Dataset files that can be readily downloaded from USGS.  I did choose to utilize the NHD as a basemap because I wanted to tie the Erosion Assessment results to a standard drainage network dataset.  There was also a City of Austin drainage network GIS file available on the web, but this network was much more detailed and refined than was necessary to show the erosion assessment data.  A bigger problem with that dataset was that the drainage network was comprised of very small segmented polylines that would have needed to be aggregated.  A reach from one stream confluence to the next is a single feature in the NHD dataset, but an equivalent reach could be as many as 10-12 segments in the Austin dataset.

One problem with using the NHD is that there was some variance between the stream reaches included in the NHD and the stream reaches used in the City study.  For example, there is one tributary to Walnut Creek included in the NHD that was not included in the Austin study.  On the other hand, there were several reaches studied for erosion assessment that were not included in the NHD.  In those cases, I tried copying the reaches from the City's GIS drainage network file into the NHD and then faking in a reachcode to use for referencing.  This worked OK, but it was tedious because the City network is comprised of so many small segments.  The copied reaches then look different from the polylines in the NHD because they are much more squiggly due to having been digitized at a different scale.  Most importantly, I don't consider it wise to add identifiers in that way to something that is nationally standardized.  I would assume that the City could work with EPA and/or USGS to add reach code identifiers for certain reaches if there was a desire to develop a network for the City that is incorporated into the NHD.

The reaches in the Austin study were divided based on similar morphology and were referenced to city streets (i.e. number of feet upstream or downstream of a street).  The NHD reaches run from one confluence to an upstream confluence or the head of the stream.  The head of the NHD reach was typically further upstream than the City's study extended.  I used relative addressing to reference the COA reaches to the NHD.  Relative addressing turns out to be a wonderful option because there is so much variance in the length measure of the stream reaches that trying to use absolute addressing would be a nightmare.  For example, a reach that the city measured as 2600 feet has a shape length of 3108 feet in the NHD and by FEMA stationing has a length of 4440.  Without the option of using relative addressing, I don't think it would have been feasible to relate the erosion assessment data to the NHD network. 

Once I verified that the two sets of stream reaches could be related in some reasonable way, I set up a table in Microsoft Excel in a format suitable for route line events. As an identifier for the City of Austin reaches, I used the unique ID already assigned in the study.  The next column contained the reachcode from the NHD.  The following two columns were the "from" and "to" measures.  It was it was necessary to determine what fraction of one or more NHD reaches comprise an Austin reach.   The simplest case would be from 0 to 100 -- meaning all of that NHD was on the given Austin reach.  More typically an Austin reach comprised some fraction of one or more NHD reaches.  In additional columns I listed the Stability Index, Ratio of Enlargement for Current Land Use, Ultimate Ratio of Enlargement for Future Land Use, Sediment Yield in Tons per Linear Foot and and Phase 1 Watersheds Rank Score for each Austin reach.  Some reaches required more than one line if they were on more than one NHD reach.  After getting the table set up, it was then saved as a dBase IV file.  From ArcCatalog, I then created a feature class from the dBASE table and imported that into the feature dataset.

It turned out to be very laborious to match up the reaches and I did not accomplish my original goal of putting all of the City watersheds into the geodatabase.  I used the Little Walnut Creek watershed as the original "pilot" effort for getting the table set up and the process worked out.  Later I added the Walnut Creek watershed data into the Hydro Line Event table.  That is one of the larger watersheds in the City and it took about 3-4 hours to match up the stream reaches and input the data.  By that point, I felt I'd learned what I wanted to know about the usefulness and practicality of the geodatabase for my purposes and opted not to try and finish up the entire city.

Hydro Line  Event Table

Adding the Hydro Line Event Table to the Geodatabase

Map of Watersheds with Stream Reaches Color-Coded by Stability Index

Conclusions

The Austin Erosion Assessment provides an extensive amount of data regarding the current state of the City’s streams.  Much of the data exists mostly in a paper format – as evidenced by a shelf full of books: one or two for each of the 17 watersheds studied.  The stability index information was handwritten on forms provided for that purpose.  Some of the cross-sections also appear to have been drawn by hand.  Tables were apparently developed in Excel and maps drawn in CADD.   One important conclusion is that utilizing a geodatabase for centralized and spatially oriented data storage would represent a major shift from the data handling methods used in developing the study. 

However, it is feasible for a much of a study such as Austin’s Erosion Assessment to be developed and stored electronically and referenced to a GIS stream network.  Besides the numerical data that I input into the geodatabase, there are also numerous cross-sections and several hundred photos.  A laptop or palmtop could have allowed the initial stability index data to have been input directly into a database.   Cross-section data could be stored in GIS.  Digital photos could be referenced to a GIS map.  All of the numerical results could be stored as one or several HydroLineEvent tables.  Besides being feasible, it would also be beneficial.  For the engineers originally performing the assessment, the use of a geodatabase would likely have greatly helped to streamline their efforts.  Having the geodatabase would also be useful for visual presentations of data for policy makers and public hearings.  The most important advantage of having the information stored as a geodatabase in contrast to a paper document is that it could become a dynamic assessment that is updated as stream improvements are made, land use changes occur and additional observations are made.

To accomplish establishment of a useful geodatabase for Austin’s stream network, it would first be necessary to amend the NHD to include all of the stream and tributaries the City deems important or to independently develop a feature class of Austin streams, dividing reaches according to its own criteria.  Such a standardized stream network would obviously have applications beyond erosion issues and would be useful for referencing flood information and water quality data also.  Though it is quite cumbersome to try and reference stream reaches used in the erosion study to the NHD, it would have been a relatively simple process to initially identify the study reaches in relation the NHD (or an independently-developed stream network feature class). 

 Options for Further Study

Using the Erosion Assessment as a Baseline for Future Investigations

The City of Austin’s Watershed Erosion Assessment provides an extensive amount of baseline data that can be used for further study of erosion processes in urban streams.  Numerous cross-sections and photos were taken to document the current channel geometry and stream conditions.  Future cross-sections will allow the actual enlargement to be compared to the predicted enlargement ratios and sediment yields.

 The Walnut Creek watershed was identified as an area most likely to experience significant stream erosion in future years due to anticipated development in the watershed and the fact that most of Walnut Creek and its tributaries are classified as alluvial.  It would be an appropriate study site for an ongoing assessment of changes in the channel cross-section.    

 Analyzing Flow Regime Changes and Correlating to Erosion

Because Austin does have an extensive amount of assessment data on stream erosion, it would be an appropriate area for analyzing the impacts of urbanization in Austin on minor system flow events and correlating the changes to stream erosion (i.e. my original project idea). 

 Assessment of Detention Impacts on Erosion

Lastly, the Austin assessment did not attempt to investigate any effects of the City’s stormwater detention policy on stream erosion.   However, it has been theorized that on-site and regional detention basins aimed at controlling major floods may in fact worsen stream erosion problems.  With enough data on similar reaches of streams with and without detention basins, it may be possible to try to analyze for impacts.  An alternative would be to model the impact of detention basins on the minor system flow events, since we already know from other studies that such changes greatly affect stream erosion.

Acknowledgements/Appreciation:

 Dr. David Maidment, CRWR

 Gerry Clayton, City of Austin Watershed Protection Department

References:

Technical Procedures for the Watershed Erosion Assessments for the City of Austin, Raymond Chan & Associates in association with Aquafor Beech Limited, September 1997.

River Course Fact Sheets, NC Stream Restoration Institute.