Lake Travis

Changing the 100 year Floodplain

A GIS Project By: Justin Kockritz and Ryan Reynolds

University of Texas at Austin - Department of Geography

A View from the Air of Lake Travis

Project Description:

Since the completion of Mansfield Dam (then the Marshall Ford Dam) in1941, the 100 year floodplain for Lake Travis has been recognized as 716 feet asl.  However, a recent Corps of Engineer study in 2003 revealed that an elevation of 722 feet asl would be more accurate.  The goal of this project was to create a new 100 year floodplain map for Lake Travis.  Not only will this data be important for predicting floods, but changing the elevation of the floodplain means that hundreds of people that were outside the "danger zone" will now be within and will be required to purchase flood insurance under the National Flood Insurance Reform Act of 1994 if they have not done so already.

 

1991 Flood Event on Lake Travis

 

Background:

The Federal Emergency Management Agency (FEMA), ranks Travis County in the top 10 percent of counties in the nation that are at risk for significant flood damage.  With this in mind, FEMA has made it a priority to redraw floodplain maps in high risk areas like Travis County.  Although work is underway on this project with the cooperation of FEMA, LCRA, TNRIS, and the City of Austin, the study will not be completed for three to four years.  Our project provides an early example of the impacts one can expect from the results of this study.    

10 DOQQ's Showing Lake Travis

Two Foot Contour Data Overlaid on DOQQ

ArcToolbox Was Used to Convert Contours to Polygons

Editing a gap in the 2 foot contour data

 

Data, Methods, and Analysis:

    The two sources for our project data were TNRIS and the City of Austin GIS Data Set.  A total of ten DOQQ's were downloaded from TNRIS that when combined, covered the entire Lake Travis area.  The following data was downloaded from the City of Austin: 2-foot contour data for each DOQQ, 1997 Building Footprints, Travis County Parks shapefile. 

One of the limitations of this project was that the most recent data available for the building footprints was from 1997.  There is no doubt that there has been a large number of structures built around Lake Travis in the past seven years, so the final product here is offered as a reference only and is not entirely accurate.  The most recent DOQQ's available were also from 1997.

The software used for analysis was ESRI ArcMap v. 8.3.  Using GIS technology to address this issue was a logical choice.  Multiple data sources can be geo-referenced and overlaid easily using ArcMap.  Through GIS, it was possible to see exactly which structures were affected by the change in the 100 year floodplain as the building footprints, 2 foot contours, and DOQQ's were layered.  The increase in the number of structures within the floodplain was calculated using ArcMap's Select by Location feature.    

 

View of ArcMap's Select by Location feature

 

This project could not have been done accurately without the 2-foot contour data.  A floodplain map needs to be as accurate as possible when it covers a residential area to provide the best prevention against flood damage.  Since the mapping required elevations of 716ft and 722ft, this data met the minimum requirements.  Some editing was required because the 2 foot contour data had gaps at various locations, mostly along the outer arms of the lake.   

In order to continue with the floodplain analysis, the gaps were repaired and both the 716ft and 722ft contour shapefiles were selected separately from the entire contour shapefile and exported to a new shapefile.  Using Spatial Analyst, these features were then converted to raster.  After that, the grid was converted to coverage, and then the coverage was converted to a polygon.  This produced a polygon of each contour elevation. 

Using the 716ft and 722ft contour polygons, the Select by Location feature was used to determine those houses that touched, or were within the 100 year floodplain.  This was done two separate times to find those houses within the old floodplain and the new floodplain.

The total area of the polygons was calculated using Microsoft Excel because each polygon was separated into several different sections.  The total values can be seen in the Surface Area Covered table.  The number of buildings in each floodplain was calculated in the same manner, and those totals can be found in the Buildings Affected table.

The only problems encountered during the data analysis were with the gaps in the two foot contour lines.  Once these were repaired, the rest of the analysis went smoothly. 

 

Table Showing Surface Area of Old and New Floodplains

Table Showing Number of Buildings Located in Old and New Floodplains

 

Floodplain Analysis Showing Old Floodplain Houses in Orange, New Ones in Red

716ft Floodplain (Light Blue) Overlaid on 722ft Floodplain (Dark Blue)

Findings: 

            With each shapefile properly georeferenced and layered, an analysis of the data can begin in order to answer the original question.  First, the area of each floodplain shapefile is calculated, representing the surface area for each.  The current floodplain, along the 716-foot contours, covered 41.21 square kilometers, or 10,183.53 acres, whereas the proposed new floodplain at 722 feet would encompass 43.18 square kilometers, or 10,670.43 acres.  This yields a difference in area between the two of 1.97 square kilometers or 486.90 acres, an increase of 4.78% from the area of the original floodplain.  The greatest difference between the two floodplains can be seen in two main areas: the Long Arm of Lake Travis, and along the northeast edge of Hudsons Bend.  These two areas are both between two inlets of the lake and have slopes that are not as steep as other areas around the lake, allowing the water to rise over the land fairly easily.

            Using the building footprints shapefile, the number of structures that would be affected (those lying wholly or partially within the floodplain) can be counted.  In the 716-foot floodplain there are 1934 buildings, but the proposed 722-foot floodplain would affect 2379.  This means that 445 buildings that lie outside of the 716-foot floodplain, but that would be affected by the change in elevations, representing a 23% increase.  The buildings that would be affected only by the new floodplain are well distributed across the lake area, but there are higher numbers on Hudsons Bend where almost a quarter of them were located, however the densities are rather low.  If the newly affected buildings had been found in high concentrations it might be feasible to initiate flood control programs in that specific area, but the low density makes that nearly impossible except on a house-by-house basis. 

            There are also some areas that can be seen on the DOQQs where no buildings are present, but where the transportation infrastructure of roads is already in place waiting for development.  While the exact number of buildings that will be affected cannot be determined from the data sources used in this research, it does appear that portions of Lakeway might be the most likely areas for development within the proposed floodplain.  Here there are some areas where a new subdivision can plainly be seen in the making, and some of the lots would certainly lie in the new floodplain.

 

Contour Line Converted to Grid Format: A Possible Source of Error

 

Potential Errors:

 

            Throughout the course of this research there are several areas where sources of error could not have been avoided, and the effect their presence had on the final results cannot be ignored.  The most likely and largest source of error stems from the simple fact that the data used for the majority of the project is seven years old, preventing a truly current study of the area.  Since 1997 when the aerial photographs for the DOQQs were taken, and the contour lines and building footprints were created the Lake Travis region has grown greatly along with the rest of Austin.  This will obviously result in a higher number of structures that would be affected by the change in floodplain level, but it can be assumed that the contours and the physical geography of the study area have changed only minimally if any over that time.

            Another source of error comes from the methods used for analysis in this research.  Since the contour lines did not always meet up exactly, or in some areas the 716 and/or 722-foot contour line may not have been present, there was some digitizing that was done.  This would change the surface area of the floodplain, but more importantly it may have been the determining factor in which properties were deemed to be within the floodplain.  The metadata associated with the contour line shapefile advises that they not be used in place of property level surveying, and that same caveat should be heeded in this research as well.  Property owners could use this research to determine the level of risk to their structures, but a detailed survey in the field should be required before the final designation is made.

            Also within the analysis portion is the loss of data during the process of converting the polyline shapefile to a polygon shapefile.  The largest error source here occurred when converting the line feature to a raster grid.  Representing the one dimensional line as a two dimensional area, however small the cell size was, will always yield skewed results.  The error for each cell is minimal, but over an area as large as Lake Travis, those tiny errors accumulate into a significant number.  There is no error in converting the raster grid to a coverage, and then to a polygon shapefile, but they obviously maintain the original error.  This may have led to discrepancies in area calculations for each floodplain, but also in determining which buildings would be affected.  The floodplain itself is not composed of small cells, so modeling them as such will affect the results, but here the error is within a reasonable margin, and valid for the large scale study of the lake itself, but not necessarily at the individual lot level.

 

New Houses Now in 100 year Floodplain in red

 

Layout of Southeast Mansfield Dam Area

Meaning:

 

            Since 1994 FEMA has required that in order to take out a loan, which is often needed in order to purchase, construct or renovate a building, on a property laying in the 100-year floodplain the owner must have flood insurance, such as the National Flood Insurance Program (NFIP) offered through FEMA, but it is often recommended that all home owners have such insurance.  However, existing structures that would be affected by the proposed change to the floodplain would likely be grandfathered in meaning flood insurance would not be mandated, and in some cases they may receive discounted rates or even tax breaks on any policies they do take out.  However, if the proposed floodplain change is accepted, any new developments within it would require flood insurance, either through the NFIP or another source, as they would in any other area.

            These results should not be a substitute for a property survey, but for current or potential homeowners along the lake it may serve as an appropriate starting point for more in depth research.  It would be wise for all property owners to investigate further to determine specifically if their property would be affected by the change, but perhaps those that have been identified here could be given first priority, followed by those closest to the new proposed floodplain.

 

 

Houses on Lake Travis

Future Research:

 

            One of the most important topics for future research would be how this proposed change to the 100-year floodplain would affect Mansfield Dam.  Being the only flood control dam in the LCRA system, Mansfield Dam is a key component in flood management and prevention, and the possible addition of 2.6 to 2.7 billion cubic feet of water in Lake Travis would have a potentially huge effect on the floodgate procedures at the dam.  It would be important for hydrological engineers to look at what could feasibly be done throughout the LCRAs jurisdiction, but specifically at Mansfield Dam, to minimize the possible damage that would be caused by a 100-year flood event, in Lake Travis and downstream in the City of Austin also. 

            FEMA could use this research in conjunction with data on building permits that have been awarded between the time of this study (1997) and the present day.  This would give them a much more accurate account of the properties in the area, and any new ones could be plotted on the floodplain maps created here to determine their risk of laying in the new floodplain.  More recent secondary data such as the DOQQs and building footprints, or new data like property tract lines, could also be used to aid in this as well.

            Individual property owners, or potential developers or home buyers might also want to use this data as an initial step in their decision in whether or not to purchase or upgrade their home.  More detailed surveying would of course be required, especially for determining if flood insurance would be necessary, but a basic, general idea about the flood risk for the property could be gained here.

 

Sources: