Lake Conroe Dam Releases' Impact on Flooded Areas: Post-Hurricane Harvey Study, Lecture notes of Communication

Download Lake Conroe Dam Releases' Impact on Flooded Areas: Post-Hurricane Harvey Study and more Lecture notes Communication in PDF only on Docsity!

Fall 2018 A Study on the Lake

Conroe Dam Release

Following Hurricane

Harvey

Emily Luomala

CE 394K: GIS IN WATER RESOURCES

Table of Contents

1.0 Introduction

1.1 Background Hurricane Harvey was a Category 4 hurricane that made landfall in Texas on August 25, 2017 [1]. While the United States has had its fair share of Category 4 hurricanes throughout history, the magnitude and size of Harvey’s destruction was unrivaled in historical records with analysts dubbing Harvey as a 1 - in-1,000-year flood [2]. Throughout the course of Harvey’s 5 - day downpour, roughly 27 trillion gallons of water were dumped over Texas and Louisiana with some weather stations in the Houston area registering more than 50 inches of rain [3]. To put this in perspective, this amount of water is equivalent to 1 million gallons of water per person in the state of Texas. This statistic is illustrated in Figure 1 [3]. The devastation incurred from Harvey included the destruction of more than 30,000 homes, the displacement of over 35,000 people to emergency shelters, and early estimates of $150 billion in recovery costs [3]. However, the physical water volume was not the only aspect of Harvey that affected the impacted communities. The existing infrastructure was put under additional strain and many experts have since called for a needed improvement in infrastructure resilience to better handle these catastrophic events [4]. The Lake Conroe Dam was one such structure that was unable to maintain safe operating levels under the threat from Harvey. The Lake Conroe Dam is located on the West Fork of the San Jacinto River, about 54 miles north of Houston. The dam was completed in 1973 and consists of an earthfill embankment just over 11,300 feet in length with an emergency spillway located near the center of the embankment [5]. The crest elevation of the dam is 212 feet above mean sea level (MSL) and the normal water level is 201 feet MSL with an additional 6 feet flowage easement allowed for extreme storm events [6]. While there is technically an additional 5 feet available before the dam overtops, the operational protocol for the safe and structurally sound operation of the dam is based on the 207 feet MSL water level [6]. During Hurricane Harvey, unprecedented water levels prompted officials from the San Jacinto River Authority (SJRA), the government entity that has control of the dam, to release 79,141 cfs of water from the dam, a volume that approached the average volume pouring over Niagara Falls [7][8]. The SJRA’s general manager, Jace Houston, made a statement about this necessary action claiming that the water would have been released Figure 1. For Scale: Water Volume per Person

regardless of their decision: either through the controlled release or over the dam’s gate, which would have put the dam at risk of failure [ 8 ]. There has been much contention regarding the decision to release such a record flowrate of water from the dam, the previous record being 33,360 cfs in 1994 [ 8 ]. Since the release, there have been a number of lawsuits with over 250 involved individuals claiming that the SJRA released the floodwaters knowing the risks of flooding downstream of the dam and failed to adequately warn the residents of imminent flooding [ 8 ]. The majority of home and businessowners believe that their properties would not have flooded due to the Harvey rains, but instead were solely impacted by the dam release. 1.2 Problem Statement This study aims to evaluate the impact of the Lake Conroe Dam release downstream of the dam, primarily through identifying flooded address points that resulted from the release flows. The SJRA provided a map of peak flows giving the peak inflow into Lake Conroe as well the peak release flowrate. This can be seen in Figure 2 below. Figure 2. SJRA Map of Peak Inflows ÔÕ ÔÕ ÔÕ ÔÕ ÔÕ ÔÕ ÔÕ ÔÕ ÔÕ **18. 5

19. 7
20. 4 31.** (^235)**. 2
21. 0
22. 1
23. 4
24. 5**

0 2. 5 5 10 Miles Release from Lake Conroe Dam 79 , 141 cfs I- 45 115 , 000 cfs Lake Creek Sendera Ranch Rd 62 , 600 cfs Spring Creek Spring, TX 82 , 100 cfs Caney Creek Near Splendora, TX 37 , 000 cfs (^) Peach Creek Splendora, TX 31 , 300 cfs West Fork San Jacinto Near Porter, TX 130 , 000 cfs East Fork San Jacinto River New Caney, TX 73 , 600 cfs San Jacinto River Authority USGS Stream Gage Direction of Flow Rainfall Totals (inches) Lake Houston Estimated Peak Inflow 400 , 000 cfs Cypress Creek Westfield, TX 24 , 100 cfs HWY 105 76 , 000 cfs Spring Creek Tomball, TX 48 , 800 cfs San Jacinto River Basin Estimated Peak Flows Hurricane Harvey August 25 - 29 , 2017 Estimated Peak Inflow Into Lake Conroe 130 , 000 cfs ÔÕ (Based on Limited Available Information)

2.0 Approach

The objectives of this study were accomplished through three main steps: (1) Creating a basemap of the area with associated characteristics (2) Determining a focus area (3) Utilizing Model Builder to perform the Height Above Nearest Drainage (HAND) analysis to assess the address points 2.1 Creation of Basemap To start this analysis, a basemap had to be created for the area of interest. The National Flood Interoperability Experiment Geospatial Database, NFIEGeo, for the Texas Gulf Coast region was first obtained from the ArcGIS online map and imported into ArcGIS Pro [ 9 ]. This database contained all necessary stream gages, flowlines, waterbodies, subwatersheds, and catchments for the entire Texas Gulf Coast region. To delineate the desired watershed boundary, the Hydrologic Unit Code 8 (HUC 8) was utilized to identify the subbasin that housed the Lake Conroe Dam. Using the USGS Watershed Boundary Dataset, 12040101 was determined to be the HUC 8 of interest [ 10 ]. This subbasin consists of 5 watersheds and 25 subwatersheds as can be seen below in Figure 3. Lake Conroe is located within the light blue watershed and the location of the Lake Conroe Dam can be seen labeled in the figure. Figure 3. Watersheds and Subwatersheds Lake Conroe Dam

Prior to any further analysis, the flowlines and land cover for the subbasin were observed to understand the basic flow of water through the subbasin and identify the different types of landcover in the area. The NFIEGeo provided the National Hydrography Dataset (NHD) flowlines and the land cover was obtained from the USA National Land Cover Dataset (NLCD) Land Cover 2011 within the Living Atlas. The graduated flowlines and land cover distribution can be seen below in Figures 4 and 5, respectively. As can be seen in Figure 4, the West Fork of the San Jacinto River cuts directly through Lake Conroe and joins up with Lake Creek downstream of the dam before flowing into Lake Houston at the southeastern most area of the subbasin. The land cover distribution in Figure 5 depicts developed areas in red, most of which are located directly around the lake and scattered downstream of the dam. This statistic was utilized to determine the primary focus areas for this study and will be discussed further in Section 2.2. The Digital Elevation Model (DEM) was then downloaded from the National Map with a 1/3 arc-second, 10 m resolution [11]. The subbasin fell between two datasets so they were merged and extracted to fit the subbasin area. From the DEM, a stream network was derived and utilized to calculate the Height Above Nearest Drainage (HAND). The HAND raster served as the basis for the evaluation of flooded stage heights and inundation depths in the analysis portion of this study. In addition to the stream network, Figure 4. Graduated Flowlines^ Figure^5.^ Land Cover^ Distribution West Fork of the San Jacinto River Lake Creek

There were two main criteria for determining a focus area: (1) Location downstream of the dam (2) Population centers (i.e. areas that would more likely be affected – flooded properties) As noted in the previous section and referencing back to Figure 5, the land cover distribution showed the majority of developed land downstream of the dam. To further refine the focus area, the US Census of Populated Places was used to identify the areas that housed the major population centers within the subbasin. This was overlaid on the catchpoly basemap and can be seen below in Figure 8. As can be seen in Figure 8, two focus areas were selected for analysis: the Conroe region located immediately downstream of the dam and the Kingwood region located at the southeastern most portion of the subbasin. While these focus areas do not encompass the entirety of either city, they were designated as “Conroe” and “Kingwood” for easy Figure 8. Study Focus Area based on US Census Conroe Kingwood

referencing. Of these focus areas, a total of 22 catchpolys were analyzed, 16 in the Conroe area and 6 in the Kingwood area. 2.3 HAND Analysis Once the initial base map was created and the focus area was chosen, the 22 selected catchpolys had to first be analyzed to obtain their flooded stage heights through the development of rating curves before the flooded address points could be identified. Model Builder was used for the majority of this section of analysis as the Object ID of each catchpoly could be parameterized. By adding the Object ID as a parameter within the model, once the steps and raster calculations were developed, the model could be run for each catchpoly by simply changing the Object ID in the Geoprocessing pane of ArcGIS. 2.3.1 Develop Rating Curves The first stage in the HAND analysis was developing the rating curves for each catchpoly of interest. The rating curve shows the relationship between stage heights and their respective discharges. As the discharge for each modeled scenario was known, the flooded stage height could be interpolated from the rating curve for each catchpoly. Prior to running the model, the length of the drainage line and the bed slope were manually identified for each catchpoly. To then develop the rating curves, the number of flooded cells, inundation depths, and slope raster were needed at various stage heights to obtain the parameters necessary for Manning’s equation to calculate the corresponding discharge. Stage heights of 1, 6, 10, and 14 m were used. These scenarios were built into the Model Builder so that the desired parameters could be extracted after one run of the model. Figure 9 to the left provides an example of one of the rating curves and Figure 10 on the following page shows a screenshot of the Model Builder. As can be seen in the rating curve, the obtained discharges were much higher than needed. For future work, a tighter range of stage heights could be used to develop a more refined rating curve. This will likely also give more accurate flooded stage heights. 0 2 4 6 8 10 12 14 16 0 100000 200000 300000 400000 Stage height (m) Discharge (cfs) Discharge vs. Stage Height Figure 9. Sample Rating Curve for Catchpoly - Object ID 506

2.3. 3 Identify Flooded Address Points Following the identification of flooded stage heights for each scenario, the flooded address points could be determined. The address points for the entire subbasin were obtained from the Hurricane Harvey story map [12]. A second Model Builder was created to identify the flooded address points by parameterizing both the Object ID so it could be run for each catchpoly and the flooded stage heights for each scenario. The model first extracted the address points for each catchpoly and overlaid the previously determined HAND raster to assess the HAND value at each address point. The flooded address points were then identified by evaluating which address points had HAND values that were less than the flooded stage height for each scenario. These steps produced a map of the address points subject to flooding for each scenario. Figures 11 and 12 show screenshots of the model builder and parameter options, respectively. Figure 11. Model Builder to Assess Flooded Address Points Identifying flooded address points – Scenario 1 Identifying flooded address points – Scenario 2 Determining inundation depths – Scenario 1 Determining inundation depths – Scenario 2 Identification and selection of catchpoly of interest

Figure 12. Parameter Options for Model Builder 2 Parameterizing the Object ID allowed for the selection of the catchpoly of interest The light blue boxes highlight the second parameter: flooded stage heights. These were used to determine the inundation depths and at-risk address points

As can be seen in Figure 15, there are more impacted address points in magenta (representing the no dam scenario) than there are impacted address points in blue in Figure 14 (representing the dam release after Harvey). This is most noticeable in the central area of the collection of catchpolys. These results are very much expected as the flow rate for scenario 2 is ~1.64 times greater than the actual peak flowrate of the dam release. The change in flowrate nearly parallels the change in number of affected address points with an increase of 1.66 times more address points at risk of flooding under the conditions from scenario 2. While these statistics are important to note when considering future potential releases or floods, the distribution of inundation depths are another important result to look into. Figures 16 and 17 show the inundation depth distribution for scenario 1 and scenario 2, respectively. Figure 14. Flooded Address Points Scenario 1 - Conroe Figure 15. Flooded Address Points Scenario 2 - Conroe

As can be seen in the pie charts, there is a much greater percentage of address points flooded more than 2 m (~ 6.5 ft) with some properties being inundated as much as 8 m (~26 ft!) under the conditions of scenario 2. Thus, in addition to more address points being subject to flooding, the no-dam scenario also increases the inundation depth of the address points, which would inflict more overall damage to the property owners. Figure 16. Inundation Depth Distribution Scenario 1 - Conroe Figure 17. Inundation Depth Distribution Scenario 2 - Conroe

Despite consisting of only 6 catchpolys, there were more than double the number of total address points in the Kingwood focus area than in the Conroe focus area. The difference in flooded address points between the dam release and no-dam scenario is also more noticeable in this focus area. The magenta (no-dam scenario) flooded address points are more dense in the central portion of the focus area and extend more into the southeastern most portion of the area. The distribution of inundation depths were observed for the Kingwood area as well. Figures 21 and 22 show the inundation depth distribution for scenario 1 and scenario 2, respectively. Figure 19. Flooded Address Points Scenario 1 - Kingwood Figure 20. Flooded Address Points Scenario 2 - Kingwood

As compared to the Conroe area, the inundation depths in the Kingwood area for scenario 1 are more concentrated within the 2 m or less range. For both areas the distribution of inundation depths is much greater under the second scenario. However, some Kingwood area address points reach an inundation depth of more than 8 m under the second scenario. These distribution pie charts demonstrate not only the variability in address point elevation and risk, but also provide a general comprehension of the damage that can be expected during releases. Many of these inundation depths are capable of completely destroying a property and putting the residents or owners at risk if they do not evacuate. This underscores the importance of good communication and forewarning prior to a release of these magnitudes. Figure 21. Inundation Depth Distribution Scenario 1 - Kingwood Figure^22. Inundation Depth^ Distribution^ Scenario^2 -^ Kingwood