Andrea M. Jackman, Ph.D. ASFPM June 3, 2015 Using NFHL Data for Hazus Flood Hazard Analysis: An Exploratory Study
What is Hazus? Hazus is FREE software distributed by FEMA Risk MAP which models structural and economic losses due to hurricanes, earthquakes, and FLOODS Download software and data for all 50 states at: http://msc.fema.gov/portal/resources/akamai 2
How does Hazus Flood work? GEOGRAPHY Demographics Structures Economy HAZARD + Type* = Severity Damage Functions RESULTS Structural damage Economic loss *Unlike the other hazard types in Hazus, the flood model requires you first model the flood hazard to generate a depth grid before you run the results model 3
Project Goal Hazus can accept depth grids from other sources Hazus currently models hydrology & hydraulics, but the results do not always match similar analysis contained in NFHL data NFHL is the official regulatory product for Risk MAP NFHL has excellent data quality, but the format is not currently supported for direct import into Hazus as a depth grid Is there a way to replace depth grid generation in Hazus with NFHL data? 4
1. AE Zone Approach 1. Extract BFE data from the FEMA AE flood zones and convert each end to point data (see Figure 1) 2. Using the point data, generate an interpolated surface using inverse distance weighted (IDW) method in GIS to approximate the 100 year flood elevation at any point 3. From the generated surface, subtract the Digital Elevation Model (DEM) to get a flood depth grid using GIS raster methods 601 600 601 600 599 599 598 598 Green = BFE polylines available from FEMA AE Zones, showing lines of equal elevation for the 100-year flood Blue = Extracted endpoints of each BFE polyline, showing points of equal elevation for the 100-year flood
1. AE Zone Approach 1. Extract BFE data from the FEMA AE flood zones and convert each end to point data (see Figure 1) 2. Using the point data, generate an interpolated surface using inverse distance weighted (IDW) method in GIS to approximate the 100 year flood elevation at any point 3. From the generated surface, subtract the Digital Elevation Model (DEM) to get a flood depth grid using GIS raster methods 601 600 601 600 599 599 598 598 Blue = Extracted endpoints of each BFE polyline, showing points of equal elevation for the 100-year flood Point data is used to interpolate a surface of 100-year flood elevation for any point in the vicinity (Image Source: QGIS)
1. AE Zone Approach 601-600 1. Extract BFE data from the FEMA AE flood zones and convert each end to point data (see Figure 1) 2. Using the point data, generate an interpolated surface using inverse distance weighted (IDW) method in GIS to approximate the 100 year flood elevation at any point 3. From the generated surface, subtract the Digital Elevation Model (DEM) to get a flood depth grid using GIS raster methods 601-600 600-599 602-602 602-600 603-602 599-598 1 1 1 2 0 1 1 600-599 1 600-598 599-598 594-592 598-596 598-596 2 1 2 2 2 598-596 2 604-604 599-595 590-586 0 4 4 Subtracting the DEM from the interpolated flood elevation surface using GIS yields a flood depth grid for use in Hazus
2. Zone A Approach (where BFEs are not available) 1. Calculate the centerline of the waterway 2. Perpendicular to the centerline of the waterway will be lines of equal elevation (opposite ends of the lines should be equal to each other) 3. Extract the values of the endpoints to point data, watching for points where an elevation is lower than the point s nearest downstream neighbor 601 600 601 600 599 599 598 598 Green = Centerline of waterway with perpendicular estimates for lines of equal elevation Blue = The endpoint of each line of equal elevation can be extracted to obtain point elevation data, and used as input for the AE Zone Approach, Steps #2-6
Study Areas and Data Methodology applied for: Buckhorn Creek in Larimer County, CO City of Boulder, CO City of Longmont, CO Columbus Junction, IA Elkader, IA 9
Findings Colorado Poor agreement between NFHL floodplain data and DEM Two dimensional flood plain boundary draped over the DEM to approximate a 3D floodplain boundary. Note the BFE polylines do not meet the floodplain boundary either the horizontal or vertical axes Similarly, the depth grid generated shows several areas of negative (red) flood elevations (dry) 10
Findings Colorado Percentage of sampled points with negative flood depths note the type of DEM used does not appear to significantly reduce errors 11
What could cause so many negative depths? Possible causes: 1. Regulatory floodplain management: islands are removed and regulatory flood plains are simplified for enforcement purposes. 2. Significant spatial displacement between the floodplain boundary and DEM elevations 3. Narrow channels with high slopes that magnified the misalignments 4. Presence of stream islands 5. Floodplain boundaries developed using terrain information distinct from the NED and LiDAR DEMs. 6. The presence of elevated structures 12
Findings Iowa Fewer negative depths, with many occurring near roadways or the center of the floodplain where there is unlikely to be any built areas to impact Hazus loss estimation Using the NFHL methodology and 3m LiDAR DEM, percentage of sampling points with negative depths was reduced to single digits Floodplain data for Iowa was available from a recentlycompleted study by the Iowa Flood Center. The same 3m LiDAR DEM used in the flood study was available, meaning the floodplain and DEM were already integrated. This resulted far fewer negative depth errors, good horizontal alignment, and was replicated in Elkader, IA where steeper terrain exists. 13
Conclusions There appears to be success where floodplain data sources are aligned, and BFE s are available, including areas of poor DEM resolution for the Iowa test cases. The Colorado test cases at all DEM resolutions had relatively high rates of negative flood depths, but may still produce products better than our current Level 1 H&H approach. For A-zones, quality issues related to the DEM and flood hazard boundary mismatches will require more work to automate a process that can filter out errors and create depth grids from the NFHL. Recommendations are provided suggesting the approach needed to evaluate and establish error thresholds, but it is still uncertain if a completely automated process could be developed. FEMA s First Order Approximation (FOA) products, since they do not have a mismatch with the DEM should be able to be used directly. 14
Next Steps 1. Establishing thresholds for acceptable input criteria 2. Analyzing additional regions with the original proposed methodology 3. Use of a reference/standard flood hazard dataset 4. Using geo-rectified flood hazard data 5. Evaluating centerlines and using NFHL cross sections 6. Performing a Hazus sensitivity analysis 15
Thank You! andrea.jackman@us.ibm.com 16