1	HAZUS-MH 1.1 Workshop Michael S. Scott, PhD msscott@salisbury.edu August 10, 2005
2	Section 1: Introduction The 1^st meeting of the Maryland HAZUS User’s Group? What is the ESRGC? Project background, goal, & method Vulnerability project results Conclusions Goals of this workshop
3	What is the ESRGC? Eastern Shore Regional GIS Cooperative Mission: To support the development of GIS technology in municipalities and counties on Maryland’s Eastern Shore Funded by the Regional Councils of the Mid-Shore and the Lower Shore www.esrgc.org
4	Project Background Vulnerability – Potential for loss Situated at the intersection of the physical hazard and the social system Unfortunately, not simply a matter of map overlay Depth of flood, recurrence, structure characteristics, mitigation techniques, etc Very complex, interrelated system
5	Project Goal Complete a General Building Stock Damage and Loss analysis for a 100-year flood event for each county in the State, including the City of Baltimore, using the latest HAZUS-MH Flood module
6	Method For each jurisdiction, calculate 100-year flood zone using HAZUS-MH 1.1, intersect with the amount and type of building stock to calculate loss potential Run both riverine and coastal where appropriate Complete a Level 1 HAZUS analysis Using default data Consistent quality across jurisdictions
7	Results Report will be made available at www.esrgc.org later today Over 1,328 square miles of Maryland are vulnerable to the 100-year flood 15 jurisdictions (of 24) are vulnerable to coastal flooding
8	Totals for Maryland Building stock exposure (in 2001): $7.99 billion Building damage in sq ft: 109,665,000 Worcester: over 21 million or 19.4% of state total Somerset: 5.7 million but 68% is predicted to be “substantial Building damage in numbers of structures: 44,755 Anne Arundel: 7,038 buildings damaged Direct economic losses: $8.12 billion Prince George’s: $1.28 billion Prince George’s, Anne Arundel, and Worcester equal almost 40% of Maryland’s total loss potential
9
10
11
12	Study Conclusions Vulnerability to flooding in Maryland is not evenly distributed Prince George’s, Worcester, Anne Arundel, Montgomery, Washington, Howard, Baltimore County, and Baltimore City disproportionately effected Areas with large vulnerable areas happen to be forecasted for increased development Dorchester, Somerset, Talbot, Caroline, Worcester There is a need to look more closely at particular trouble spots to further refine the analysis
13	Goals of the Workshop Introduce the capabilities and limitations of HAZUS-MH Discuss the functionality of HAZUS-MH Examine the steps necessary to complete a Level 1 HAZUS-MH model run Analyze the results of the recently-completed HAZUS-MH model runs for each county Discuss the next steps in vulnerability modeling
14	Section 2: Model Expectations HAZUS-MH 1.1 – Level 1 What it is and what it isn’t Capabilities Variables for consideration How it works How you get it Hardware requirements Software requirements Installation hints
15	HAZUS-MH 1.1 – Level 1 Designed to estimate a wide range of losses from flood (as well as other hazards) Internally, several stochastic flood models provide the engine, as well as ArcGIS 9.0.1 Uncertainty/error are significant but not quantified in a Level 1 analysis New version scheduled to be released in late summer/early fall
16	What it is… An rough estimate of potential losses A very useful and consistent tool to measure differences in vulnerability across and between counties A dynamic modeling tool that allows the user to ask “what if” questions Helpful to prepare for the inevitable Complicated and computer-intensive
17	What it isn’t… A vulnerability crystal ball Calibrated to your specific economic situation Any better than its inputs A substitute for a engineering-based flood study Easy to use or interpret
18	Capabilities Calculating flood depths in both riverine and coastal contexts Modeling losses to the census block level Examining multiple dimensions of loss Analyzing multiple recurrence intervals or specific discharge amounts Viewing up to 4 counties simultaneously
19	Variables for Consideration 30-meter digital elevation model General Building Stock 2000 Census of Housing data for buildings Dun & Bradstreet data for non-residential buildings US Department of Energy for regional differences in square footage, construction types, etc Limited number of stream gauges
20	How it works: Determine Riverine Flood Depth Creates flow grid from DEM Identifies stream reaches from flow grid Associates each reach with a drainage area Identify stream gauges in drainage area Approximate the floodplain for each reach Create a set of flood depth cross sections Interpolate from cross sections to grid cells
21	How it works: Determine Coastal Flood Depth Characterize the shoreline (i.e. bluffs, dunes, erosion control) Enter flood depths per recurrence interval plus wave height (if appropriate) Run one or more coastal flood models Dune/bluff erosion, wave height, and wave runup) From models, determine flood depth
22	How it works: Loss Estimates Uses depth-damage curves that need: Occupancy class, foundation type, and assumed first floor elevation Depth of flooding throughout the census block Uses NFIP claims to create depth-damage curves for “typical” construction types Matches up buildings in a block and depth within a block to depth-damage curves to estimate damage
23	How you get it Request a copy from FEMA http://www.fema.gov/hazus/ Set of DVDs and CDs Data DVDs are regional
24	Hardware Requirements FEMA recommended configuration 2.5 Ghz processor 512 GB RAM 80 GB HDD 800x600 video card ESRGC configuration 3.4 Ghz processor 2 GB RAM 80 GB HDD 128 MB, 1600x1200 video card Still takes approximately 8-10 hours per model run
25	Software Requirements HAZUS-MH 1.1 Avoid version 1.0 if at all possible Processing takes 3-4 times as long (several days for one county) Limit on raster processes Very buggy ArcGIS 9.0 – Service Pack 1 Current SP is 3…must keep old version of Arc 9 Must have ESRI’s Spatial Analyst Windows XP, Service Pack 1 SP2 causes problems
26	Installation Hints Start with a fresh copy of Windows and ArcGIS (no SP 2) Copy the entire data DVD on to the hard drive Must manually change the registry entry to point to the new location Increase your virtual memory to at least 4 GB If you are getting more than a couple hydrology/hydraulic errors in a county, it’s possible you have corrupted installation disks If you have disks from before Feb 28, you need a major patch Be sure to have administrator privileges
27	Section 3: Modeling steps Create a study region Study region set-up Study case set-up Riverine flooding parameters Coastal flooding parameters Determine riverine and coastal hazard Analysis/loss estimation Report results Or…import a model run that’s already done!
28	Creation of a Study Region Choose your state Choose your county All flood modeling is done at the county level using census blocks, even if you don’t want to analyze an entire county Maximum of 4 counties Implications in studying watershed vulnerability
29
30
31
32	Study Region Setup Decide on type of flooding Riverine, coastal, or both Makes a difference in the size and shape of the DEM you’ll need Download DEM from NED online The button to “Navigate directly to NED download” does not seem to be working Go to http://seamless.usgs.gov and enter the bounding coordinates manually Set the DEM data paths Will begin to build the hillshade visualization
33
34
35
36
37
38
39
40
41	Study Case Set-Up Calculate stream reaches Must chose minimum drainage area Essentially a level of precision Limited to computer processor and time Doesn’t modify the chosen floodplains Create study case Name and description Can have multiple study cases per county Different parameters per study case Selecting reaches and coastlines to be included in the study case. Be sure to select all the reaches and coastlines up front…you can’t add more later
42
43
44
45
46
47
48	Riverine Flooding Parameters Hydrologic analysis Individual drainage basins can be run separately All in a county can be selected at once If the analysis fails, there is a log to check what stream reach it failed on This allows you to remove the offending reach
49
50	Coastal Flooding Parameters Shoreline characterization Type of shoreline Rocky, large/small dunes, flood control structure Controls what models will run Flood height With or without wave setup 10, 50, 100, and 500 year flood
51
52
53
54
55	Determine Riverine and Coastal Hazard By recurrence interval 10, 50, 100, and 500 Single recurrence interval By discharge value By annualized loss
56
57
58
59
60
61
62
63	Analysis/Loss Estimation Possibilities range from building stock damage to casualties to essential facility damage to debris removal costs This study focused on general building stock damage and loss, as well as indirect economic losses Includes damage by type and by occupancy, count of damaged buildings, and economic losses such as structural damage, contents damage, relocation costs, wage losses, etc. Some analysis results are dependent on others but the model accounts for that automatically
64
65	Report Results Results are reported with maps and tables Every variable calculated can be mapped by census block Crystal Reports automatically generates summary reports Exportable to Excel, etc
66
67
68
69
70
71
72
73
74	Import a Study Region Luckily, most of this work has been done for you! Can import the model run used for the state study to examine your county’s vulnerability Any changes to the model parameters will require a re-run.
75
76	Major HAZUS Errors/Bugs In the past: Counties without any results Stream reaches that cannot have hydrologic analysis completed Stream reaches that cannot have hydraulic analysis completed “Riverine and Coastal” model runs returns only coastal results Most of the major errors have been fixed with the patch released in March 2005 Now, the concerns turn to the more insipid error: those within the results
77	Section 3: Analysis of Results Examine the output in ArcGIS 8.3 Understanding the Flood Depth Grid Working with Census Blocks
78	Let’s get into the GIS! Insert CD and copy contents to the D: drive under studentfiles Open ArcMap Load in: reg_bound/reg_bound_utm18 flood_bound/flood_bound_utm18 V:\hazus_reports\grids\yourcounty\rpd100*
79	Understanding the Flood Depth Grid Difference between the flood elevation and the ground elevation at each grid cell Water depth at each cell Some results are suspect Backwater areas Pools at confluences Problems with 30-meter DEM Bridges as dams, for example Won’t generic LiDAR make this worse?
80	Working with Census Blocks Load: dam_occ/res.shp Thousands of square feet of residential building damage dam_occ_cnt/res.shp Numbers of residential buildings damaged dam_type/wood.shp 1000s of square feet of wood building damage dam_type_cnt/wood.shp Numbers of wood buildings damaged econ_full_total/total.shp Thousands of dollars in direct economic losses from buildings
81	Working with Census Blocks Smallest unit of census data released to the public Blocks are not a uniform size but try to use visible landscape features for definition Have to assume that population is evenly distributed It’s possible to have a block intersect the flood zone but have no damage
82	Working with Census Blocks Dam_occ/res.shp (1000s of sq ft) CensusBloc – Block ID number Length and area are WRONG Totalsquar – total amount within flood area (including undamaged) Sqftwithsu – amount with substantial damage (>50% damaged) UndamagedS – amount undamaged SqftDmg1to – amount damaged between 1 and 10% SqftDmg11to – amount damaged between 11 and 20% TotalDam – total amount damaged
83	Working with Census Blocks Dam_occ_cnt/res.shp (Numbers of structures) CensusBloc – Block ID number Length and area are WRONG TotalBuild – total amount within flood area (including undamaged) BldgswithS – amount with substantial damage (>50% damaged) UndamagedB – amount undamaged PctDmg1to1 – amount damaged between 1 and 10% PctDmg11to – amount damaged between 11 and 20% TotalDam – total amount damaged
84	Working with Census Blocks econ_full_total/total.shp (Thousands of dollars) TotalLoss – total direct economic losses BuildingLo – cost of building damage ContentsLo – cost of damage to building contents InventoryL – cost of damage to inventory Relocation – cost of temporary relocation IncomeLoss – cost of lost owner’s income due to displacement RentalInco – cost of lost rental income WageLoss – cost of lost wage income DirectOutp – unknown Employment – unknown
85
86	Section 4: Next Steps Level 2 Analysis Special tools Data needs Comparisons Recurrence Intervals Digital vector flood maps Let skepticism be your guiding force
87	Level 2 Analysis Much more detailed and realistic Abandons national, default datasets for local, specialized information BIT and FIT Building Information Tool Used to get detailed building stock info into HAZUS Flood Information Tool Used to get flood extents and depths into HAZUS Time consuming and technical but probably worth it.
88	Level 2 Data Needs More accurate DEM LiDAR, 2-meter cell size? Unsure about the effect of local minima and maxima Detailed inventory of the built environment Structures Location, occupancy, construction, height to 1^st finished floor, characteristics Essential facilities High potential loss facilities Lifelines
89	Level 2 Data Needs More accurate flood hazard data Detailed DEM Series of flood elevation cross sections Boundary of flood zone Detailed shoreline characterization
90	Comparisons Comparison between 50-, 100-, and 500-year flood zone extents and depths Often considered more vulnerable when a large distance between 100- and 500-year is found (people tune out to flood warnings) Comparison between HAZUS output and DFIRMs/Q3 data St. Mary’s example Comparison between HAZUS output and MDProperty View Be very careful…
91	And finally… Be very skeptical of HAZUS-MH 1.1 until you have a reason not to be With the number of critical errors present in the first iteration of this release, one has to wonder what errors lie beneath No quantification of uncertainty It’s just too new…hasn’t been properly vetted by the scientific community in a variety of situations