Hazards & Vulnerability Research Institute (HVRI)


 

SHELDUS™ | Frequently Asked Questions

  1. Why do I have a different number of counties from decade to decade?
  2. Why do I only see 10 results on the web though it seems that my request should contain many more records?
  3. Why can't I find a certain event in the database?
  4. Will my request contain double records if I search for two or more different hazard types?
  5. What data sources have been used to compile the database?
  6. Does the database contain information on Puerto Rico, Guam, or other U.S. territories?
  7. What do the different columns of my downloaded .txt file represent?
  8. Why can't I download the complete database?
  9. How can I map downloaded data?
  10. What does each record represent?
  11. What is a FIPS code?
  12. How can I find out to what state a certain county belongs?
  13. Can I use Excel to manipulate and analyze the data?
  14. How do I import the data into Excel?
  15. Why are there injuries and fatalities with decimal places?
  16. Can I compare dollar losses across years?
  17. There are numerous cases where the damage figures are far less than $50,000. Is this only due to the fact that damages have been divided based on the geography or have more specific sources been used?
  18. Why can't I join the SHELDUS™ attribute table with a spatial table via FIPS codes?
  19. Why are county names sometimes incorrect in Alaska?
  20. Why are losses for Hurricane Katrina much smaller than other official loss estimates?
  21. What is the GLIDE number?
  22. What is a presidential disaster declaration?
  23. Why do losses in my county include coastal hazards when my county is not a coastal county?
  24. How are coastal hazards defined?
  25. How do you categorize data?

1. Why do I have a different number of counties from decade to decade?
This is caused by changes in county boundaries. The changes are documented in the metadata and are the following.

1960: There were 3,068 counties, 30 independent cities in Virginia, and the independent cities of Baltimore and St. Louis for a total of 3,100 enumeration units.

1970: There were 3,067 counties (Princess Ann County, VA was absorbed by Virginia Beach city in 1963; Menominee County, WI was carved out of Shawano County, WI in 1961), 38 independet cities in Virginia (Salem/51775 became independent of Roanoke/51161 in 1968; Bedford City/51515 became independent of Bedford County/51019 in 1968; Emporia became independent of Greensville County/51081 in 1967; Lexington/51678 became independent of Rockbridge County/51153 in 1965; Fairfax/51600 became independent of Fairfax County/51059 in 1961; Arlington was a county in 1960 but changed status to an independent city in 1970), and the independent cities of Baltimore, St. Louis, and Carson City (Ormsby, NV was consolidated with Carson City/32510 in 1969).

1980: There were 3,065 counties (Washabaugh, SD/46131 merged with Jackson, SD/46071 in 1979; City of Nansemond, VA became Nansemond County in 1972, then merged with the City of Suffolk, VA in 1974). 41 independent cities in Virginia (Poquoson/51735 became in dependent of York County/51199 in 1976), and the independent cities of Baltimore, St. Louis, and Carson City.

1990: There were 3,067 counties (La Paz, AZ/04012 was formed from part of Yuma County, AZ in 1982; Cibola, NM/35006 was formed from part of the western portion of Valencia County, NM in 1981), 41 independent cities of Virginia, and the independent cities of Baltimore, St. Louis, and Carson City.

[Top]

2. Why do I only see 10 results on the web though it seems that my request should contain many more records?
The ten results displayed on the web represent the records with the highest property damage of your request. It is just an example of how the requested data will look when displayed as a spread sheet. The downloadable subset will include all records that match your search.

[Top]

3. Why can't I find a certain event in the database?
The database only includes those events that generated more than $50,000 in either crop or property damage between 1960 and 1995. If the event you are looking for generated less damage, you will not be able to locate it in this database. Sometimes, however, damage information was just not available from the utilized data sources even if the event caused high monetary losses. For information on the data sources, please see the metadata section.

Please consider, that the event may have affected several counties. Consequently, you need to locate all these counties by searching the database by date and hazard event in order to get a loss estimation for a certain event.

[Top]

4. Will my request contain double records if I search for two or more different hazard types?
No, events composed of multiple hazard types will not be listed more than once in your data request. However, you will receive double, triple etc. records when querying the database separately for each hazard type. For instance, if you query the database for Floods, you will receive any record related to flooding. If you query the database again for a different hazard type such as Severe Thunderstorms, the database will not know that you would like to omit flood events based on your previous query. Instead, the database will return all thunderstorm-related events including combined thunderstorm and flood events. To avoid double records, please query the database simultaneously for all the hazard types that you are interested in.

[Top]

5. What data sources have been used to compile the database?
The main data sources were "Storm Data and Unusual Weather Phenomena" by the National Climatic Data Center, information from the National Geophysical Data Center, and the Storm Prediction Center. Please see the metadata section for more detailed information.

[Top]

6. Does the database contain information on Puerto Rico, Guam, or other U.S. territories?
No, the database refers to U.S. counties only.

[Top]

7. What do the different columns of my downloaded .txt file represent?
The columns represent information on hazard events (beginning and ending), hazard type(s), spatial information (FIPS code, county, state, and sometimes additional local remarks), and damage information (crop and property damage, fatalities, injuries). Please see the metadata section for more detailed information.

Example: If you are searching for drought events in South Carolina between August 1, 1990 and August 1, 1995, the database will return an event that will be visualized on our website as following:

The corresponding download file will come in a delimited text format (pipe or tab) including a header and looks the following: HAZARD ID | HAZARD_BEGIN_DATE | HAZARD_END_DATE | HAZARD_TYPE_COMBO | NAME | POSTAL_CODE | FIPS_CODE | FATALITIES | INJURIES | PROPERTY_DAMAGE | CROP_DAMAGE | LOCATION | REMARKS 96371 | 05/01/1995 | 05/31/1995 | Abbeville | SC| 45001 | 0.00 | 0.00 | 0.00 | 434782.61 | DROUGHT

[Top]

8. Why can't I download the complete database?
Download restrictions are required due to limited server capacities. For more information contact the Hazards & Vulnerability Research Institute.

[Top]

9. How can I map downloaded data?
The easiest way to map the downloaded information is by generating a spreadsheet depending on the software you are using (e.g. .dbf for ESRI ArcView and ArcGIS) and then linking it with spatial information through the FIPS code. Such spatial information may be already provided by your software package. However, by using such default data you won't be able to map data prior to 1982 as the number of counties is different. Spatial information (shapefiles) are available from our products section.

[Top]

10. What does a record represent?
Each record from 1990 trough 1995 refers to a hazard event affecting a county and generating total losses higher than $50,000 of either property or crop damage. For instance, a thunderstorm event affecting Richland and Lexington County in South Carolina and causing property damages of $50,000 will be entered in the database as an event affecting Richland County with $25,000 and Lexington County with $25,000 worth of damage. However, a thunderstorm event causing $5,000 in Richland County alone will not be included as the damage does not exceed the threshold value of $50,000.

Note, from 1960 to 1989 and 1995 on, every event listed in NCDC's storm data set that had exact damage figures assigned was entered into the database. This methodology modification was necessary as NCDC changed its reporting procedure in the course of 1995. Originally, NCDC had classified damages into logarithmic categories such as $0-50, $50-500, $500-5000, $50,000-500,000, $5,000,000-50,000,000. In addition, the spatial resolution of the reports was low, i.e. most frequently damages were documented for larger regions instead of singular counties. In the course of 1995, NCDC started reporting exact dollar figures such as $126,000 and the resolution increased as well, meaning damages were immediately assigned to a specific county if possible. Thus, many counties missed the $50,000 threshold from 1995 trough 2000. Major damages in a state would be lost had we maintained the threshold, hence the decision to adjust the methodology.

[Top]

11. What is a FIPS code?
The Federal Information Processing Standard (FIPS) provides the names and codes that represent the counties and other entities treated as equivalent legal and/or statistical subdivisions of the 50 States, the District of Columbia, and the possessions and freely associated areas of the United States. The last three digits of the FIPS code refer to the county whereas the preceding one or two digits identify the state. Changes of FIPS codes are published by the National Institute of Standards and Technology (NIST) after approval by the Secretary of Commerce. Questions concerning the list of entities and their assigned codes are to be addressed to the Maintenance Agency: Office of the Chief, Geography Division, Bureau of the Census, Washington, DC 20233.

For quick access to FIPS code/county name changes go to the metadata section and to the products section.

[Top]

12. How can I find out to what state a certain county belongs?
The last three digits of the FIPS code identify the county whereas the preceding one or two digits refer to the state. The following list provides the state identifiers.

FIPS Code State Name
02 ALASKA
01 ALABAMA
05 ARKANSAS
60 AMERICAN SAMOA
04 ARIZONA
06 CALIFORNIA
08 COLORADO
09 CONNECTICUT
11 DISTRICT OF COLUMBIA
10 DELAWARE
12 FLORIDA
13 GEORGIA
66 GUAM
15 HAWAII
19 IOWA
16 IDAHO
17 ILLINOIS
18 INDIANA
20 KANSAS
21 KENTUCKY
22 LOUISIANA
25 MASSACHUSETTS
24 MARYLAND
23 MAINE
26 MICHIGAN
27 MINNESOTA
29 MISSOURI
28 MISSISSIPPI
30 MONTANA
37 NORTH CAROLINA
38 NORTH DAKOTA
31 NEBRASKA
33 NEW HAMPSHIRE
34 NEW JERSEY
35 NEW MEXICO
32 NEVADA
36 NEW YORK
39 OHIO
40 OKLAHOMA
41 OREGON
42 PENNSYLVANIA
72 PUERTO RICO
44 RHODE ISLAND
45 SOUTH CAROLINA
46 SOUTH DAKOTA
47 TENNESSEE
48 TEXAS
49 UTAH
51 VIRGINIA
60 VIRGIN ISLANDS
50 VERMONT
53 WASHINGTON
55 WISCONSIN
54 WEST VIRGINIA
56 WYOMING

[Top]

13. Can I use Excel to manipulate and analyze the data?
Yes, you can import the downloaded file without any problem into Excel (see FAQ #14). However, there are two things to consider:

1. Excel does not handle more than 65,536 records (rows). Often enough, it does not notify you if records were discarded. Hence, either make a note of how many records you downloaded or view the data in a different program such as SPSS, ArcView or ArcGIS.

2. It is easy to mix up records in Excel! Selecting and sorting by columns only instead of selecting the whole dataset and then sorting, results in a totally disordered and worthless dataset. The following tables provide you with an example of what happens when sorting the data incorrectly:

Step 1: Importing the data

Beginning Date County State Hazard Type Property Damage
5/15/1990 Richland SC Thunderstorm 1,250.00
9/27/1984 Pinellas FL Hurricane 50,000,000.00
1/31/1978 Los Angeles CA Wildfire 1,000,000.00

Step 2a: Incorrect sorting by property damage only

Beginning Date County State Hazard Type  Property Damage
5/15/1990 Richland SC Thunderstorm 1,250.00
9/27/1984 Pinellas FL Hurricane 50,000,000.00
1/31/1978 Los Angeles CA Wildfire 1,000,000.00

Step 3a: Incorrect output

Beginning Date County State Hazard Type  Property Damage
5/15/1990 Richland SC Thunderstorm 50,000,000.00
9/27/1984 Pinellas FL Hurricane 1,000,000.00
1/31/1978 Los Angeles CA Wildfire 1,250.00

Step 2b: Correct sorting by property damage

Beginning Date County State Hazard Type Property Damage
5/15/1990 Richland SC Thunderstorm 1,250.00
9/27/1984 Pinellas FL Hurricane 50,000,000.00
1/31/1978 Los Angeles CA Wildfire 1,000,000.00

Step 3b: Correct output

Beginning Date County State Hazard Type  Property Damage
9/27/1984 Pinellas FL Hurricane 50,000,000.00
1/31/1978 Los Angeles CA Wildfire 1,000,000.00
5/15/1990 Richland SC Thunderstorm 1,250.00

[Top]

14. How do I import the data into Excel?
In Excel, go to the file menu and Open... Make sure that in the "Files of Type" drop-down list you have selected "All Files" in order to be able to see your downloaded .txt file. Opening your .txt file will prompt the Text Import Wizard window. Chose delimited and start the import at row 1. Click on the Next button. Select Tab or Others - enter | if Others - depending on what type of delimitation you have chosen. If the data preview looks reasonable proceed to the next step. Leave the column data format at General and click the Finish button. Now, you should see the column titles in the first row and every record should appear in a separate cell.

[Top]

15. Why are there injuries and fatalities with decimal places?
Often casualties and damage information are listed without sufficient spatial reference. For instance, the damage caused by a singular natural hazard could be listed as:

Delaware (statewide) - January 20, 1988 - Snow Storm - 1 fatalities - 6 injuries - $100,000 property damage

In order to assign the damage amount to a specific county, the fatalities, injuries and dollar losses need to be divided by the number of counties affected from this event. In the snow storm example provided above, the losses would be split between Delaware's three counties as the hazard had affected the whole state. Thus, the event would enter the database as:

Kent - January 20, 1988 - Snow Storm - 0.33 fatality - 2 injuries - $33,333.33 property damage

New Castle - January 20, 1988 - Snow Storm - 0.33 fatality - 2 injuries - $33,333.33 property damage

Sussex - January 20, 1988 - Snow Storm - 0.33 fatality - 2 injuries - $33,333.33 property damage

[Top]

16. Can I compare dollar losses across years?
Yes, our database now has a function that adjusts losses for inflation, allowing for comparability between years.

[Top]

17. There are numerous cases where the damage figures are far less than $50,000. Is this only due to the fact that damages have been divided based on the geography or have more specific sources been used?
The National Climate Date Center (NCDC) has changed its reporting procedures in 1995. During this year both categorical as well as exact dollar losses have been reported by NCDC. Thus, the majority of the records from 1995 onwards are exact damage figures that have been reported as such by NCDC and that have not undergone any processing by us (exemption: events affecting multiple counties).

In addition, NCDC has also improved its spatial reporting system. Instead of reporting affected regions and an associated damage figure that we would have distributed across the affected counties, NCDC has moved on to reporting every single county and its associated damage separately.

Thus from 1960-1989, we have included EVERY event that caused property or crop damages. This change in our methodology was necessary due to NCDC's change in reporting. Consequently, you will find many small damage figures like $500, $1000, etc. in these events.

Hence, from 1990 - 1995 we have only selected events with property or crop damage higher than $50,000 (equals NCDCs logarithmic category 5=$50,000 to $500,000), whereas from 1995 onwards we have included all property or crop damage-causing events reported in NCDC's Storm Data publications.

[Top]

18. Why can't I join the SHELDUS™ attribute table with a spatial table via FIPS codes?
The most likely reason for this problem is that the FIPS code columns are formatted differently. The SHELDUS™ FIPS code column is formatted as number.

[Top]

19. Why are county names sometimes incorrect in Alaska?
The following Alaskan counties may experience incorrect county names after downloading data from SHELDUS™: Aleutians (East/West, Anchorage, Bethel, Haines, Outer Ketchikan, Ketchikan Gateway, and Upper Yukon. Mismatches between FIPS codes and county names occur mostly in requests for historic information.

In order to reassign the correct county name either:

a) use GIS to join the downloaded dataset to a historic shapefile (available in the Products section) via the FIPS code. Please make sure to use the shapefile matching your time period.

b) use Excel, Access, SPSS or other software products that can handle large datasets and change the county names manually. County names and changes from 1960-2000 are listed here.

[Top]

20. Why are losses for Hurricane Katrina much smaller than other official loss estimates?
The data on Hurricane Katrina are current as of December 2005 as reported by the National Climate Data Center (NCDC). SHELDUS™ relies on NCDC data and we did not consider government reports (e.g. FEMA, GAO), newspaper information or any other sources to supplement NCDC's Storm Data publications. NCDC plans to update loss figures for Hurricane Katrina in the future. Once these updates become available, they will be incorporated into SHELDUS™. Similar is true for other large-scale events.

[Top]

21. What is the GLIDE number?
The GLIDE number is a internationally recognized global identifier for large-scale events. The GLIDE number is fairly young initiative. Between 2002 and 2003, GLIDE numbers were issued for events that are included in the international hazard database (EM-DAT) maintained by the Centre for Research on the Epidemiology of Disasters (CRED) of the University of Louvain in Brussels (Belgium). Since 2004, GLIDE numbers are issued by members of the GLIDE initiative and upon request. SHELDUS™ does not include all U.S.-related GLIDE numbers nor is there a GLIDE number for every large-scale U.S. event. SHELDUS™ includes only GLIDE numbers for selected major events, given a GLIDE number exists. For more information see www.glidenumber.net

[Top]

22. What is a presidential disaster declaration?
FEMA issues numbers for presidential disaster declarations by state. For every state, FEMA lists so-called designated counties affected by the event. SHELDUS™ does not include all PDDs. SHELDUS™ includes only PDDs for selected major events. For more information on presidential disaster declarations see http://www.fema.gov/news/disasters.fema.

[Top]

23. Why do losses in my county include coastal hazards when my county is not a coastal county?
Because of inaccuracies in the spatial identification of hazard events within NCDC Storm Data reports, there are instances within the SHELDUS™ database where inland counties will contain property and/or crop losses from coastal hazard events. This issue mainly occurs when an event is labeled as a “statewide” event by NCDC in which each county receives an equal percentage of losses irrespective of spatial location or hazard event type. Hence, a hazard event identified as “flooding, winter weather, coastal storm” which is given a statewide locational identifier is treated as an event which impacted each county equally. Therefore, inland counties may (at times) be attributed with events and losses that seem incapable of occurring.

[Top]

24. How are coastal hazards defined?
Coastal hazards not specifically referencing a major event type (such as a hurricane or flood) sometimes fall into this more broad category. The coastal category can sometimes include losses pertaining to flooding, severe storm, or hurricane events but more often includes other coastal hazards referenced as high tide, rouge tide, erosion, high seas, heavy surf, etc. We advise users to look at coastal hazards for any events that may fall within either your temporal frame of interest or your hazard specific interest as additional losses from such events may be allocated to this category. For more information on hazards that are classified as coastal in SHELDUS please see this document: Coastal Event Types

[Top]

25. How do you categorize data?
Generally, data is easily parsed into one of the HVRI 18 hazard categories based on the NCDC hazard classification. We do not change or modify the classifications created by NCDC but rather aggregate multiple hazard type classifications into a more broad hazard event class including the following event types: 1 - Avalanche 2 - Coastal 3 - Drought 4 - Earthquake 5 - Flooding 6 - Fog 7 - Hail 8 - Heat 9 - Hurricane/Tropical Storm 10 - Landslide 11 - Lightning 12 - Severe Storm/Thunder Storm 13 - Tornado 14 - Tsunami/Seiche 15 - Volcano 16 - Wildfire 17 - Wind 18 - Winter Weather. To learn more about NCDC hazard classifications please visit our metadata page and search for NCDC Storm Data Preparation: SHELDUS Metadata Page