In mid-2014 the National Drought Mitigation Center began providing estimates of the number of people living in areas impacted by drought. Below is an explanation of how these numbers are calculated, and how they should be interpreted.
In order to calculate the number of people living in areas that the USDM classifies as being in drought, the first step was to determine a data source. In this case, the population numbers were taken from the 2010 U.S. Census, as those data are standardized and freely available. It was also decided that population statistics would be calculated for smaller areas and aggregated to determine the values for larger areas. In this case counties were used as the smallest area of calculation because they were able to be aggregated into larger areas of analysis such as states.
The population density for each county was determined by dividing the population of the county by the area of the county in square miles.
Then to compute population in drought statistics each week, this population density value is multiplied by the percent of the area in a given USDM category to determine the number of people living in the area of each USDM category. This is done for each county in the U.S. (including Alaska, Hawaii and Puerto Rico) that is in drought. Once the county-level statistics are calculated, the numbers are aggregated up to the state level by summing the values for each county within the state. The state values are then aggregated to the national level.
Due to the fact that there is quite a bit of estimation throughout the process, these numbers are not exact values. There are several assumptions made in order to simplify the calculation process, necessary to ensure that the numbers can be available when the map gets released each week. A few points are mentioned below:
- We assume that population is distributed evenly across each county, although in reality, it isn’t.
- The population of an area is not static for long periods of time. Population is fairly fluid and real-time population numbers for an area are constantly changing. For the purpose of these calculations, it is assumed that the population of an area has not changed since the last census was taken.
- The USDM boundaries are considered “fuzzy.” For example, this means that the magnitude of the drought does not necessary change drastically along the boundary between D2 and D3. These areas should be considered more of a transition zone. The observations and experiences of people in an area labeled D2 may not differ meaningfully from the experiences of people in an area labeled D3.
- Drought can affect people living outside of the areas it is directly affecting. For example, a drought in a major agricultural region might increase food prices even in areas that aren’t in drought. And a lack of mountain snow can affect the water supply for cities hundreds of miles away.
How to Use the Numbers
Any population figures provided should be considered a rough estimate. The numbers are presented with a high degree of precision due to the calculation process, but should be considered artificially precise. It is up to the user to determine how much rounding should be necessary for a given purpose. For example, instead of using an exact figure of “1,026,493 people” it would be better to say “more than a million people.”
Due to the way the numbers are calculated, there are also instances where the numbers don’t seem logical. For example, it might be possible that the calculations determine that 5 people are living in the area of a USDM category. Again, this is simply due to the way that the numbers are calculated and should be interpreted that the USDM category in question is basically non-existent in the area.
The purpose of these numbers is to give an idea of the magnitude of the number of people living in a drought-stricken area. A drought that covers an area with a population of 15 million people is most likely going to have a different impact than a drought that covers an area of 500 people.
There are several potential improvements under consideration for the process. One of the questions that remains to be answered is how to deal with population changes over time. The U.S. Census only occurs every 10 years and population can change drastically on a much shorter time scale. It may be possible to use other population data that are collected more frequently, such as population estimates made between the decennial censuses.
Along those same lines, there is also the question of how to deal with historical information. Older versions of the U.S. Census and other population data may be used to re-calculate values for dates prior to 2010.
Currently the population statistics are limited to areas based on counties. However, population data for non-conforming areas such as watersheds would be very useful. Future research is being conducted on the best method to estimate population for these areas.