Drought and Fire

Numerous commentators have blamed the number of acres burned in recent years on (your choice) increased fuels from past fire suppression, increased fuels from timber cutting, and environmentalist obstructions to fuel treatments. But a close look at the data reveal that the main factor responsible for fires today is drought.

The National Oceanic and Atmospheric Administration (NOAA) has data showing the percentage of the U.S. that is "severely or extremely dry" by month since 1900. These numbers can be compared with the number of acres burned in wildfires each year since 1960.

Such a comparison reveals a strong correlation between the amount of drought in July, August, and September and the number of acres burned that year. Calculating r, the correlation coefficient, reveals that drought is responsible for half the variation in acres burned from year to year. When examined on a decade-by-decade basis, drought is responsible for 98 percent of the variation in acres burned in each decade from the 1950s through the 1990s.

On the average, the number of acres burned each year (in millions) is about half the percentage of the U.S. that is under severe or extreme drought. The year 2002 was the droughtiest year on record (by this measure) since 1956, with 36 percent of the U.S. under severe or extreme drought. The second droughtiest year was 2000, at 34 percent. Thus, the surprise is not that so many acres burned those years but that so few acres burned: about 7 to 8 million instead of the 17 to 18 million that would be expected if the average of past years had been met.

Those who are interested can download a Excel data file with all available data on drought and acres burned. The following information can be found in the file:

• The raw "dry and wet" data (percentage of U.S. that is severely or extremely dry or wet) are in columns A through D. (These columns are in an invisible pane that you can access by choosing "unfreeze panes."
• Column E shows the year, column F the average percent dry in that year, column G the average percent wet in that year.
• Column H is the acres burned in that year (available only since 1960).
• Columns I through T is the percent dry in the months of January through December of each year.
• Column U is the average of August and September, column V is the average of July August and September.
• Columns W through AH is the percent wet in the months of January through December (which aren't evaluated further).
• Row 106 shows r-squareds for each column against the number of acres burned (so are for 1960 through 2002). The highest r-squareds are for August and September; July is a bit lower.
• Cells Q107:W117 show acres burned and drought indices (Aug-Sep in column U, July-Aug-Sep in column V) by decade. R-squareds are shown for decades in U119:V122. W119:X122 are the square roots of the r-squareds for the Aug-Sep and July-August-September indices.

The "correlation coefficient" (r) is the square root of r-squared, so an r-squared of .25 indicates a 50 percent correlation. That is, one of the variables explains about half of the variation in the other variable.

Here are a few points that are worthy of note or further study:

• The decade-by-decade correlation is strongest for the 1950s through the 1990s.
• When the 1940s are added, the correlation falls dramatically: more acres burned than would be expected based on the drought index. One possible reason is that World War II depleted the firefighting labor force and led to more escaped fires than usual.
• Adding the 1920s and 1930s brings the correlation back up. The 1930s turn out to be, by far, the droughtiest decade of the twentieth century and also the decade that saw the most acres burned in wildfires.
• Looking at individual years, 2002, 2000, and 1988 were the droughtiest years since 1960 and also three the four years when the most acres burned.
• The year-by-year correlation isn't perfect, however. NOAA recorded nearly zero drought in 1969, yet 1969 saw the fifth most acres burned since 1960. Anyone who remembers the 1969 fire season would be welcome to contact Randal O'Toole with some ideas on why so many acres burned that year.

Anyone else with comments is welcome to email me as well.