Fire and Drought in the United States

I recently stumbled across statistics for the total acreage in the United States consumed by fire each year since 1960.  Graphed over time, the data look as follows:

Here I have expressed the acreage burned as a percentage of the total area of the United States. Notice the pattern of fire decreasing through the 1960s and 1970s and then increasing fitfully from there.

That reminded me loosely of the PDSI data for the US:

Note how the US became wetter after the 1960s and then drier after the 1990s.  And of course, it makes  sense that drier conditions would lead to more and bigger fires.

To make a closer comparison I took the PDSI data for the contiguous US averaged over June-August and then made a scatter plot of the fire area versus PDSI:

As you can see - there's definitely a relationship with around three times more burned in the driest years than the wettest years.  However, it's far from the only thing going on with substantial scatter away from any simple relationship.  Probably much of the variation is caused by regional fluctuations - one half the country very dry and the other half very wet will result in a lot more fires than everywhere uniformly normal, but the national average PDSI will be about the same.  The PDSI may also not capture the fire-relevant sense of "drought" perfectly.

The exponential trendline shown had a bit better fit than a straight line (R² of 33% vs 28%) but I wouldn't necessarily have high confidence in extrapolating it very far.  Still it seems clear that if the US continues to dry out, there will be a lot more acreage burned in fires.

A note for the especially statistically cautious or skeptical amongst you; the original data source for the fire burned area warns us as follows:

Prior to 1983, sources of these figures are not known, or cannot be confirmed, and were not derived from the current situation reporting process. As a result the figures above prior to 1983 shouldn't be compared to later data.

To assess this, I also tried the plot and fits with the pre-1983 and post-1983 data separately:

The fitted trendlines are pretty similar so I feel comfortable that, whatever issues in reporting there may have been prior to 1983, it's fine to combine all the data for this purpose.  Note that the post-1983 data have four years more dry than any year in 1960-1982 and these were all bad fire years.

This also tends to rule out the idea that the upsurge in fires in the last decade or two is primarily due to changes in fuel load because of past fire suppression.  That would show up as a shift upwards in the fire/drought relationship with more fire for the same amount of drought in the post 1983 data: that doesn't seem to be the case (the blue line is slightly steeper than the red one, but only slightly and there's no evidence of upward shift overall).

Finally, it's worth musing on the implications of this picture:

Climate models suggest that in thirty years the new normal in the US will be about PDSI of -4 on the historical scale.  So if we extrapolate the relationship from earlier, and figure that in 30 years we'll be running from -9 to +1, instead of -5 to +5, then we'd be in the region of the graph either side of the red line:

Again, this assumes the extrapolation is valid, and it's probably only roughly so.  There will be a lot more fires.  Not raging inferno across the whole country, but something like 50-75% more fire area burned per year than during past decades (with large variations between good and bad years, as now).

Of course, if we don't start making much more serious efforts, climate change won't be stopping or slowing down in the 2030s, and we'll be heading further and further off the left side of that graph as the century goes on.