Precipitation and wildfire analysis of Reno, Nevada

I downloaded gridmet using code to pull files from a url where the files were listed directly, referencing published code (Ahmadalipour 2022). I downloaded the daily precipitation from 1979 to 2022. Once I determined that I only had fire perimeters through 2020, I restricted the precipitation dataset to 2020 as well. The historical fires data set I downloaded from the blm website as a shapefile (Bureau of Land Management). Refer to the figure below for a visualization of the wildfire dataset restricted to 1979- 2020 and the Reno area with the roads shown as well to illustrate the developed extent. The roads dataset is from the Washoe County GIS portal.

I calculated the area burned by year by using the group by function on the discovery year for the fire attributes. One complication of the data is that there was not a fire every year for the time span. In order to make the data comparable to the average annual rainfall, I appended a “0” to every year that was missing in the historical wildfire dataset. I found the average annual rainfall over the area by averaging with the dimension latitude and longitude.

This shows the annual rainfall per year and the total burned area in acres per year, not including the added zeroes. This visualization is encouraging as it shows some evidence of the relationship found in Pilliod et al. (2017) in which a proportionally large amount of total acres burned follows a proportionally high rainfall year.

To assess the relationship between one year’s rainfall and the total burned acres for the following year, which is the relationship observed in Pilliod et al. (2017), I used the same average annual rainfall averaged over the reno area per year. To create a lagged total burned area dataset, I added a 0 to the end and dropped the first value in order to shift the values forward one year, to compare 1979 average rainfall with 1980 total burned area in acres and so on. This relationship is much more compelling, with a slope of 26.182 and an r-squared value of 0.2351, as seen in the figure below. The p-value of 0.001 indicates that this correlation is significant.