Downloading and cleaning Monitoring Trends in Burn Severity (MTBS) fire data

Requirements

Downloading and processing the data requires that you have R installed, and the following R packages:

dplyr
rgdal
rgeos
tidyr

If you do not have these packages, you can try to install them with install.packages(), e.g.,

install.packages('tidyr')

The script get-mtbs-data.R pulls data from the internet, so you'll need an internet connection.

Running the script

To run the R script, you can use Rscript from the terminal:

Rscript get-mtbs-data.R

Or, if you prefer, you can open the file in R or Rstudio and run it interactively.

What it produces

The script will produce a directory with the following structure:

data/
  - processed/
    - fire-events.csv
    - fire-summary.csv
  - raw/
    - mtbs_fod_pts_data/
      - ...
    - us_eco_l3/
      - ...

Event data

The file fire-events.csv has the raw data from MTBS where each row is a fire event:

fire_id	year	month	day	lat	long	acres_burned	na_l3name	na_l2name	na_l1name	yearmonth	mean_potential_et	total_precip	max_air_temp	mean_wind_speed
AL3038308812219980404	1998	4	4	30.383	-88.122	705	Southern Coastal Plain	MISSISSIPPI ALLUVIAL AND SOUTHEAST USA COASTAL PLAINS	EASTERN TEMPERATE FORESTS	1998-04	4.757972	50.62457	299.6400	4.281175
AL3041008830120050916	2005	9	16	30.410	-88.301	2914	Southern Coastal Plain	MISSISSIPPI ALLUVIAL AND SOUTHEAST USA COASTAL PLAINS	EASTERN TEMPERATE FORESTS	2005-09	4.952767	90.42460	305.1668	3.975596
AL3067208833720050207	2005	2	7	30.672	-88.337	696	Southeastern Plains	SOUTHEASTERN USA PLAINS	EASTERN TEMPERATE FORESTS	2005-02	2.275852	110.09175	289.2758	4.016748
AL3105208777220050211	2005	2	11	31.052	-87.772	7334	Southeastern Plains	SOUTHEASTERN USA PLAINS	EASTERN TEMPERATE FORESTS	2005-02	2.275852	110.09175	289.2758	4.016748
AL3110108837320050218	2005	2	18	31.101	-88.373	757	Southeastern Plains	SOUTHEASTERN USA PLAINS	EASTERN TEMPERATE FORESTS	2005-02	2.275852	110.09175	289.2758	4.016748

The columns are as follows:

fire_id: unique event identifier
year, month, day: year, month, day of ignition
lat, long: lat and long of ignition
acres_burned: number of acres burned
na_l3name, na_l2name, na_l1name: level 3, 2, and 1 ecoregion names
yearmonth: year and month concatenated (e.g., '2010-01')

Both of these files have preprocessed MACA climate data summarized at the level of month and ecoregion, with four variables: mean daily potential evapotranspiration (mean_potential_et), total precipitation (total_precip), mean daily maximum air temperature (max_air_temp), and mean daily wind speed (mean_wind_speed).

Summary data

The file fire-summary.csv has the number of fires and total burn area summarized for each of the 85 EPA level 3 ecoregions and 382 months over the range of dates in the MTBS data (for a total of 382 X 85 = 32470 rows), where each row is an ecoregion by month combination.

yearmonth	na_l3name	year	month	mean_potential_et	total_precip	max_air_temp	mean_wind_speed
1984-02	Acadian Plains and Hills	1984	2	1.1805210	116.813903	275.4773	4.962357
1984-02	Arizona/New Mexico Mountains	1984	2	2.7397637	3.008436	285.1002	3.876059
1984-02	Arizona/New Mexico Plateau	1984	2	2.2535006	3.420493	283.5001	3.200126
1984-02	Arkansas Valley	1984	2	2.5848122	94.998374	288.3537	4.530387
1984-02	Aspen Parkland/Northern Glaciated Plains	1984	2	0.8037277	15.325724	274.3547	3.577555

Computing polygon centroids and distance matrices

library(rgdal)
library(rgeos)

# load ecoregion shapefile
ecoregion_file <- list.files(pattern = 'us_eco_l3.shp$',
                             recursive = TRUE,
                             full.names = TRUE)
ecoregions <- readOGR(ecoregion_file)

Computing centroids

Next, we'll use the gCentroid function from the rgeos package to compute centroids for each polygon in the ecoregion shapefile:

centroids <- gCentroid(ecoregions, byid = TRUE)

Last, because some ecoregions are discontinuous, a subset of our ecoregions now have multiple centroids. To get just one centroid per ecoregion, we can average across all of the centroids for those ecoregions with multiple centroids:

library(dplyr)

ecoregion_centroids <- as.data.frame(centroids) %>%
  tbl_df %>%
  mutate(na_l3name = ecoregions$NA_L3NAME) %>%
  group_by(na_l3name) %>%
  summarize(x = mean(x), 
            y = mean(y))

This will return a data frame with the centroids for each ecoregion. As a final step, you may need to compute a distance matrix, which provides pairwise distances among all centroids.

Computing a distance matrix

D <- ecoregion_centroids %>%
  select(x, y) %>%
  dist %>%
  as.matrix

# associate ecoregions with rows and columns
rownames(D) <- ecoregion_centroids$na_l3name
colnames(D) <- ecoregion_centroids$na_l3name

# preview part of D
D[1:3, 1:3]

	Acadian Plains and Hills	Arizona/New Mexico Mountains	Arizona/New Mexico Plateau
Acadian Plains and Hills	0	3443179.1	3514975.6
Arizona/New Mexico Mountains	3443179	0.0	200954.1
Arizona/New Mexico Plateau	3514976	200954.1	0.0

mbjoseph / mtbs-data