Our examination of MODIS-detected fires supports Butry and Thomas’ hypothesis that few of these incidents are reported in NFIRS.
Butry and Thomas assumed that MODIS observations were wildfires and those that occurred in local direct protection areas should be recorded in NFIRS as wildfires. Since NFIRS does not have an incident type for "wildfires," Butry and Thomas developed criteria using incident type, actions taken, area of origin and suppression factors to identify NFIRS fires that were candidates for MODIS matching and matched them based on the county and date. They found that 32% of the MODIS observations occurred on the same date and in the same county as a fire from their NFIRS subset.
USFA analysis used geocoded NFIRS data and looked for any NFIRS incident (including unauthorized and prescribed fires) and found that only 1.6% of MODIS incidents occurred within 1 kilometer and 24 hours of any NFIRS incident. A summary is provided in Table 10.
Table 10: Comparison of Approaches and Conclusions
| Factor | Butry/Thomas analysis | This analysis |
|---|---|---|
| Timeframe | 2003-2014 | 2017 |
| MODIS observations used | Only those located in California Local Direct Protection Areas | Entire U.S. |
| Consolidated multiple MODIS observations to single incidents | Not done | Grouped MODIS observations within 24 hours and 1 kilometer |
| NFIRS Incidents used for comparison |
| Incident Type in 100 series or 561, 631, 632 |
| MODIS-NFIRS matching technique | Whether a fire department in the same county as a MODIS observation reported an NFIRS incident (meeting criteria above) on the same day | Whether a MODIS observation occurred within 24 hours and 1 kilometer of an NFIRS incident |
| Conclusion | 32% of MODIS observations are reported in NFIRS | 1.6% of MODIS incidents (1.1% of MODIS observations) are reported in NFIRS |
We distinguish between individual MODIS observations and MODIS incidents by grouping all MODIS observations that occurred within 1 kilometer and 24 hours together as a single incident. Slightly over half of MODIS observations (52%) can be grouped with another observation to create an incident, which suggests that MODIS can be useful for identifying fires with large burn areas and lengthy durations, likely large named wildfires.
We also found that 6.4% of MODIS incidents occur within national parks and national forests, grasslands, and wilderness areas. However, the vast majority of MODIS incidents involve a single observation and most of these are not reported in NFIRS. Additional research should be done to determine whether these are fires in public or nonpublic lands, instances of crop or trash burning, or false positives.