the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Mar 2026
CONFEX: A Database for CONUS Fire EXtent
Abstract. This article presents the CONUS Fire EXtent (CONFEX) database. CONFEX is based on the VIIRS S-NPP 375 m active fire product and provides wildfire perimeters, centroids, ignition locations, and start and end dates for 2012–2024 across CONUS and Alaska. The algorithm extracts hotspot detections from VIIRS S-NPP, clusters them into fire events using a spatio-temporal DBSCAN approach and derives a perimeter polygon and centroid for each event, assigning event-level attributes. When evaluated against MTBS using regional area-threshold filters, the algorithm obtained area-based F1 scores of 67 % (western CONUS; >1000 acres), 77 % (Alaska; >1000 acres), and 26 % (eastern CONUS; >500 acres); and when evaluated against California FRAP (excluding the 2020 tuning year), an area-based F1 of 69 % was found, demonstrating the ability to aggregate VIIRS hotspots into spatially and temporally coherent fire events. To our knowledge, CONFEX is the first publicly available, moderately high-resolution wildfire-extent dataset developed for the CONUS and Alaska regions using the VIIRS S-NPP 375 m active fire product. The database provides a valuable resource for researchers to study fire regimes in CONUS and Alaska.
- Preprint
(1667 KB) - Metadata XML
-
Supplement
(418 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2026-116', Anonymous Referee #1, 10 Apr 2026
-
AC1: 'Reply on RC1', Raja Zubair Zahoor Qadiri, 21 Apr 2026
Dear Reviewer 1,
Please find attached our detailed response to your comments. We sincerely thank you for your careful review and constructive feedback, which helped us improve the clarity, structure, and scientific presentation of the manuscript.
Sincerely,
Zubair Qadiri
-
AC1: 'Reply on RC1', Raja Zubair Zahoor Qadiri, 21 Apr 2026
- RC2: 'Comment on essd-2026-116', Anonymous Referee #2, 11 Apr 2026
- RC3: 'Comment on essd-2026-116', Anonymous Referee #3, 21 Apr 2026
-
RC4: 'Comment on essd-2026-116', Anonymous Referee #4, 21 Apr 2026
Qadiri and Cerrai present a new fire event dataset for the continental US (CONUS) and Alaska using clusters of active fire detections from the Suomi-NPP VIIRS instrument. The CONFEX dataset provides an estimated size and duration of individual fire events on an annual basis, with corrections to the estimated area of active fire clusters to remove areas without burnable fuels. CONFEX provides a set of attributes derived from the clustered fire detections for each fire perimeter, including start date and location, end date, duration, and size. Data from the Monitoring Trends in Burn Severity (MTBS) dataset were used to provide initial validation of large fire events in the overlapping time period (2012-2023).
The paper provides summary statistics for initial comparisons with MTBS data, but the limited information on matched and unmatched fires does not provide sufficient context for the potential strengths or limitations of the CONFEX data compared to MTBS. For example, the processing approach for CONFEX is retrospective, and the time lags for data production are not specified, such that it is unclear what the potential advantage of CONFEX would be for fires that also appear in the MTBS record. Further, the dataset provides a small subset of characteristics for each event, with few derived estimates of fire behavior compared to existing databases like FIRED, FEDS, or the Global Fire Atlas. What do clusters represent in this analysis, and do they provide further insights into periods of growth or other patterns? What potential uses are there for the specific attributes associated with each CONFEX cluster?
In addition, the descriptive material in the Results and Supplement lacks depth and context. What drives the differences between CONFEX and MTBS data by region, by state, or by year? How would one interpret the increase in fire events in VIIRS for specific seasons or land cover types relative to material in MTBS? What might be the advantages of CONFEX in these circumstances? These concepts are not clearly explored or explained.
Overall, the clustering approach provides a reasonable reproduction of final fire extent as estimated from MTBS in Alaska and the western US, with more mixed performance in eastern CONUS where fires tend to be smaller. Without further investigation of specific strengths and limitations of the CONFEX dataset, it is not clear that this methodology or dataset provides a substantial advance in our understanding of individual fire events across CONUS or Alaska.
Additional Comments:
- The introduction is somewhat generic, and does not provide a clear motivation for the development of CONFEX relative to other VIIRS-based products, or for active fire detection based products over burned area products. FIRED is already publicly available over the full extent of this study.
- Methods: the full history of active fire detection by satellite is not needed.
- Methods: the calibration of the clustering approach (2000 m, 48 h) uses FRAP data from 2020. Does this approach work universally across the three primary study regions (Alaska, Western US, Eastern US)? Existing databases use much longer time intervals to account for observational constraints and episodic changes in fire behavior (FIRED 11 days (from burned area), FEDS 5 days). What impact does this shorter clustering interval have on fragmentation of fire events in CONFEX?
- Methodology issues: Clustering vs. tracking. Very low “object” scores compared to area scores indicates an issue with the matching methodology, likely due to the overly strict 2 day threshold. This is discussed around L320: “Taken together, these results demonstrate that CONFEX adds value to FRAP by providing detection-inferred ignition location and characterization of fires with complex growth behavior at substantially higher temporal resolution than incident-based perimeter products”... is CONFEX supposed to be a final fire perimeter dataset, or a fire progression dataset? Inconsistent throughout.
- It is unclear why the Short et al. 2022 (FPA-FOD) is mentioned, as it does not support the analysis and is explicitly intended as a point occurrence dataset. The authors may wish to consider NIFC’s WFIGS fire perimeters dataset as a reference for the “best available” manually collected fire perimeters.
- The paper lacks clear comparisons with the methods and results from similar recent efforts. Several recent products use DBSCAN–how does this inform the selection of this methodology for CONFEX, and what, if any, differences are there in the implementation or outcomes for the CONUS domain? Likewise, given that various perimeter delineation methods have been proposed and evaluated in depth (e.g. Bhuian et al. 2024), some justification of the choice of the alpha hull approach would be appreciated.
- Conceptually, or analytically, how does CONFEX compare to existing data (FEDS, FIRED)? Given that previous work (e.g., Chen et al. 2022) has published open source code for fire event tracking using VIIRS active fire detections, why not use an existing algorithm to develop a dataset covering this time period and spatial extent? It is not clear from the manuscript why the authors chose to develop a new algorithm rather than applying an existing one.
Citation: https://doi.org/10.5194/essd-2026-116-RC4
Data sets
CONFEX: CONUS and Alaska Fire EXtent database Raja Zubair Zahoor Qadiri and Diego Cerrai https://data.mendeley.com/datasets/sk6jwy7xmg/3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 127 | 32 | 10 | 169 | 20 | 11 | 8 |
- HTML: 127
- PDF: 32
- XML: 10
- Total: 169
- Supplement: 20
- BibTeX: 11
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
pls see attached pdf