NASA FIRMS MODIS Active Fire Central America 24h

Rows: 308
Columns: 13
Data range: 10.0513 to 9.9773
Auto-refresh: Re-checks its source every 30 days
Last refreshed with new data: 2026-10-30

Regional active fire time-series for Central America.

AI analysis

Dataset Analysis: NASA FIRMS MODIS Active Fire — Central America (24h)

1. What this dataset represents and its value for correlation studies

This dataset is a near-real-time snapshot of satellite-detected active fire pixels across Central America, sourced directly from NASA's Fire Information for Resource Management System (FIRMS) API, specifically the MODIS Collection 6.1 product. The source URL (firms.modaps.eosdis.nasa.gov) confirms this is an authoritative, operationally-maintained government feed rather than a static or curated dataset — the "24h" designation and narrow time coverage window (spanning essentially one diurnal cycle, reflected in only 2 distinct acqdate values) indicate this is a rolling snapshot rather than a historical archive. This makes it highly valuable for near-real-time correlation work — e.g., linking fire detections to weather conditions, drought indices, land-use change, or air quality readings on the same day — but less suited for long-term trend analysis without repeated ingestion over time.

2. Data quality observations

Data quality appears strong on the surface: zero null cells across all 580 rows and 13 columns, and no columns show disproportionate missingness. However, duplicate-row detection is flagged as pending ("Phase D — recomputed against Parquet"), which is worth flagging as an open verification item — FIRMS feeds are known to occasionally produce overlapping detections from adjacent satellite passes, so duplicate-row confirmation should be prioritized before using this for aggregate fire-count analysis. The version column has only 1 distinct value, meaning it carries no analytical signal and should be excluded from modeling. Similarly, satellite and daynight each have only 2 distinct values, appropriate for categorical/dummy encoding.

3. Key columns and their distributions

The frp (fire radiative power) column is the most analytically interesting: heavily right-skewed (skew=4.74) with mean (22.7) far exceeding median (11.77), Q1/Q3 of 7.63/21.12, and 65 outliers — consistent with the expected long tail where a small number of intense fires dominate energy output. Brightness (mean 318.35, median 316.23, skew=1.78, 29 outliers) and scan (skew=2.65, 77 outliers) show similar right-skew patterns typical of thermal anomaly detection data, where most pixels register modest signal and a minority represent large or intense burns. Confidence is left-skewed (skew=-0.525, mean 65.5, median 67) with a solid IQR (53–80), suggesting most detections are moderate-to-high confidence, useful for filtering noise. Geographically, latitude (7.1–33.4°) and longitude (-115.9 to -62.0°) confirm full regional coverage from southern Mexico/Central America down toward northern South America, with a slight rightward longitude skew (0.251) suggesting denser detections toward the western part of the range.

4. Recommended join key columns

latitude and longitude (both flagged as Join Keys) are the natural spatial join keys for merging with other geospatial datasets — ideally combined with acqdate/acqtime as temporal keys to enable spatiotemporal joins. For coarser aggregation (e.g., joining to country- or region-level datasets), consider deriving a rounded lat/lon grid or reverse-geocoded admin region, since raw coordinates at 580 distinct points are too granular for many external datasets.

5. Suggested pairings for correlation discovery

Strong candidates include: (a) meteorological data (temperature, humidity, wind speed, precipitation/drought indices like SPI or the Keetch-Byram Drought Index) to test fire-weather relationships; (b) land cover/deforestation datasets (e.g., Global Forest Watch, Hansen tree-cover loss) to distinguish agricultural burning from wildfire; (c) air quality/PM2.5 monitoring data downwind of fire clusters; (d) socioeconomic or agricultural calendar data, since Central American fire activity is often tied to slash-and-burn agricultural cycles; and (e) longer-term NASA FIRMS historical archives (VIIRS or MODIS multi-year) to contextualize this 24h snapshot against seasonal/historical baselines.

Columns

latitude (decimal)
longitude (decimal)
brightness (decimal)
scan (decimal)
track (decimal)
acq_date (date)
acq_time (integer)
satellite (string)
confidence (integer)
version (string)
bright_t31 (decimal)
frp (decimal)
daynight (string)