Why Data Source Matters More Than the Answer

Attorneys and claims professionals often arrive at a weather case with a specific question: Was it raining? How fast was the wind? Was there ice? Those questions have answers. But the answers are only as strong as the data behind them. Each data source measures something specific, from a specific vantage point, with a specific margin of error.

A forensic meteorologist’s job is not to say “yes it was raining” and move on. The job is to identify which instruments were operating near the incident, what those instruments physically measured, how well those measurements represent conditions at the incident location, and where the gaps are. That documented chain from raw data to conclusion is what makes weather evidence defensible.

What follows is a plain-English guide to four major data types: what each one measures, what it cannot measure, and what a competent forensic analysis does with each.

Doppler Radar: Storm Structure and Precipitation Estimates

The U.S. WSR-88D NEXRAD network covers most of the contiguous United States. It is one of the most useful tools in forensic meteorology, and one of the most commonly mischaracterized.

Here is what the radar physically measures: reflectivity (the energy returned by precipitation targets aloft) and radial velocity (the component of target motion toward or away from the radar beam). From those measurements, meteorologists can draw conclusions about storm structure, precipitation presence, approximate precipitation intensity, and radial wind patterns within the storm.

Here is what it does not measure directly:

• The full wind vector at the surface. Radial velocity is one component of motion in one direction from a fixed point. Reconstructing a complete wind field requires additional analysis and, where available, corroborating surface observations.
• Rainfall at the ground. Reflectivity is measured aloft, then converted to a rainfall-rate estimate using an empirical relationship called Z-R. That relationship varies by storm type, drop size distribution, and precipitation regime. The result is a useful estimate, but not a gauge reading.
• Near-surface conditions at long range. Radar beam elevation increases with distance. At 40 nautical miles from the radar, the center of the lowest-tilt beam is typically 3,000 to 4,000 feet above ground level. The data describe the atmosphere at altitude, not at the surface.

Dual-polarization technology, standard on U.S. NEXRAD radars since 2013, improves precipitation-type discrimination and reduces some estimation errors. It does not eliminate the physical constraints above.

What a competent analysis does: Use radar to establish storm timing, spatial extent, approximate intensity, and wind patterns. Then corroborate the surface-level conclusions with ASOS or COOP station data. Label any radar-derived rainfall figure as an estimate and note the relevant limitations for the specific site geometry.

Surface Station Data: Point Observations With Defined Sampling

ASOS (Automated Surface Observing System) stations are the backbone of surface observation in the United States. They produce METARs (standardized aviation weather reports) at least once per hour and more frequently during changing conditions. For forensic purposes, they provide time-stamped, archived, authenticated records that can be obtained directly from NOAA’s National Centers for Environmental Information (NCEI).

Two technical details matter more than almost anything else when using this data in litigation:

COOP (Cooperative Observer Program) stations provide daily precipitation totals, typically from a 24-hour accumulation window ending at 0700 local time. They extend the observation network significantly, particularly in rural areas, but their daily resolution cannot establish whether rain fell at the specific hour of an incident. Hourly ASOS data or radar provides that temporal resolution.

Satellite Imagery: Context, Not Surface Evidence

GOES-East and GOES-West provide continuous imagery of the continental United States. This data is valuable for establishing the large-scale storm environment: cloud-top temperatures, moisture patterns, storm evolution, and timing.

It is not surface evidence. Satellite measures radiances at the top of the atmosphere: visible reflectance during daylight, and infrared brightness temperature that reflects cloud-top emission, not ground conditions. A geostationary satellite cannot observe surface precipitation, road-surface temperature, or ground-level wind speed.

One technical detail that matters when spatial precision is important: parallax displacement. A tall thunderstorm cloud may appear shifted from its actual ground position in geostationary imagery because the sensor views the cloud top at an angle rather than directly overhead. At moderate distances from the satellite’s subsatellite point, tall cloud features can appear displaced by several miles. When an analysis involves a specific address or a short road segment, any satellite imagery used to argue storm position should account for this effect.

What a competent analysis does: Use satellite to document storm evolution, timing, and synoptic context. Rely on surface stations and radar (not satellite) as the primary evidence for surface conditions at the incident location.

Reanalysis and Gridded Products: Filling Gaps, Not Replacing Observations

When no surface station exists within a reasonable distance of the incident, a common situation in rural Montana, the Nevada high desert, or parts of the Southern Appalachians, forensic meteorologists may use reanalysis products to estimate conditions. ERA5 (ECMWF), NARR (NOAA NCEP), and similar products blend model output with assimilated observations to produce gridded, physically consistent reconstructions of the atmosphere at regular time steps.

These products are scientifically credible tools. They are not observations. Three points matter for litigation:

• Spatial resolution has limits. ERA5’s standard grid is approximately 28 km (about 17 miles). A feature smaller than that grid, such as a localized convective downburst, a narrow fog layer, or a valley drainage wind, may not appear in the gridded output even if it occurred.
• They carry inherent uncertainty. Reanalysis products are model-observation blends. Ensemble-based versions provide uncertainty estimates explicitly. A single gridded value reported without a confidence range understates what the product itself acknowledges about its own precision.
• Their role is corroboration and context. Where surface stations exist, they take precedence. Reanalysis fills the gap and provides synoptic context. When a report relies primarily on reanalysis for a site-specific wind or precipitation claim, the confidence rating should reflect that reliance: Medium or Low, not High.

How Authentication Works and Why Source Matters

Weather data is not self-authenticating. For it to support expert testimony, the proponent must establish that the records are what they purport to be, retrieved from a reliable custodian, and unaltered from the original.

For NOAA-archived data, NCEI operates a formal data certification service that produces documentation specifically designed to meet authentication requirements for court records. This is a direct, reliable path. When data is certified through NCEI, the question of where it came from and whether it is unaltered has a documented answer.

Data does not have to originate from the National Weather Service to be admissible. State agricultural mesonets, FAA aviation networks, university research stations, and properly maintained private sensor systems can all support expert analysis when source documentation, metadata, and chain-of-custody records are preserved. The evidentiary question is reliability and authentication, not the specific issuing agency.

On the expert opinion side, federal courts apply Federal Rule of Evidence 702 under Daubert v. Merrell Dow Pharmaceuticals and Kumho Tire v. Carmichael, requiring that the testimony rest on sufficient facts, reliable methods, and reliable application to the case. Several states apply a Frye general-acceptance standard or a hybrid. Attorneys should confirm which standard governs in their jurisdiction before retaining a weather expert witness.

What Weather Evidence Establishes and Where the Line Is

A forensic meteorologist reconstructs atmospheric conditions. The analysis answers: What were the weather conditions at or near this location and time? Were those conditions consistent with icing, high winds, reduced visibility, or flooding? How confident is that conclusion, and why?

The meteorology does not establish whether a specific individual acted reasonably, whether a property was adequately maintained, or whether weather caused a specific incident. Those are legal and factual determinations that belong to the fact-finder. The weather expert provides the atmospheric context that informs those determinations.

This boundary is not a weakness. It is the source of the opinion’s durability. A meteorologist who stays within the atmospheric record and states clear, data-supported conclusions is far more useful to counsel and far more resilient than one who overclaims.

Frequently Asked Questions

Quick Recap

• Radar, satellite, surface stations, and reanalysis data each measure something specific and different. A competent forensic analysis uses the right source for each conclusion and states the limits of each.
• ASOS wind data is a 2-minute average at a fixed, sited point. Representativeness to the incident location is always a required analysis step.
• Weather evidence establishes atmospheric conditions. The meteorologist does not determine liability, breach of duty, or causation. Those are determinations for the fact-finder.

Technical Appendix

Chain of Custody