Ground truth is a direct surface or in-situ observation used to check what a radar signal, satellite image, model field, or forecast was showing at a given moment. It confirms what actually happened on the ground.
Why it matters
Remote sensing tools infer weather from proxies. Radar measures reflected energy and Doppler velocity, not raindrops and wind directly. Satellite measures radiances, not visibility. Models produce fields based on assumptions. Each of these tools has strengths and limits, and none of them are trusted without observations to anchor them.
Ground truth closes the loop. A confirmed hail measurement lets a forecaster relate the observed size to the dual-pol signatures overhead. A measured wind gust matched to a bowing radar signature builds calibration. A confirmed tornado tied to a specific rotation couplet gives forecasters and researchers a case they can learn from.
Where it comes from
Trained SKYWARN spotters are the largest source, and their observations flow into Local Storm Reports and post-event Storm Data. Additional ground truth comes from mesonet and ASOS stations, cooperative observers, law enforcement, emergency management, chasers, media, and post-event damage surveys performed by NWS office teams.
Because a good report requires accurate timing and location, offices ask for a clear observation, the time of the event, a location referenced to a road or landmark, and a brief description. The NWS spotter guidance stresses reporting what has happened, not what a spotter thinks is about to happen.
How forecasters use it
During warnings, ground truth influences the next warning decision. A confirmed tornado can upgrade a Tornado Warning to a Tornado Emergency and expand the warning polygon. A confirmed wind or hail report can support extending or reissuing a Severe Thunderstorm Warning downstream.
After events, ground truth verifies the forecast. Storm Data, storm surveys, and archived reports feed the climatology and the performance metrics of the warning system. That is how the NWS knows whether a warning verified and how researchers separate signal from noise in future case studies.
Important limits
Ground truth is only as reliable as the observer, the timing, and the location. A misidentified feature, a report with the wrong time, or a location off by miles can mislead everyone downstream. When in doubt, forecasters weight ground truth against what the radar and other data are showing rather than accepting a report outright.