Weather radar is remarkably good at detecting where precipitation is falling, but its estimates of how much rain or snow is hitting the ground carry meaningful error. The raw radar signal can differ from what a rain gauge on the ground measures by 30% to 50% or more, depending on distance from the radar station, the type of storm, and how the data is processed. Understanding why helps you interpret those colorful radar maps with a more realistic eye.
How Radar Turns Signals Into Rain
Weather radar works by sending out pulses of microwave energy and measuring what bounces back. The strength of the return signal, called reflectivity, tells forecasters something is out there. But converting that signal strength into an actual rainfall rate requires a mathematical formula with built-in assumptions about drop size and storm type.
The National Weather Service uses different conversion formulas for different conditions. For widespread, steady rain (stratiform storms), one set of coefficients is applied. For intense summer thunderstorms with large drops, a different set is used. The problem is that real storms rarely fit neatly into one category. A single storm can contain regions of light drizzle, heavy convective rain, and mixed precipitation all at once. Choosing the wrong formula for a given patch of sky can easily skew the rainfall estimate by 30% or more in either direction.
Resolution: What the Radar Actually “Sees”
The U.S. NEXRAD radar network, which includes about 160 stations, scans the sky in slices. With the upgraded Super Resolution mode, each data point covers a 250-meter-deep slice along the beam’s path and a 0.5-degree wedge of the sky. That 250-meter range resolution stays constant whether you’re 50 km or 200 km from the station.
The catch is the beam’s width. Because radar energy spreads out like a flashlight, the area illuminated by each pulse grows with distance. At 50 km from the station, a single beam slice might sample a patch of sky a few hundred meters across. At 200 km, that same slice covers well over a kilometer. Everything within that volume gets averaged into one reading. A small but intense thunderstorm cell far from the radar can be blended with the calm air around it, making it look weaker and broader than it really is.
The beam also rises higher above the ground with distance. Close to the station, the radar samples precipitation near the surface. At 200 km, it’s sampling the atmosphere thousands of feet up. Rain can evaporate before reaching the ground, or light precipitation aloft can intensify as it falls. What the radar “sees” at altitude may not match what’s happening at street level.
Why Radar and Rain Gauges Disagree
When researchers compare radar rainfall estimates to ground-level rain gauges, the two rarely match perfectly. But a surprising amount of that disagreement isn’t the radar’s fault. A study published in the Journal of Applied Meteorology and Climatology found that 50% to 80% of the discrepancy between radar and gauge measurements for hourly totals comes from what’s called “gauge representative error.” A rain gauge captures rainfall at a single point, while the radar averages over a grid cell roughly 3 km across. Rain simply doesn’t fall uniformly over that area, so the gauge and the radar are measuring slightly different things.
Even after accounting for this, about 30% of the apparent radar uncertainty genuinely stems from the gauge’s limitations as a reference standard rather than from radar error. This doesn’t mean radar is perfectly accurate. It means the true error is smaller than a naive comparison suggests. In practice, radar rainfall estimates for a single hour at a single location can be off by 30% to 50%, but when averaged over larger areas or longer time periods, accuracy improves considerably.
False Echoes and Ghost Rain
Sometimes radar shows precipitation where none exists. The most common cause is anomalous propagation, which happens when a temperature inversion forms near the surface. Normally, air temperature drops with altitude, and the radar beam curves gently upward and away from the ground. When a layer of warm air sits on top of cooler, moist air near the surface, the beam bends downward instead, bouncing off buildings, hills, or the ground itself. The radar interprets these returns as precipitation.
This phenomenon, often called “AP” by meteorologists, is most common on clear nights when the ground cools quickly and an inversion develops. You might check a radar app on a calm evening and see scattered green splotches that look like light rain. If the sky is clear outside your window, anomalous propagation is the likely explanation. Modern radar processing filters out much of this clutter automatically, but some still slips through, especially near the station where ground returns are strongest.
How Heavy Rain Weakens the Signal
Radar signals lose energy as they pass through precipitation, a problem called attenuation. The severity depends on the radar’s operating frequency. The NEXRAD network uses S-band radar, which is relatively resistant to this effect. During heavy rainfall of about 60 mm per hour (roughly 2.4 inches), S-band signals lose about 1.5 decibels per kilometer of heavy rain they travel through. That’s noticeable but manageable.
Smaller, more portable radars operating at higher frequencies suffer more. C-band radars lose about 3 decibels per kilometer under the same conditions, and X-band radars lose around 5 decibels per kilometer. This means a heavy storm close to an X-band radar can effectively block the radar’s view of anything behind it. The result is an underestimate or complete blindness to precipitation on the far side of a heavy storm cell. Dual-polarization technology, now standard on NEXRAD stations, helps correct for some of this attenuation by comparing signals sent in horizontal and vertical orientations.
What Dual-Polarization Changed
Before 2013, NEXRAD radar sent and received pulses only in the horizontal plane. The upgrade to dual-polarization added a vertical pulse, letting the radar measure the shape of whatever is reflecting the signal. Raindrops flatten as they fall, snowflakes tumble irregularly, and hail tends to be round or jagged. By comparing horizontal and vertical returns, the radar can distinguish rain from snow, identify hail, and even detect debris lofted by tornadoes.
For rainfall estimation, dual-pol data provides a second, independent way to calculate how much rain is falling. This cross-check helps correct errors from the traditional reflectivity-based approach, particularly in heavy rain where the standard formulas tend to overestimate, and in mixed precipitation where they struggle the most. The improvement is most dramatic for localized heavy rain events, where traditional methods could miss the true intensity by a wide margin.
Scan Speed and Fast-Moving Storms
Current NEXRAD stations use a mechanically rotating antenna that takes 5 to 10 minutes to complete a full volume scan of the atmosphere. For slow-moving frontal systems, that’s fine. For rapidly evolving thunderstorms or tornadoes, a lot can change in 5 minutes. A tornado can form, touch down, and shift direction between scans.
Phased array radar, which steers its beam electronically rather than physically rotating, can complete a full volume scan in about 1 minute. Some experimental designs operating at higher frequencies have achieved scans in as little as 20 seconds. The National Weather Service has been testing phased array technology as a potential replacement for the aging NEXRAD network. Faster scans won’t change how precisely the radar measures reflectivity, but they will give forecasters a much more current picture of what storms are doing right now, which matters enormously for tornado and severe thunderstorm warnings.
What This Means for Your Radar App
The radar image on your phone has already been through several layers of processing: raw reflectivity data converted to estimated rainfall, ground clutter filtered out, multiple radar stations stitched together, and the image smoothed for display. Each step introduces small decisions that affect what you see. Some apps apply their own algorithms on top of the National Weather Service data, which can make images look sharper or more detailed than the underlying data actually supports.
As a practical guide, trust radar most for answering “is it raining over there?” and treat it with more skepticism for “how hard is it raining?” Radar is excellent at showing the location, movement, and general structure of storms. It’s less reliable at pinpointing exact rainfall totals at a specific spot, especially far from a radar station. If you’re 150 km or more from the nearest NEXRAD site, the radar is sampling the atmosphere well above your head, and what it reports may not match what you see out the window.
Light rain and drizzle are particularly hard for radar to detect at distance, since small droplets return very weak signals that fade quickly with range. Snow also reflects radar energy much less efficiently than rain, so winter precipitation often appears lighter on radar than it actually is until the flakes begin to melt on the way down, creating a bright band of enhanced reflectivity that can look misleadingly intense.