Seattle Weather Radar How Realtime Data Is Transforming Local Forecasts And Emergency Preparedness
There's a NEXRAD station at Skagit Regional Airport, about 60 miles north of downtown Seattle. That's the KHGX site, part of the WSR-88D network, and it's the primary source of radar coverage for the Puget Sound region. The data comes in volume scan products every four to six minutes, and if you know where to look, it tells you more than the standard GOES satellite loop most people watch on the news. Most people see a radar image and interpret the colors without understanding what they're looking at. Base reflectivity is the simplest product — it shows where precipitation exists and roughly how intense it is. Green means light rain, yellow moderate, red heavy. But that's only the surface-level read. The real useful information comes from storm-relative velocity and differential reflectivity, which aren't shown in most consumer weather apps. I spent years working emergency management in King County, and one of the first things I learned was that the standard radar display dramatically underestimates precipitation in the urban core. The terrain around Seattle causes significant beam blockage. When the radar beam hits the Olympic Mountains to the west or the Cascades to the east, you get partial to complete shadow zones. The Skagit station handles this better than older installations because of its elevation and positioning, but the south end of the Sound — Tacoma, parts of Bellevue — still sits in a decent amount of ground clutter and blocking depending on the scan angle.
The workaround I used involved pulling the S-PolQ data. That's the mobile C-band radar the University of Washington operates out of Shoreline. It's not as powerful as the NEXRAD for long-range detection, but for the immediate Seattle metro area it gives you much finer resolution at lower altitudes. Where NEXRAD's lowest scan might be 3,000 feet above ground level after 60 miles, S-PolQ is scanning right over your head at a few hundred feet. When we had flash flooding concerns in June 2021 along the I-90 corridor, the NEXRAD wasn't showing the convective initiation until it was already over 10,000 feet MSL and too late to act on. S-PolQ caught the development 20 minutes earlier because it was looking at the layer where the moisture actually was.
What the velocity data actually tells you
Storm-relative velocity mode flips the colors — red for precipitation moving away from the radar, green for precipitation moving toward it. This sounds like basic stuff, but the important thing is the couplet. A tight red-green couplet side by side indicates rotation. In the Seattle area most of us think about tornado risk, but the more common threat is microbursts from thunderstorms. The 2015 July microburst event in South Lake Union destroyed temporary structures and knocked down trees along the waterfront. The velocity data showed the divergence pattern about three minutes before the gust front hit. NWS Seattle issued a warning based on that, which was unusually fast for the area. Base velocity alone is tricky though. If you're not in storm-relative mode, the entire field looks noisy and hard to interpret. The radial velocity component means precipitation moving perpendicular to the radar beam appears nearly stationary on the display, which can mask important rotational signatures. Always check that the product is in storm-relative mode before making any operational decisions based on it.
Get the Full Details
Dealing with the bright band problem
Seattle gets a lot of stratiform rain, and that's where the bright band becomes a major issue. When snow melts into rain about 8,000 to 10,000 feet up, the coating of water makes the ice crystals extremely reflective. The radar interprets this as heavy precipitation at the surface, but what's actually falling is a steady light drizzle. I've seen base reflectivity values of 45 to 50 dBZ in the bright band over the Sound and then zero actual accumulation on the gauge below. It's misleading on every level. The fix is using the dual-pol products. Differential reflectivity, or Zdr, shows the ratio of horizontal to vertical reflectivity. Raindrops are flattened spheres — they show positive Zdr. Hail is tumbling and roughly spherical, so Zdr drops near zero. Snowflakes are dendritic and elongated, giving high Zdr values at low reflectivity. By cross-referencing reflectivity with Zdr, you can separate the bright band echo from real precipitation and adjust your forecasts accordingly. The NWS now offers corrected reflectivity products that attempt to filter out the bright band, but they're not always reliable at the range distances covering the metro area.
Where the system fails completely
I need to be straightforward about the limitations because relying on radar alone has caused problems. The first is range. The Skagit radar's effective range for precipitation detection is about 120 miles. Beyond that, the beam is so high above the ground that light precipitation simply isn't sampled. When storms develop over the Cascades or the Olympic Peninsula, you're largely flying blind until they move into the coverage area. I've had situations where afternoon convection built up in the foothills and moved west into Snohomish County without triggering any radar signature until it was already producing hail the size of quarters. The second failure mode is attenuation. C-band radars, which is what S-PolQ uses, suffer signal loss when precipitation is heavy. A strong rain core can shadow what's behind it, creating artificial gaps in the data. This matters most during the heavy Pacific winter storms when you need the most accurate picture. NEXRAD is S-band and doesn't attenuate as badly, but it still happens during intense convective situations. Third, there's the lag time. Even at the fastest volume scan rate, you're getting data that's roughly four minutes old. For real-time emergency response — flash flooding, severe wind events — that delay is meaningful. By the time you see the echo on your screen, the event has already progressed several kilometers. This is why we layered in rain gauge networks and automated weather stations for the actual decision-making. Radar tells you where the precipitation is; ground truth tells you what's happening right now.
Accessing the data yourself
The NWS makes their radar products available through various feeds. The most accessible route is the NOAA Weather Radar Lab data portal, which provides raw and derived products in standard formats. You can subscribe to the WMO-compliant data streams that feed into systems like WeatherBug Enterprise or the National Digital Forecast Database. For the S-PolQ data, the UW homepage maintains an archive and near-real-time feed, though it requires some familiarity with meteorological data formats to work with directly. If you're setting up an automated monitoring system for emergency preparedness, I'd recommend starting with the NWS API for KHGX base reflectivity and velocity, then layering in S-PolQ for the high-resolution metro coverage when the mobile unit is deployed. The two data sources combined give you better situational awareness than either alone, but you need to account for the different spatial resolutions and update rates. NEXRAD updates every four to six minutes. S-PolQ can do volumetric scans in two to three minutes when actively tracking a storm.
How emergency operations has changed
The shift from manual interpretation to automated detection algorithms has been the biggest change in my experience. Ten years ago, a meteorologist would watch the radar loop and call out developments. Now the NWS has automated nowcasting tools that flag severe weather signatures in near real-time and push alerts directly to emergency coordinators. The Severe Weather Information Center aggregates radar, satellite, and surface data to produce unified threat assessments that feed into our county EOC systems. What hasn't changed is the judgment call. Algorithms miss context. They don't understand that a particular valley in Kitsap County funnels wind in a way that amplifies downdrafts, or that a specific street in Georgetown floods when the radar shows more than half an inch per hour for twenty minutes. The radar data improved the speed and accuracy of alerts significantly, but the people who actually make the decisions — whether to evacuate a neighborhood, close a road, or activate flood pumps — still need the ground-level knowledge that no radar can provide.
