What do you do when you hear a rumble of thunder? Storm chasers would dart out the door, camera in-hand, but most people — including those not in the weather business — would check the radar to see what storm is lurking around the corner. We load our apps and click links to see nearby storms without giving the technology behind it a second thought. Radar is a supreme convenience if you want to know whether you need an umbrella for the commute, but it also serves as vital, life-saving technology to the millions of people threatened by severe weather every year.

We’re fortunate to have a robust network of Doppler radars in the U.S., but that spaced-out array is slowly aging and prone to failures at inopportune times. We desperately need to invest in expanding our network of weather radars before its flaws become a serious issue.

A world without weather radar would be a terrifying place to live, and we’re lucky the technology came into use when it did. Like so many invention stories, using radar to track the weather came about by accident — soldiers scanning the skies for enemy aircraft during World War II instead found a sky full of rain and snow showing up on their screens, a discovery that sparked a meteorological revolution.

The widespread use of weather radar began in the 1950s and 1960s, allowing forecasters to look at the intensity and expanse of precipitation, greatly increasing our knowledge of life-threatening storms like supercells. Recent hardware upgrades allow us to use radar to see just about anything you can think of: precipitation, wind, tornado debris, swarms of bugs, flocks of birds, airplanes, cold fronts, mountain peaks, wind turbines, traffic and even the sun’s rays. We’ve come a long way in our ability to watch storms bubble up on the horizon (or gross out at the bugs swarming overhead), but there’s always room for improvement.

Our network of weather radars in the United States — called NEXRAD, or Next Generation Radar — consists of 160 sites that cover all 50 states, Puerto Rico, Guam and a couple of military bases in Asia and the Azores. This invaluable coverage is able to detect precipitation over just about every populated area in the U.S. that isn’t obstructed by terrain, and most heavily-populated areas have low-level radar coverage.

The low-level coverage in itself is the reason we should expand our radar network. We may have figured out how to keep tabs on Mother Nature, but Earth gets the last laugh — the curvature of our planet is a radar’s Achilles heel. When the beam is sent out from the antenna, the lowest scan is conducted at about a 0.5-degree angle from the horizon, so the elevation of the beam gradually gets higher in the sky the farther it travels from the radar site. For instance, the lowest beam sent out by the radar in Sterling is higher when it passes over the National Mall — scanning the sky at around 1,500 feet — than it is when passes over Centreville, Va., where the beam passes overhead at around 500 feet.

Until we can figure out how to conquer the curvature of the Earth, we need to install more radar sites to close the wide gaps in coverage and stave off future issues presented by downtime. The above map from the National Centers for Environmental Information shows a rough approximation of low-level radar coverage from each site in the lower 48 east of the Rocky Mountains. When you factor in coverage into the mid-levels of the atmosphere — higher than 6,000 feet — almost everyone is covered to some extent, but when you whittle it down to just low-level coverage, the gaps in our network are glaring and pretty unsettling.

Two years ago, I wrote here at the Capital Weather Gang about the worst gap in our radar network, which exists over central North Carolina and puts cities like Charlotte and Greensboro in a vulnerable position.  The four nearest sites — Greenville, S.C.; Columbia, S.C.; Raleigh, N.C., and Blacksburg, Va. — provide some coverage for the area, but not much at that crucial level close to the ground.

This lack of near-ground service can cause meteorologists to miss tornadoes, especially quick ones that don’t form as the result of a supercell. The issue came to a head on March 3, 2012, when an EF-2 tornado touched down near Charlotte without warning, damaging hundreds of buildings and injuring four people. The bulk of the rotation in the storm occurred beneath the radar beam, causing it to go unnoticed until reports of damage filtered in.

There are also quite a few areas in tornado-prone areas like the Central Plains and Southeast that have precious little coverage near the ground, including portions of central Alabama, much of Missouri and eastern Texas, as well as a large chunk of real estate in southeastern Kansas. It’s pretty easy to see the rotation in one of those huge supercells that explodes over the Plains—so missing a tornado isn’t as big of a problem there, but confirming the strength of the rotation and possible debris is a huge benefit of low-level coverage out in tornado country.

The existing gaps are bad enough without any other issues coming into play. Radars have an ugly habit of breaking down, and it’s dangerous when they decide to quit working when there’s bad weather on the horizon.

Fortunately, the National Weather Service is actively investing time and money into the radar network, “replacing obsolete and high-failure equipment with state-of-the-art technology that will improve system reliability, data quality and add new science from the research community while sustaining the network through at least 2030,” says Terry Clark, acting director of the National Weather Service’s Radar Operations Center in Norman, Okla. That much-needed deployment is set to begin nationwide next year.

Outside of technical failures that can be mitigated, there are numerous instances of radars getting knocked offline due to severe weather itself. The radar site in Mobile, Alabama, was knocked offline after a direct lightning strike on April 29, 2014, in the middle of a historic flash flood disaster that plagued southern Alabama and the Florida Panhandle, just when they needed that critical low-level coverage the most. On top of that, within the past two years, tornadoes have passed within spitting distance of the radars covering Jackson, Mississippi, and Kansas City, Missouri—any wiggle in the storm’s path and those radars likely would have disappeared.

In addition to addressing the hardware failures, strategically installing new radar sites around the country would both solidify the network by protecting communities from being heavily affected by downtime during severe weather, as well as filling in the gaps that exist during normal operations. It all sounds good on paper, but the task won’t be easy.

The newest radar in the network sits in Langley Hill, Washington, keeping watch over storms roaring ashore from the Pacific. However, it came about after nearly two decades of lobbying and wrangling in Congress to secure funds to get it up and running. And while not officially part of the NEXRAD network, the National Weather Service will soon have access to a new Doppler radar going up at the University of Missouri in Columbia, filling a significant gap in the center of the Show Me State.

Cost is huge factor, and it’ll be a tough sell to convince officials to fund such an ambitious endeavor. According to a 2008 study by the National Research Council, one new radar site would cost about $10 million to purchase and install, and about $500,000 per year to maintain. Even without considering inflation or increased costs, $10 million is a lot of money, and funding doesn’t come easy in Washington.

In a perfect world, we would have dozens more sites to ensure that there are absolutely no gaps in coverage, but reality is far from perfect, and that’s an impossible task to achieve. Short of letting private companies pick up the tab—this tornado warning brought to you by Conglomocorp!—it’s going to be a challenge getting one more radar, let alone as many as are needed.

This raises the big question: who gets their hands on some of these big, new, expensive radar sites? Above is a map showing fifteen possible sites for a new radar, complete with a rough approximation of the extent of their low-level coverage. Central North Carolina is an obvious choice for the top of the list, followed behind by dead zones in Virginia, Pennsylvania, Louisiana, and a big swath of eastern Texas. Many of these theoretical sites would close significant gaps in areas with hundreds of thousands of inhabitants—however, it remains to be seen whether politicians and decision-makers would support the idea, given the high costs associated.

In the meantime, the Weather Service has utilized existing technology to improve forecast lead times in severe weather situations. In addition to the Nexrad radars that the Weather Service operates, it also relies heavily on radars around airports, operated by the Federal Aviation Administration. The additional data in combination with new scanning improvements that allow forecasters to rapidly scan low levels of the atmosphere have enabled forecasters to better detect things like tornadoes, hail and downbursts.

Still, the critical gaps in our valuable network of weather radars are too great to ignore; unfortunately, many people don’t notice them until one site goes down at the wrong time, or forecasters miss a tornado because the rotation quite literally slipped under the radar. The entire mission of the National Weather Service is to protect life and property, and there’s no better way to accomplish that goal than to give forecasters the additional eyes in the sky they need to easily and efficiently spot hazards without worrying about the limits of our existing network.