Every roofer has had the same argument. A homeowner swears their neighborhood got hit. The map shows a bright band two streets over and nothing on their block. Somebody is wrong, and it matters who — because one answer means a roof and the other means an hour you will not get back.
Hail maps are the single most useful tool in storm restoration and the single most misread. They are not damage reports. They are not measurements. They are a weather model's best guess at what a storm was capable of dropping, drawn at a resolution fine enough to route a crew and coarse enough to be wrong about one house.
This guide covers what you are actually looking at, what each color means for a roof, where the map lies, and how to turn it into a canvassing route.
What a hail map is actually made of
Almost every hail map a roofer will ever use — free or paid, ours included — traces back to the same public source: NOAA's radar network, processed through a system called MRMS, or Multi-Radar Multi-Sensor.
Here is the chain, in plain terms.
A weather radar sends out a pulse and listens for what comes back. Big, dense, irregular objects — hailstones — throw back far more energy than raindrops. MRMS stitches together every radar in the country onto one grid roughly a kilometer across, updated every couple of minutes, and looks specifically at how much of that strong return is sitting above the freezing level in the storm. Hail lives up there. Rain does not.
From that, an algorithm produces a number called MESH — Maximum Estimated Size of Hail. That number is what almost every color band on almost every hail map represents.
Three things follow from that, and they are the three things people get wrong:
- The map describes the storm, not the roof. MESH says "this storm was capable of producing a stone about this big over this grid square." Whether that stone actually reached the ground at that spot, and at what angle, and into what condition of shingle, is a separate question.
- The unit of measurement is a grid square, not an address. A kilometer is roughly ten to fifteen residential lots. When a map colors a square, it is making one claim about all of them.
- It is a maximum, not an average. MESH estimates the largest stone the storm could produce there. Most of what fell was smaller.
None of that makes the map unreliable. It makes it a prospecting instrument. Used as one, it is extraordinarily good.
The size bands, and what each one means for a roof
Hail maps are colored by estimated stone size, usually in quarter-inch steps. Roofers describe hail in coins and sports equipment; meteorologists describe it in inches. Here is the translation, plus what each size tends to mean in the field.
| Size | Common name | NWS severe? | What it typically means for a roof |
|---|---|---|---|
| 0.75″ | Penny | No (was, before 2010) | Rarely functional damage on asphalt. Can mark soft metals and screens. Worth noting, not worth a crew. |
| 1.00″ | Quarter | Yes | The severe threshold. Enough to bruise aging or three-tab shingles. New architectural shingle often survives it. |
| 1.25″ | Half dollar | Yes | Damage becomes common on worn roofs. Gutters and vents start showing dents you can photograph. |
| 1.50″ | Ping pong ball | Yes | Most asphalt roofs of any age start taking functional damage. This is where canvassing gets easy. |
| 1.75″ | Golf ball | Yes | Functional damage is the norm. Soft metals are obviously dented. Adjusters expect to find something. |
| 2.00″+ | Hen egg and up | Yes | Widespread, obvious damage. Total losses are common. Expect the whole market to be canvassed within days. |
Two notes on that table.
The severe threshold moved. The National Weather Service raised the definition of severe hail from three-quarters of an inch to a full inch in January 2010. Older data, older articles, and some older tools still key off the smaller number. If you are looking at anything historical, check which threshold it uses before you compare it to a modern map.
Size is not the whole story. A one-inch stone driven sideways by seventy-mile-an-hour wind does more to a west-facing slope than a still-air 1.5-inch stone does to anything. Wind-driven hail concentrates damage on one or two elevations, which is why a map showing "only" an inch over a neighborhood that also took a wind event is worth a look rather than a skip. Most good hail maps carry a wind swath layer next to the hail for exactly this reason.
Why swaths are shaped like stripes
New users expect hail to land in blobs. It lands in stripes, and understanding why tells you how to canvass one.
A severe storm produces hail from a fairly small region — in a supercell, the area near the updraft where stones cycle up and down until they get heavy enough to fall out. That region is a few miles across at most. The storm itself, though, is moving, often at thirty to fifty miles an hour. As the hail-producing region tracks across the ground, it paints a long, narrow corridor behind it.
That gives you three practical rules:
- Work along the axis, not across it. The swath's long dimension is the storm's direction of travel. A route that runs parallel to the stripe keeps your reps inside the damage. A route that crosses it wastes half the day on undamaged streets.
- The edges are sharp and they are real. It genuinely is possible for one side of a street to have been hit and the other not. When a homeowner tells you their neighbor got a roof and they did not, the map is often not the thing that is wrong.
- The widest, brightest part is usually not the whole story. Storms pulse. A swath frequently has two or three hot cores along its length with weaker sections between them. Canvass the cores first; they close faster and they fund the rest of the week.
The three layers, and why you want all of them
A single radar-derived number is a lonely piece of evidence. The better hail tools show you at least three views of the same storm, and the disagreements between them are where the useful information lives.
| Layer | What it is | What it is good for | Where it falls down |
|---|---|---|---|
| MESH swath | Radar-estimated maximum stone size, on a ~1 km grid | Drawing the shape of the event and ranking streets by severity | Estimates the storm, not the ground. Known to over-read in some storm structures |
| Spotter reports | Human observations logged with the National Weather Service — trained spotters, law enforcement, and the public | Ground truth. Someone physically measured or compared a stone | Sparse and biased toward where people are. Empty countryside produces no reports no matter what fell |
| POSH | Probability of Severe Hail, a percentage rather than a size | Sanity-checking the edges of a swath and the fringes radar over-reads | Tells you whether, not how big |
The way to use them together is a habit, not a formula. Look at the swath to find the shape. Look for spotter reports near your streets to see whether a human confirmed anything close to the estimated size. Look at probability when a square looks surprisingly hot on its own, out at the edge of the event, with nothing around it — that is the classic signature of radar reading something other than hail.
When all three agree, you have a market. When the swath is hot and the probability is low and there is not a spotter report within twenty miles, you have a lead worth a drive-by before you have a lead worth a crew.
Where hail maps get it wrong
Being straight about this is worth more than another paragraph of features.
The fringes. Every swath has a boundary, and the boundary is a modeled estimate, not a fence. Expect a band of uncertainty around the edge of any colored region. Streets right at the boundary are worth an inspection precisely because nobody else is knocking them.
Bias at both ends of the range. The bias is not in one direction. Work by Murillo and Homeyer (2019) evaluating the original MESH equation against reported hail sizes found it under-estimates smaller stones and over-estimates larger ones — which is why they published recalibrated versions fit to the 75th and 95th percentiles of the report distribution. Practically: a square showing 2.75 inches is telling you "very large hail," not "the stones measured 2.75 inches," and a square showing an inch may have produced a bit more than that.
Certain storm structures fool it. NOAA's own Warning Decision Training Division lists specific cases where MESH runs low: storms tilted over in strong vertical wind shear, left-moving supercells, supercells with a very large bounded weak echo region (which can show up as a suspicious hole of low values right where the worst hail was), and storms dropping low-density, dry stones. If the swath looks like it has a bite taken out of the middle, that is worth a drive-by rather than a skip.
Melting and the fall. Radar sees hail in the cloud. What reaches the ground can be smaller after a long warm fall, and where the freezing level is unusually high, small hail can melt entirely on the way down.
Beam geometry. Radar beams rise as they travel outward from the site. Far from a radar, the beam is sampling much higher in the storm than it is nearby, which can change what the algorithm sees. This is a real limitation of every radar-derived product, not of any one vendor.
Nothing about the roof. The map does not know the roof is nine years old, or three months old, or that the previous owner already collected on it. That is the inspection's job.
Reading time, not just space
The second axis on a hail map is the one people forget. Every swath belongs to a date, and the date matters as much as the shape.
For canvassing, storm age governs your approach entirely. In the first seventy-two hours you are the first knock and the conversation is about getting an inspection scheduled. At three weeks the homeowner has been knocked a dozen times and the conversation is about why you are different. At six months the conversation is about deadlines. Most carriers put a limit on how long after a loss a claim can be filed — frequently one year, sometimes two, and it varies by state and by policy — so an old swath is a real opportunity with a real clock on it.
For claims, the storm date becomes the date of loss. Getting it right is not a formality: a wrong date of loss is one of the more common reasons a legitimate claim comes back denied or gets partially paid. If your neighborhood took hail on both the 14th and the 22nd, the map is what tells you which event you are filing on.
This is where most standalone map subscriptions stop being useful. They can tell you the date; they cannot carry it into the job record, the estimate, or the supplement letter. That handoff is where information gets typed in wrong. It is also the reason we built the hail map and the CRM as one product rather than selling a map that hands you off to another vendor.
Turning a map into a route
Here is the actual workflow, in the order that works.
1. Start with the biggest band and work outward. Filter the map to the top size band in your market and look at what streets sit inside it. That is day one. Do not start at the edge of an event because it is closer to the office.
2. Check the spotter reports before you commit a crew. If there is a human-confirmed report of 1.75-inch hail three miles from your target subdivision, your reps can open with a fact instead of a question.
3. Check the age of the housing stock. A 2-inch swath over a subdivision built last year is worth much less than a 1.25-inch swath over roofs from 2009. The map does not know this. You do, and it is one of the few real edges available to a local contractor.
4. Draw the territory before the reps get in the truck. Assign specific streets, not "that area." Overlap and gaps are both expensive, and both are invisible without a map everybody is looking at.
5. Pin every door, not just the yeses. A no-answer is not a dead door, it is a callback. The pins are what let you run the same neighborhood a second time on a Saturday instead of guessing.
6. Record the storm date on the job the moment it is created. Not later. Later is where it gets typed in wrong.
We wrote a longer field guide on the knocking half of this — see how to canvass a neighborhood after a hailstorm.
Five mistakes worth avoiding
Treating a county-level alert as a hail map. A county hail warning covers hundreds of square miles. If your tool shades whole counties, it is a weather app, not a canvassing instrument.
Ignoring the wind layer. Plenty of storms that produce mediocre hail produce excellent wind, and wind claims are claims. Look at both.
Canvassing a swath without checking whether it has already been worked. In a competitive market, a three-week-old 2-inch swath has been picked over. A one-week-old 1.25-inch swath forty minutes further out often has not.
Promising a homeowner they have damage because the map is red. You cannot know that from a map, and saying it is the fastest way to become the contractor the adjuster distrusts. The honest line — "the radar shows a significant hail event over your address on this date, and I would like to get on the roof and find out what it actually did" — is also the more persuasive one.
Believing a size to two decimal places. It is an estimate. Use the bands, not the digits.
What to do next
If you want to see how this looks on a real address rather than in the abstract, the free hail tracker will check any US address and show you every hail day on record nearby with the date, the estimated size, and how far away it landed. No account, no card.
If you want the map itself — the swath polygons, the size bands, the storm-day picker, and the ability to drop pins on it — that is the paid product, and the entry plan is $79 a month for one whole state.
And if you are still deciding which hail tool to buy, we wrote an honest comparison of the five hail apps roofers actually use, with every published price sourced and linked, including the ones that are more expensive than ours.