Reading Hail History

In short: Hail history lists every date since October 14, 2020 on which radar estimated hail of 0.5 inch or larger at a place or within 5 miles of it. Each date carries four sizes: at the location, and the largest within 1, 3 and 5 miles. The sizes come from MESH, a NOAA radar estimate, so they say where hail very likely fell and roughly how big, not what landed on one roof. The NWS storm reports under each date are what people on the ground saw, kept apart from the radar numbers.

What to look at first.

• The big number at the top counts dates at the location alone. The smaller line under it counts dates within 1, 3 and 5 miles.
• The At location view lists only the dates that reached the location. Within 5 mi lists every date with all four sizes, in the layout hail tables commonly use.
• Open a date for its NWS storm reports within 10 miles, and Show on map for that date’s swath.

What MESH is

MESH, the Maximum Estimated Size of Hail, comes from NOAA’s MRMS system, which blends every NEXRAD radar into one national grid of 1 km squares. It estimates the biggest hail a storm is producing from how much strong reflectivity sits above the freezing level, weighted toward the colder levels where hail grows. Hail on Radar covers how it works; MRMS Explained covers the grid.

NOAA publishes a file every day that holds, for each square, the largest MESH value of the previous 24 hours. Hail history is built from those files: one date per file, every square of 0.5 inch or larger kept, back to the first file on NOAA’s public archive, October 14, 2020. Nothing older exists there, which is why the list starts in 2020.

Why a radar estimate is not a measurement

Radar sees the storm in the air, often thousands of feet up. MESH turns that into a size with a formula; no stone is weighed or measured. Researchers who compare it with hail on the ground keep finding the same thing:

• It is not meant to be exact. A 2019 study for the American Meteorological Society compared radar hail products with ground observations collected by the Insurance Institute for Business & Home Safety (IBHS). It describes MESH as “not meant to be a 1-to-1 estimate”, and found the radar products did poorly at estimating a discrete size.
• It can run well high. In a 2013 case from NOAA’s National Severe Storms Laboratory testbed, MESH showed 3.31 inches where the largest stone reported was golf ball size, 1.75 inches. MESH was fit so that about three quarters of the largest observed stones are smaller than the estimate (Witt and colleagues, 1998), so a high reading is the design, not a fault.
• It is still the best radar guide to where hail fell. A 2018 study of a large set of ground hail observations found MESH the best radar tool for telling severe hail from smaller hail. Use it for where and for roughly how big.

Hail also melts on the way down and drifts with the wind, and the biggest stones often fall in a narrow streak. One 1 km square is about 0.6 miles on a side, larger than a neighborhood block. So a size printed for a location is the largest estimate for that square over 24 hours, not a reading taken at a house.

That is why every size in Hail history either sits under the caption “RADAR-ESTIMATED MAXIMUM HAIL SIZE” or ends in “est.”, why sizes move in quarter-inch steps (0.5, 0.75, 1, 1.25 and so on, or the nearest 5 mm), and why a radar size never carries a coin or ball name. The sheet’s foot, and every picture shared from it, carries the line “Radar-estimated hail size from NOAA MRMS, not a damage report.”

How to read the four columns

• The numbers never fall from left to right. Each ring includes the smaller ones, so a ring’s number is never smaller than the location’s.
• A dash means none of 0.5 inch or larger in that column. It does not mean no hail at all: radar estimates below 0.5 inch are not kept, and small hail can fall where radar shows nothing.
• The ring numbers are not the location’s. The 5 mile number is always the largest in the row because it covers about 78 square miles. The At location column is the only one about the place itself, which is why it comes first and is the heaviest.
• A city or ZIP is its center. If you looked up a town rather than a point, the At location column is the square at the town’s center. Tap the map on the house itself (in Show on map) and “Hail history for this point” re-reads the whole list there.

On the map, the location’s own square is outlined, the 1, 3 and 5 mile rings are thin circles around it, and the date’s squares are drawn in the Hail Size colors, unsmoothed, so the coarseness of the estimate stays visible. A dashed line marks the edge of the area drawn for the date, and the low sheet’s last line says how far it reaches every way (about 20 miles in most of the country); nothing is drawn past it. While a past date is on the map, the live radar, satellite pictures, warnings and every other layer from another time come off, and come back when the sheet closes.

How the NWS storm reports differ

The NWS storm reports under each date are reports of hail people saw: spotters, emergency managers, the public, collected by the National Weather Service and listed by the Storm Prediction Center. They are preliminary, and the sheet says so.

• A report is one person at one spot and time. It has a distance and direction from the location, a town as the SPC lists it, and a clock time with its weekday.
• A report keeps the NWS size word when its size is one on the SPC’s table (penny 0.75 in, quarter 1 in, golf ball 1.75 in, baseball 2.75 in, and the rest). A radar size never gets one.
• The two are never merged or averaged. A report is never counted into a ring column. Where they disagree, both are shown: a report without a radar size can be hail too small or too brief for the radar, and a swath without a report can be a place where no one was looking.
• No report is not no hail. Many hail falls are never reported, most of all at night and in open country. “No NWS hail reports within 10 mi on this date” is a fact about the reports, not about the hail.

The 7 AM date window

A date is not midnight to midnight. NOAA’s daily file covers the 24 hours from 12 UTC, which is 7 AM Central daylight time (6 AM in winter), and the storm reports are cut the same way. Each date is named by the day its window starts, and every row prints its window in your phone’s time, for example “7 AM Fri to 7 AM Sat”.

So a storm at 1 AM Saturday belongs to Friday’s date. Its storm report reads “Sat 1:10 AM” under Friday, the weekday showing why. If you know the night a storm came through, look at the date before it as well as the date of.

The dates with no data

A few dates in the archive have nothing to show, and the list prints them at their place rather than skipping them, so a missing date is never mistaken for a clear one.

• Radar artifact, date left out. A bad radar file can paint giant false sizes over wide areas. A date with more than 20,000 squares over 2 inches in the raw file is left out whole (two since October 2020), and any single square over 6 inches is dropped as an artifact.
• Not published by NOAA. NOAA did not publish the day’s file (two since October 2020). A date that fails to load is marked the same way until it is read.

New dates are added each morning after NOAA’s file lands; the line at the foot of the sheet names the last date held.