South Texas, Nov 8–9, 2026
Verified storm.
This storm was selected on Nov 6.
Built Nov 18, 16:14 UTC from runs through Nov 7 12Z.
1 day before: the 00Z and 12Z runs of Sat Nov 7, averaged.
Rain that fell per radar estimates as well as observations at station gauges
What fell at each of the 39 gauges over the storm, on the same colors as the models' maps. Between them is a radar estimate.
Which model came closest
Lead at which model predicted a stormMost models predicted the storm at 6 or more days out, AIGFS and GFS only from 4 (AIGFS first at 6).
WhereToo close to call: AIGFS 61 mi and GFS 65 mi from the satellite estimate's center of the heaviest rain on average.
How much rainAIFS came closest, by 0.09 in over ECMWF, and was closer than ECMWF at 24 of 39 gauges.
Sustained windGEM came closest, by 0.7 mph over ECMWF, and was closer than ECMWF at 9 of 24 stations. GFS, ICON and UKMO are not in it, having predicted the storm at too few leads to pool.
GustECMWF came closest, by 0.6 mph over GEM, and was closer than GEM at 7 of 18 stations. GFS, ICON and UKMO are not in it, having predicted the storm at too few leads to pool.
What happened
A heavy rain event in South Texas dropped > 6 in of rain in some locations. All the models saw the storm for at least 4 days prior to the event, some 6+ days. Most models under-forecast the rain; AIFS generally performed the best. Wind gusted over 30 mph in some locations; ECMWF performed the best there.
Jonathan, Nov 12
Rain
1.64 in
on average at the 39 gauges; 0.00 to 6.58 in
- Brownsville Intl6.58 in
- Alice Intl Airport5.30 in
- Kingsville NAS5.15 in
- Port Isabel-Cameron County Apt4.63 in
Rain per six hours, averaged over the 35 gauges that reported every hour. Wettest from Sun Nov 8, 6 am.
Wind
37 mph
peak gust, at Corpus Christi Intl; sustained to 24 mph, at Corpus Christi Intl
- Corpus Christi Intl37 mph
- Alice Intl Airport36 mph
- Rockport/Aransas Co33 mph
- Cotulla Municipal32 mph
Each station's highest gust over the storm, at 18 stations. The sustained wind was highest across them at Sun Nov 8, 8 am.
Every map
Choose a map to show it large above.
Where the rain went
Satellite estimate, ranked
- Some
- More
- Most
NASA's IMERG Late, Nov 8 to Nov 10, ranked across this map rather than in inches: Most is its wettest 10%, More is the next 15%, Some is the next 25% of the map, and the rest is unshaded. The dashed outline is the region scored. The cross is the satellite estimate's center of the heaviest rain in it, and each model's track joins its own center from its earliest forecast to the day before, the last dot larger.
How far off the heaviest rain was
Each model's center of the heaviest quarter of its rain in the region, distance from the satellite estimate's, by days before the storm. A model is drawn only at leads it predicted the storm. The gauges do not sit under most of the heaviest rain, so the satellite estimate is the truth here: it ranks rain well across a region and is one more opinion at tens of miles.
Lead at which each model predicted a storm
A filled dot: the model's window for that lead holds the storm. An open one: no storm at that lead. The earliest scored is 6 days before.
The score
| How much rain | 1 day | 2 days | 3 days | 4 days | 5 days | 6 days | 1–4 days |
|---|---|---|---|---|---|---|---|
| AIFS | 0.65 | 0.64 | 0.67 | 0.75 | 0.93 | 1.18 | 0.68 |
| ECMWF | 0.49 | 1.01 | 0.72 | 0.88 | 1.36 | 1.10 | 0.77 |
| AIGFS | 0.80 | 0.73 | 0.79 | 1.16 | no storm | 1.46 | 0.87 |
| GFS | 1.10 | 0.73 | 0.71 | 1.28 | no storm | no storm | 0.96 |
| GEM | 1.20 | 1.30 | 0.90 | 0.70 | 1.20 | 1.39 | 1.03 |
| No rain | 1.64 | 1.64 | 1.64 | 1.64 | 1.64 | 1.64 | 1.64 |
How much rain: mean error at the 39 gauges, in inches, by days before the storm. Outlined: closest at that lead. Highlighted: closer than forecasting no rain. “no storm” is not scored on how much rain. Choose a lead for its tables.
| Bias | 1 day | 2 days | 3 days | 4 days | 5 days | 6 days | 1–4 days |
|---|---|---|---|---|---|---|---|
| AIFS | −0.05 | −0.28 | −0.39 | −0.63 | −0.88 | −1.13 | −0.34 |
| ECMWF | −0.23 | +0.42 | −0.32 | −0.54 | −1.33 | −1.10 | −0.17 |
| AIGFS | −0.56 | −0.45 | −0.74 | −1.13 | no storm | −1.36 | −0.72 |
| GFS | −0.47 | −0.55 | −0.17 | −1.14 | no storm | no storm | −0.58 |
| GEM | −0.98 | −1.30 | −0.76 | −0.59 | −1.20 | −1.39 | −0.91 |
Bias: each model's rain minus what fell, averaged over the 39 gauges, in inches, by days before the storm. Positive: the model forecast more rain than fell. Negative: less.
The same mean error as a line per model, by days before the storm, broken where a model had no storm. Dashed: what forecasting no rain at all scores, a reference and never ranked.
1 day before
Rain
| Model | Mean error, in inches | Bias, in inches | Closest, of 39 |
|---|---|---|---|
| ECMWF | 0.49 | −0.23 | 12 |
| AIFS | 0.65 | −0.05 | 10 |
| AIGFS | 0.80 | −0.56 | 7.3 |
| GFS | 1.10 | −0.47 | 2.3 |
| GEM | 1.20 | −0.98 | 7.3 |
39 gauges, 25 of them wet.
The rain score
Each model's storm total over its own window at this lead, against what the gauge measured over the storm. Mean error is the typical miss, bias its sign, and closest the gauges where the model was nearest, a tie shared. The storm's winner pools leads 1, 2, 3, 4, and is named only when it is also closer than the runner-up at more gauges than not.
Sustained wind
| Model | Mean error, mph | Bias, mph | Closest, of 24 |
|---|---|---|---|
| GFS | 2.0 | −0.1 | 10.5 |
| GEM | 2.7 | +0.1 | 4 |
| ECMWF | 3.1 | −2.9 | 5.2 |
| AIGFS | 3.6 | −2.8 | 1.5 |
| ICON | 4.3 | −4.3 | 1.9 |
| AIFS | 4.5 | −4.5 | 0.2 |
| UKMO | 5.1 | −5.1 | 0.7 |
24 stations.
Sustained wind
Each model's peak over its own window at this lead, against the station's peak over the storm's days. Mean error is the typical miss, bias its sign, and closest the stations where the model was nearest, a tie within 1 mph shared. The storm's winner pools leads 1, 2, 3, 4, 6, and is named only when it is also closer than the runner-up at more stations than not.
Gust
| Model | Mean error, mph | Bias, mph | Closest, of 18 |
|---|---|---|---|
| ECMWF | 2.9 | +0.9 | 5.6 |
| GEM | 3.9 | −0.2 | 4.2 |
| GFS | 4.2 | 0.0 | 4.3 |
| ICON | 4.9 | −4.1 | 1.2 |
| UKMO | 6.0 | −5.6 | 2.6 |
18 stations.
Gust
Each model's peak over its own window at this lead, against the station's peak over the storm's days. Mean error is the typical miss, bias its sign, and closest the stations where the model was nearest, a tie within 1 mph shared. The storm's winner pools leads 1, 2, 3, 4, 5, 6, and is named only when it is also closer than the runner-up at more stations than not.
When the models saw it coming
Each model's forecast rain averaged over the 39 gauges, from 6 days before the storm to 1, against what fell. A model is drawn only at leads it predicted the storm.
When the rain came
Bars: what fell. Lines: each model 1 day before. Rain at the 35 gauges, in inches per six hours. Times are CST.
When the rain came
The bars are what fell, averaged over the 35 gauges that reported every hour of the storm, and each line is a model's rain at those same gauges, per six hours, with the lead's 00Z and 12Z runs averaged. The thick stretch is the model's window at this lead, the stretch of its rain that every measure is scored over, and the dot is its peak. A window is proposed by a rule and corrected by hand where review disagreed, and the operator's notes say why.
Each model's window, 1 day before
| Model | Starts | Peak rain | Ends | Against the gauges' peak | Rain, in inches | Wettest gauge, in inches |
|---|---|---|---|---|---|---|
| GFS | Sat 12 pm00Z12Z | Mon 12 pm00Z12Z | Tue 12 pm00Z12Z | 24 h late | 1.18 | 4.39 at San Antonio/Stinson (1.25 fell) |
| ECMWF | Sat 12 pm00Z12Z | Sun 6 pm00Z12Z | Mon 6 pm00Z12Z | 6 h late | 1.41 | 6.24 at Alice Intl Airport (5.30 fell) |
| ICON | Sat 6 pm00Z12Z | Mon 12 pm00Z12Z | Wed 12 pm00Z12Z | 24 h late | 1.22 | |
| AIFS | Sat 12 pm00Z12Z | Sun 12 pm00Z12Z | Wed 12 pm00Z12Z | on time | 1.59 | 4.39 at Port Isabel-Cameron County Apt (4.63 fell) |
| AIGFS | Sat 12 pm00Z12Z | Sun 12 pm00Z12Z | Tue 12 am00Z12Z | on time | 1.08 | 5.25 at Port Isabel-Cameron County Apt (4.63 fell) |
| GEM | Sat 12 pm00Z12Z | Sun 6 pm00Z12Z | Mon 6 pm00Z12Z | 6 h late | 0.67 | 4.81 at Cotulla Municipal (4.11 fell) |
| UKMO | Sat 12 pm00Z12Z | Sun 12 pm00Z12Z | Mon 6 pm00Z12Z | on time | 0.63 |
Each time links both runs' sea-level pressure and rain chart at that hour.
These maps are drawn from model output we store ourselves, based on data and products of NOAA, the European Centre for Medium-Range Weather Forecasts (ECMWF), Environment and Climate Change Canada and NASA's Goddard Space Flight Center, and from Open-Meteo for ICON and UKMO. The gauges and stations are Iowa State's IEM.
Sources and licenses
- NOAA: GFS and AIGFS, from NOAA's NOMADS server. US government data, not subject to copyright (NOAA/NWS terms of use).
- ECMWF: IFS and AIFS Single, from ECMWF's open data. This service is based on data and products of the European Centre for Medium-Range Weather Forecasts (ECMWF). Source: www.ecmwf.int. This ECMWF data is published under a Creative Commons Attribution 4.0 International (CC BY 4.0). ECMWF does not accept any liability whatsoever for any error or omission in the data, their availability, or for any loss or damage arising from their use.
- Environment and Climate Change Canada: GEM, from the MSC Datamart. Data Source: Environment and Climate Change Canada, under its Data Services End-use Licence.
- NASA: IMERG Late, from the GES DISC archive. Freely available under GPM's data policy; the reference is below.
- NOAA: the radar estimate, MRMS Multi-Sensor QPE (Pass 2), from NOAA's open data on AWS. NOAA Multi-Radar/Multi-Sensor System (MRMS) was accessed on Nov 12, 2026 from https://registry.opendata.aws/noaa-mrms-pds.
- Open-Meteo: ICON and UKMO, from Open-Meteo's open data (CC BY 4.0).
- Iowa State's IEM: the gauges' rain and the stations' wind, from the Iowa Environmental Mesonet.
Huffman, G. J., E. F. Stocker, D. T. Bolvin, E. J. Nelkin and J. Tan (2023): GPM IMERG Late Precipitation L3 1 day 0.1 degree x 0.1 degree V07, edited by A. Savtchenko. NASA/GSFC, Greenbelt, MD, USA, NASA Goddard Earth Sciences Data and Information Services Center (GES DISC). Accessed Nov 11, 2026 at doi:10.5067/GPM/IMERGDL/DAY/07.
What we change: each model's rain is cropped to the storm's map and summed over its window, the disagreement map is the difference of those, and the satellite estimate is summed over the storm's days, cropped and ranked, and the radar estimate is summed over the gauges' days, cropped, averaged onto a coarser grid and hidden where its own radar quality index is low. These are our maps of the providers' numbers, not their products, and none of them has reviewed or endorsed this site.
Leads 1 to 6 are drawn. GFS and AIGFS come from NOAA; ECMWF and AIFS from ECMWF; GEM from Environment and Climate Change Canada; ICON and UKMO from Open-Meteo's open data.
About this page
00Z and 12Z change the maps and nothing else: the timing table, the score and the winner stay on the two runs averaged. A run is read over its lead's window, and where a lead has none the map shows that run's own nearest rain event and says the score does not use it.
Lead 0, the runs of the storm's first day, is not drawn: no window at that lead was reviewed. AIGFS's rain is the sum of its six-hour files, because its filed run total reads short. Times are CST.