Module 02 · Confirm · Be An Earth System Trend Detective!
Radar sees the flood through the clouds. This viewer makes it readable.
Interactive and fully static: everything runs in your browser from precomputed data. Use the "Experiment briefing" button inside the tool for a guided walkthrough.
During a flood the sky is covered, so optical satellites see cloud, not water. Radar sees through it day and night, but responders and the public can't read radar images, and one flood map can't say if floods are a trend.
Floods arrive with the storm, so optical satellites usually see only cloud. Synthetic aperture radar sees through cloud, day or night, but radar images are hard to read for anyone who isn't a specialist.
SAR Change takes the 29 October 2024 DANA flood in Valencia, Spain, and makes the radar change visible and explainable, with a path to NISAR L-band.
The data story · Confirm
Five beats: the raw radar, what changed, how sure we are, where it gets subtle, and who acts on it.
01 · The signal
On 29 October 2024 a DANA storm dropped extreme rain on Valencia. The Spanish government counted 230 deaths. Under that cloud, optical satellites saw nothing useful.
Sentinel-1 radar works through cloud and at night. Smooth water reflects the radar pulse away, so it shows up dark. SAR Change turns those images into a water map anyone can read.
02 · What changed
NASA/JPL's OPERA surface-water product, made from the same radar, counts water outside the lagoon and the sea. It rose from 30.7 km² on 26 October to 77.7 km² on 31 October, two days after the storm.
26 Oct 202431 Oct 202403 · How sure we are
Our U-Net, trained on hand-labelled Sen1Floods11 chips, reaches 61.7% water IoU on 90 test chips: 3.0× the standard Otsu threshold (20.3%). OPERA is built separately by NASA/JPL and shows the same jump on 31 October, then 30–60 km² of standing water for three weeks.
On the longer scale, rain and storage tell different stories. Satellite rainfall (GPM IMERG, 1998–2024) shows no trend (annual p = 0.62). GRACE says the ground itself has been losing water: −0.39 cm a year, p = 0.002, a pattern that would turn up by chance about 2 times in 1,000 if nothing were changing.
Are Valencia's rain extremes rising?
Reanalysis and satellite agree: no trendIs eastern Spain's stored water changing?
One record: falling04 · The nuance
The Albufera rice paddies were already filling before the storm (4 km² on 1 October, 31 km² on 26 October), so only the jump to 77.7 km² is the flood. Radar misses water between buildings, so the urban flooding in l'Horta Sud is under-detected, and much of it drained before the first pass.
NISAR's L-band GCOV products start in October 2025, after this flood, so the viewer is NISAR-ready rather than NISAR-built. One extreme event isn't a trend.
05 · Who acts on it
Civil protection, responders and insurers need to know where water is standing days later, not just where it fell. A radar map available through cloud tells them where to send pumps, check access and verify claims first.
The trend layer adds the slow story: rain extremes aren't trending, but stored water is falling. Planners should prepare for the extreme event itself rather than expect a steady rise in rain.
Launch the dashboardWhat this means
77.7 km²of new surface water on 31 OctoberUp from 30.7 km² five days earlier (OPERA DSWx-S1, 30 m).
The Albufera paddies were already rising; urban water is under-detected.26 Oct 202431 Oct 2024Are Valencia's rain extremes rising?
Reanalysis and satellite agree: no trendIs eastern Spain's stored water changing?
One record: fallingWhat this means
77.7 km²of new surface water on 31 OctoberUp from 30.7 km² five days earlier (OPERA DSWx-S1, 30 m).
The Albufera paddies were already rising; urban water is under-detected.Who uses it
The path through the dashboard, from the situation they start in to the decision it changes.
Valencia / l'Horta Sud
After 29 October 2024 the sky is closed, the public can't read a radar image, and there's a risk of treating one disaster as proof that autumns are getting wetter.
Viewer tab: swipe before/after on the same orbit (26 Oct vs 7 Nov), then scrub the time-lapse.
Read "Why radar works here".
On the module page: OPERA new water, 30.7 → 77.7 km², with the paddy and urban caveats.
Trend layer: no rain trend in POWER or IMERG; GRACE storage falling.
Brief the flood as a cloud-proof, quantified jump in new water, with urban flooding undercounted, and not as evidence of a wetter climate.
A swipeable radar story, an OPERA number with a date, and tested climate context that prevents a false trend claim.
An illustrative journey built from what the dashboard does today, not a field-tested outcome.
Launch the dashboardValencia region
They need a footprint for drainage design or a loss estimate, and a view on whether this is the new normal.
Swipe the radar and read the OPERA series.
Apply the caveats: paddies already filling before the storm, urban water under-detected.
Read the trend layer and the GRACE mascon map: drying storage, no rain trend.
Treat urban loss as larger than the radar footprint, and plan for an extreme event on a drying water-storage background, not a proven rise in extreme rain.
A bounded statement: what the radar saw, what it missed, and what the 20- to 45-year tests do and don't support.
An illustrative journey built from what the dashboard does today, not a field-tested outcome.
Launch the dashboardU-Net water overlap on Sen1Floods11's official hand-labelled test split.
The standard automatic threshold, on the same chips. A single tuned VV threshold reaches 46.8%.
Channel ablation: adding VH and the VV/VH ratio each help.
Per 512×512 tile on a CPU. INT8 kept accuracy but was about 9× slower on that CPU, so FP32 stays.
Sentinel-1 radar passes in the Valencia viewer.
NISAR L2 GCOV granules found via NASA CMR over Valencia, Toronto and Jasper.
Held-out test data, compared against baselines. One seed per run, not a published benchmark.
Before and after
26 Oct 2024 (3 days before)7 Nov 2024 (9 days after)
Visuals
Earthdata upgrade · NASA OPERA
26 Oct 2024 (before)31 Oct 2024 (2 days after)
We added NASA/JPL OPERA DSWx-S1, a surface-water map made from Sentinel-1 radar at 30 m, downloaded with an Earthdata login. 18 radar passes between 1 October and 19 November 2024 cover the coast south of Valencia. Outside water that was already there in early October, new water rose from 30.7 km² on 26 October to 77.7 km² on 31 October, two days after the flood, then stayed between about 30 and 60 km² for the next three weeks.
Read it with care. Most of the new water sits in the Albufera rice paddies around the lagoon, and it was already rising before the flood (from 4 km² on 1 October to 31 km² on 26 October), so the flood signal is the jump to 77.7 km². Radar water maps miss water between buildings, so the urban flooding in l'Horta Sud is under-detected, and much of it had drained by the first pass. Passes with under 80% valid pixels are left out.
New water outside usual water, km² · OPERA DSWx-S1, 30 m
Trend layer · Water
A flood map shows one event. To ask if heavy rain is changing, we tested 45 years (1981–2025) of NASA POWER daily precipitation over eastern Spain (around Valencia) and southern Ontario (around Toronto, a NISAR site in the viewer): annual totals, the wettest day of each year, and heavy-rain days.
Mann-Kendall (α = 0.05) and Sen's slope with 95% CI, one value per year. Hover or tap the chart for yearly values.
Where it is changing
The approach
26 Sentinel-1 RTC passes over four orbit tracks around Valencia (Sept–Dec 2024). Each orbit shares one colour stretch, so the time-lapse doesn't flicker.
Before/after swipe, time-lapse and a VV dB change map for three same-track pairs that bracket the flood.
A U-Net trained on Sen1Floods11's hand-labelled chips, tuned with 14 Optuna trials, and compared against Otsu and a global VV threshold.
Why calm water looks dark, what VV and VH mean, and why NISAR's L-band can see flooding under vegetation where C-band can't.
NISAR L2 GCOV granules found through NASA CMR (Valencia 55, Toronto 67, Jasper 76), with public browse images in the viewer and hooks for full NISAR files.
Limits we know about
| Dataset | How it's used | Access | Link |
|---|---|---|---|
| NASA-ISRO NISAR L2 GCOV (metadata + browse) | Granule search via NASA CMR; public browse images from ASF DAAC | Public, no login | source |
| Sentinel-1 RTC (ESA Copernicus) | VV/VH backscatter, 10 m, Valencia, Sept–Dec 2024; RTC via Microsoft Planetary Computer | Public, no login | source |
| Sen1Floods11 (CC-BY-4.0) | Hand-labelled Sentinel-1 flood chips (Bonafilia et al., 2020) | Public, no login | source |
| JRC Global Surface Water | Permanent-water reference overlay on the test chips | Public, no login | source |
| NASA POWER daily PRECTOTCORR (trend layer) | Daily precipitation 1981–2025 (MERRA-2 based), 35 cells per region; Mann-Kendall + Sen's slope | Public, no login | source |
| GRACE/GRACE-FO JPL mascon RL06.3 CRI v04 (trend layer) | Monthly total water storage 2002–2026, land cells over eastern Spain and southern Ontario; deseasonalised annual means | Earthdata login | source |
| GPM IMERG V07 monthly GPM_3IMERGM (trend layer) | Satellite rainfall, 0.1°, eastern Spain and southern Ontario boxes, 1998–2024: annual, Sep–Nov and wettest-month totals | Earthdata login | source |
| NASA/JPL OPERA DSWx-S1 (module upgrade) | Radar-derived surface water, 30 m, 18 passes 1 Oct–19 Nov 2024 over the Valencia flood zone | Earthdata login | source |
For the judges
Floods are among the most costly disasters, and radar is often the only clear view during the event. Making SAR legible helps responders, journalists and the public.
Pairs an interactive swipe and time-lapse with a plain-language physics explainer, and is honest about what C-band can't see and L-band NISAR can.
Scored on Sen1Floods11's official hand-labelled test split, against two classical baselines, with a channel ablation and CPU timing.
Trend Detective: puts the Valencia flood in a 45-year NASA POWER rainfall record and asks whether extremes are trending there and in southern Ontario. The viewer uses real NISAR GCOV granules found via NASA CMR, with browse imagery in the app.
Swipe, time-lapse and a 'why radar works here' tab turn a specialist dataset into a story anyone can follow.
First designed around the related Space Apps 2026 brief Dancing with the SARs; here it is the confirm module of one system in Be An Earth System Trend Detective!