q53·intermediate

Did the power go out after this storm — and where?

citieshazards Datasets: 3 15–30 min
✓ Verified answer A computed answer to this question — with its caveats.

A Before-After-Control-Impact natural experiment on real satellite data — the affected area vs. a comparable unaffected control, before vs. after. The change has to clear an empirical noise floor to count as a real signal; a genuine null is reported as a null. Computed on Google Earth Engine — not yet scientist-verified.

Augusta, Georgia

Hurricane Helene power outages — Sep 2024 before 2024-08-15–2024-09-20 → after 2024-09-27–2024-10-04
Measured signal — the trustworthy variables changed in the expected direction.

Night-time lights over Augusta, Georgia changed by -44% after the event, control-corrected vs. a nearby less-affected city. A clear drop in light, consistent with widespread power loss.

Observational analog (a natural experiment over comparable places) — evidence, not a prediction for any one spot. Each bar shows the change versus the natural jitter between undisturbed control pixels; it must clear the tick to count as a real signal.

STRONG (direct, local)Seen up close, at the cut itself — trust it.
night-time lights
-9.872 nW/cm²/sr (-44%) darker (likely outage) SNR 1.3 938 def / 2,172 ctrl px
clears the tick → trust it (for its tier) below the tick → treat as no clear change noise floor estimated, not measured the bar a real signal must clear (1× noise floor)
Provenance & full trace — reproducible

Every number came from this exact query on Google Earth Engine. Same query → same number.

Night-time lights over Augusta, Georgia changed by -44% after the event, control-corrected vs. a nearby less-affected city. A clear drop in light, consistent with widespread power loss.

San Juan, Puerto Rico

Hurricane Maria power outages — Sep 2017 before 2017-08-20–2017-09-15 → after 2017-09-21–2017-10-08
Measured signal — the trustworthy variables changed in the expected direction.

Night-time lights over San Juan, Puerto Rico changed by -33% after the event, control-corrected vs. a nearby less-affected city. A clear drop in light, consistent with widespread power loss.

Observational analog (a natural experiment over comparable places) — evidence, not a prediction for any one spot. Each bar shows the change versus the natural jitter between undisturbed control pixels; it must clear the tick to count as a real signal.

STRONG (direct, local)Seen up close, at the cut itself — trust it.
night-time lights
-8.283 nW/cm²/sr (-33%) darker (likely outage) SNR 0.51 922 def / 432 ctrl px
clears the tick → trust it (for its tier) below the tick → treat as no clear change noise floor estimated, not measured the bar a real signal must clear (1× noise floor)
Provenance & full trace — reproducible

Every number came from this exact query on Google Earth Engine. Same query → same number.

Night-time lights over San Juan, Puerto Rico changed by -33% after the event, control-corrected vs. a nearby less-affected city. A clear drop in light, consistent with widespread power loss.

New Orleans, Louisiana

Hurricane Ida power outages — Sep 2021 before 2021-08-05–2021-08-25 → after 2021-08-30–2021-09-08
Measured signal — the trustworthy variables changed in the expected direction.

Night-time lights over New Orleans, Louisiana changed by -26% after the event, control-corrected vs. a nearby less-affected city. A clear drop in light, consistent with widespread power loss.

Observational analog (a natural experiment over comparable places) — evidence, not a prediction for any one spot. Each bar shows the change versus the natural jitter between undisturbed control pixels; it must clear the tick to count as a real signal.

STRONG (direct, local)Seen up close, at the cut itself — trust it.
night-time lights
-11.054 nW/cm²/sr (-26%) darker (likely outage) SNR 1.11 829 def / 2,345 ctrl px
clears the tick → trust it (for its tier) below the tick → treat as no clear change noise floor estimated, not measured the bar a real signal must clear (1× noise floor)
Provenance & full trace — reproducible

Every number came from this exact query on Google Earth Engine. Same query → same number.

Night-time lights over New Orleans, Louisiana changed by -26% after the event, control-corrected vs. a nearby less-affected city. A clear drop in light, consistent with widespread power loss.

Fort Myers, Florida

Hurricane Ian power outages — Sep 2022 before 2022-09-05–2022-09-25 → after 2022-09-29–2022-10-08
Not measurable — too few comparable pixels in this box to give a trustworthy answer.

Not measurable for Fort Myers, Florida: too few cloud-free Black Marble nights (clouds and the storm obscure the surface), so an outage can't be separated from noise.

Observational analog (a natural experiment over comparable places) — evidence, not a prediction for any one spot. Each bar shows the change versus the natural jitter between undisturbed control pixels; it must clear the tick to count as a real signal.

STRONG (direct, local)Seen up close, at the cut itself — trust it.
night-time lights ⚠ control overshoot — not a trustworthy estimate reported, not judged floor estimated — not measured here
-22.502 nW/cm²/sr (-101%) not measurable SNR 0.79 1,739 def / 1,299 ctrl px
clears the tick → trust it (for its tier) below the tick → treat as no clear change noise floor estimated, not measured the bar a real signal must clear (1× noise floor)
Provenance & full trace — reproducible

Every number came from this exact query on Google Earth Engine. Same query → same number.

Not measurable for Fort Myers, Florida: too few cloud-free Black Marble nights (clouds and the storm obscure the surface), so an outage can't be separated from noise.

Houston, Texas

Hurricane Beryl power outages — Jul 2024 before 2024-06-15–2024-07-05 → after 2024-07-08–2024-07-15
Not measurable — too few comparable pixels in this box to give a trustworthy answer.

Not measurable for Houston, Texas: too few cloud-free Black Marble nights (clouds and the storm obscure the surface), so an outage can't be separated from noise.

Observational analog (a natural experiment over comparable places) — evidence, not a prediction for any one spot. Each bar shows the change versus the natural jitter between undisturbed control pixels; it must clear the tick to count as a real signal.

STRONG (direct, local)Seen up close, at the cut itself — trust it.
night-time lights ⚠ only 1 clear night(s) — not a trustworthy estimate reported, not judged floor estimated — not measured here
-43.65 nW/cm²/sr (-56%) not measurable SNR 6.39 2,207 def / 1,535 ctrl px
clears the tick → trust it (for its tier) below the tick → treat as no clear change noise floor estimated, not measured the bar a real signal must clear (1× noise floor)
Provenance & full trace — reproducible

Every number came from this exact query on Google Earth Engine. Same query → same number.

Not measurable for Houston, Texas: too few cloud-free Black Marble nights (clouds and the storm obscure the surface), so an outage can't be separated from noise.

See all verified answers, with maps →

Real events · NASA Disasters / VEDA

Analysis-ready products for actual events that this question maps to — open each in the catalog, or browse them on the NASA Disasters Portal.

Find the data for your area

Draw a rectangle to pick your area of interest, then see what NASA data covers it (live, here in your browser) or download a ready-to-run notebook with your AOI pre-filled. The notebook runs in any Python environment — it needs a free Earthdata Login to fetch the data.

Current AOI: -82.15, 33.36 → -81.9, 33.55 (Augusta, Georgia)

When a hurricane knocks out the grid, the lights themselves become the data. NASA's **Black Marble** measures how much light each place emits at night — so a city going dark after a storm is a visible, mappable signal of where power was lost. This is the analysis behind the Disasters Portal's Hurricane-Helene power-outage maps.

When a hurricane knocks out the grid, the lights themselves become the data. NASA’s Black Marble measures how much light each place emits at night — so a city going dark after a storm is a visible, mappable signal of where power was lost. This is the analysis behind the Disasters Portal’s Hurricane-Helene power-outage maps.

What you can answer

  • Where the lights went out. Compare a clear pre-storm night to a post-storm night: neighbourhoods that dimmed or vanished are the likely outage areas.
  • How recovery progresses — track the lights coming back over the following nights.
  • Relative severity — bigger radiance drops over more people = higher-priority areas.

What you can NOT answer (be careful)

  • It’s a proxy, not a meter. Black Marble measures light, not the grid directly. Clouds, moonlight, snow cover, and even wildfire smoke distort nighttime radiance — which is why the product is BRDF-corrected and why you compare clear nights and read it cautiously.
  • Exact customer counts — light loss correlates with outages but isn’t a utility outage feed.
  • Under thick cloud — a storm’s own clouds can blind the sensor; wait for the first clear night.

How you’d approach it

Take VNP46A2 for a clear night before and after, difference the radiance over your AOI, mask cloud-flagged pixels, and overlay population to rank affected areas. Supports the Respond phase of the NASA Disasters program. The interactive nightlights guide explains the method.

How a scientist answers this
Parameters
NASA Black Marble VNP46A2 daily BRDF/atmosphere-corrected nighttime radiance (nW·cm⁻²·sr⁻¹); Black Marble HD for neighbourhood-scale detail; VIIRS Day/Night Band raw radiance. Compare a clear pre-storm night to a clear post-storm night over the same AOI, using the product's QA/cloud flags.
Method
Difference post-storm minus pre-storm radiance per pixel (or as percent drop), mask cloud/snow/stray-light-flagged pixels using the VNP46A2 QA layers, then rank dimmed areas by radiance loss weighted by underlying population.
Validation
Read it as a proxy, not a grid meter — moonlight, snow, smoke, and clouds distort radiance (hence BRDF correction and clear-night comparisons); validate against utility outage reports where available and wait for the first cloud-free post-storm night.
In plain EnglishCompare how bright a city looks at night before and after a storm; neighbourhoods that went dark are the likely power-outage areas.

Make it yours → Set your AOI and pick clear pre-storm and post-storm dates, then overlay population to prioritize the hardest-hit areas.

Run the core method · no login

The Before-After-Control-Impact vs a noise floor at the heart of this question — runnable on synthetic data, right here. The full earthaccess code template further down does it on real NASA data (needs an Earthdata login).

editable · runs in your browser