--- title: what radar can see canonical_url: https://ensurance.app/guide/what-radar-can-see markdown_url: https://ensurance.app/guide/what-radar-can-see.md subtitle: weather radar watches the storm. imaging radar looks through the cloud. neither one is the gauge in the creek category: ecosystem-services --- # what radar can see *weather radar watches the storm. imaging radar looks through the cloud. neither one is the gauge in the creek* Synthetic aperture radar (SAR) is a satellite or aircraft instrument that sends a microwave pulse at the ground and records what comes back. Because the platform is moving, a short antenna can be processed as if it were a much longer one — that is the synthetic aperture. It images at night and through cloud. Copernicus Sentinel-1 carries a C-band SAR of exactly this kind. That is the answer to the search. Here is the part that usually gets skipped: a radar pass is a picture plus a model, and the picture is of a consequence. A creek, a soil profile, a canopy, and a storm already produce a signal whether or not a network stores it. A probe, a lab assay, a camera, a radar pass, and an indicator sheet are readings of that living system. **[ensurance](/?from=guide)** keeps the reading beside what the place claims, so funding can cite the condition. It does not replace anyone's sensors, and it is not the living system. The storm forms. The channel fills. A satellite passes over and records water sitting where water does not usually sit. The flood was real before the image existed. The image is evidence about the flood. It is not the flood, and it is not the stage of the creek. ## two instruments share one word "Radar" names two different jobs, and conversations about environmental data go sideways when the two get mixed. One looks up at the weather. The other looks down at the ground. | | weather radar | imaging radar (SAR) | |---|---|---| | **what it looks at** | precipitation in the atmosphere | the surface of the ground | | **example system** | NOAA NEXRAD, S-band Doppler | Copernicus Sentinel-1, C-band | | **what the return gives you** | an estimate of precipitation over an area | an image of how the surface scattered the pulse | | **common products** | rainfall estimates | flood extent, surface change, a soil-moisture retrieval | | **what it is not** | a gauge in a channel | a probe in the ground | ### weather radar watches the air NEXRAD is NOAA's Next Generation Weather Radar network — S-band Doppler radar that estimates precipitation in the atmosphere. That sentence carries the limit built into the instrument. The radar is looking at hydrometeors aloft. It is not looking at the tipping bucket on the ridge, and it is not looking at the channel downstream. A rain gauge catches water that landed at one point. A radar estimate covers area, which the gauge cannot, and infers rate from a return, which the gauge does not have to. Both are readings of the same storm, answering slightly different questions, and a hydrologist who has corrected one against the other for years already knows the gap by heart. ### imaging radar looks at the ground Sentinel-1's C-band SAR images day and night and through cloud, which is the whole reason it matters for storms and floods — the sky is usually not cooperating when you most want a look. The physics that makes flood mapping work is almost disappointingly simple: calm open water scatters the pulse away from the sensor, so calm open water shows up as a dark patch of low return against the land around it. Water under trees or between buildings can go the other way — bright, from a double bounce — so a dark-patch map can miss flooded forest and flooded streets. So a flood-extent map is a classification of a radar image. Somebody, or some model, drew the line between water and not-water, and that line moves if you change the threshold, the reference image, or the day. Wet roads, smooth bare soil, and radar shadow behind terrain produce low return too. The map is good. The map is still a decision made about a picture. Soil moisture from the same instrument is a further step out. A C-band soil-moisture figure is a retrieval of roughly the top few centimeters, and vegetation, surface roughness, and the crop all move the number. That is not a knock on it — no probe network matches that coverage. It is a different object from a probe at a known depth in one soil profile, and treating the two as interchangeable is how an argument gets lost in front of an auditor. :::johnson **radar sees through the cloud. it does not replace the plate in the creek.** The satellite's conclusion and the field reading are two sources about one place. Both arrive as rows. Neither one deletes the other. [how a stack joins →](/guide/how-to-plug-a-sensor-stack-in?from=guide) ::: ## a retrieval is a model, and it keeps that label Flood extent, surface change, and soil moisture from radar are all conclusions drawn from a return signal. They can be excellent. They are still inferences, and the honest version of a radar row says so in the method field. A reading that arrives here carries the result, the unit, the time it happened, the place it belongs to, and the method — a person, a lab, or a model. A radar retrieval is a model. Naming the mission and the band there is not a disclaimer; it is what makes the row reusable later, when someone wants to know whether the number came from a pulse or from a probe. If you hear the word *tier* used about readings in this context, read it as one thing: how sure the row is allowed to sound. A Sentinel-1 flood classification can be a strong row. It is still a model, labeled as one. ## the plate in the creek stays A staff plate bolted to a bridge pier, read by a person who writes the number in a book, is not obsolete because a satellite exists. It is the other source, and the other source is what makes a radar conclusion worth anything. Say the SAR scene shows water across a field on the left bank, and the plate on that reach says the creek crested below the bank. Several things could be true: the water came from irrigation return or a side tributary, the pass caught a moment the plate reading missed, the classification picked up smooth wet ground, or the plate was read wrong. That argument is the useful part of the record, and it only happens if both rows survive. So both stay. There is no blended score here, and no confidence number we compute out of a pass. A gauge-corrected rainfall product, or a classification that already carries its own uncertainty, can arrive as its own labeled row. A sensor does not write a price. A person with local knowledge decides what the disagreement means. ## what a radar conclusion sends, and what it does not The conclusion travels. The scene does not. Nobody needs a copy of a radar granule in a record about a place — a pointer to where the file lives is enough, and the file stays with whoever holds it. So a radar arrival looks like this: the product's own name for the measure, the result and its unit, the time of acquisition, the method stated as a radar retrieval with the mission named, and the place it belongs to — a reach, a parcel, a catchment the account already claims. A result with no place attached can be kept as a note. It does not count as evidence. And the honest part, because a reader here starts looking for a login: the public observation intake is not switched on. There is no URL to post a feed to, no endpoint on ensurance.app for a satellite product, and no public observation log that refreshes. The shape of the row is decided. The public intake is not. [The place keeps the record](/guide/the-place-keeps-the-record?from=guide) says the same thing to a corporate reader, and that has not changed. ## who is already reading both pictures A water utility watching the catchment above a reservoir reads rainfall estimates in the hours after a storm and surface change in the weeks after a fire. A floodplain administrator reads extent maps against the gauge network the county already maintains. A reinsurer scoping exposure reads both, then asks what was actually measured on the ground. The radar work is already being done, and done well, by people who know their instrument. What tends to be missing is the place-level record where a satellite conclusion, a field reading, and the claim someone made about that reach sit next to each other without being averaged into one number nobody can defend. Optical imagery answers a different set of questions, and [what earth-observation data shows](/guide/what-earth-observation-data-shows?from=guide) covers that ground. ## frequently asked questions ### what is synthetic aperture radar? Synthetic aperture radar is an instrument on a satellite or aircraft that transmits a microwave pulse and records the return to build an image of the ground. Because the platform moves along its track, a short antenna can be processed as a much longer one. It works at night and through cloud. Copernicus Sentinel-1 is a C-band SAR mission. ### can radar measure soil moisture? Radar can produce a soil-moisture retrieval, which is not quite the same claim. A C-band figure describes roughly the top few centimeters, and vegetation, surface roughness, and the crop change it. It covers area no probe network can cover. It does not replace a probe at a known depth in a known profile, and it should stay labeled as a model. ### what is the difference between weather radar and sar? Weather radar such as NOAA's NEXRAD is S-band Doppler radar that estimates precipitation in the atmosphere — it looks up. Synthetic aperture radar such as Sentinel-1 images the surface of the ground — it looks down. Different band, different target, different product. Neither is a gauge in a channel. ### can a radar flood-extent map replace a stream gauge? No. A flood-extent map is a classification of a radar image showing where water appeared across a landscape. A recording gauge reports stage — water height — at one point through the whole event. A staff plate gives a dated reading each time someone looks. The map gives breadth, the gauge gives continuity, and when they disagree the disagreement is information. Keep both. ## read next **[how to plug a sensor stack in →](/guide/how-to-plug-a-sensor-stack-in?from=guide)** — the doors, the fields a note should carry, and what is not switched on yet. Or look at how a named place is held in the first place: [natural assets](/natural-assets?from=guide). ## sources [ESA — 10 ways Sentinel-1 data lets us see our world](https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-1/10_ways_Sentinel-1_data_lets_us_see_our_world) — Sentinel-1 C-band SAR, day and night, through cloud [Copernicus SentiWiki — Sentinel-1 applications](https://sentiwiki.copernicus.eu/web/s1-applications) — flood mapping and low return from calm open water [Water (MDPI) 2024, 16(1), 40](https://www.mdpi.com/2073-4441/16/1/40) — C-band soil-moisture retrieval depth, vegetation and roughness effects [NOAA NCEI — Next Generation Weather Radar (NEXRAD)](https://www.ncei.noaa.gov/products/radar/next-generation-weather-radar) — S-band Doppler radar estimating precipitation in the atmosphere ## the series 1. [what your sensors already measure](/guide/what-your-sensors-already-measure?from=guide) — five arrivals, one record 2. [a claim is not a reading](/guide/a-claim-is-not-a-reading?from=guide) — the sentence, the reading, the gap 3. [what mrv actually is](/guide/what-mrv-actually-is?from=guide) — three uses of one row 4. [the radio stays outside](/guide/the-radio-stays-outside?from=guide) — instrument, carrier, record 5. **what radar can see** — this post 6. [how to plug a sensor stack in](/guide/how-to-plug-a-sensor-stack-in?from=guide) — the doors and the note to send 7. [what lidar can see](/guide/what-lidar-can-see?from=guide) — a laser times a return; height is that distance