The open map

What the open map is

The open map is one anonymous, platform-wide map of every public sensor whose owner has chosen to list it — across every group, with no sign-in. It is where the Maps item in the public bar leads, it is the first card on the maps hub when you are signed in, it is linked from the landing page, and it is meant to be found by search engines. It draws the same page a saved map does: pins beside sensor cards, a plot window with the last week of readings and a statistics table on every pin. There are no GIS overlays on it; those belong to the maps a group builds for itself (Maps & the map builder).

Nothing appears on it by default. A sensor is listed only when all three hold: it is Public, it has coordinates, and List on the open map is ticked on its card.

Listing a sensor (opt in)

1. Open Settings → Sensors and find the sensor's card. You need edit rights in the Settings area for that sensor (the group owner has them; a member has them when their Settings permission is Read & edit).
2. Tick Public if it is not already — a private sensor is never listed, whatever else is ticked.
3. Make sure the sensor has a latitude and longitude (set on the card, or by placing it in the map builder).
4. Tick List on the open map and save. The sensor appears on the open map on the next page load; there is no cache to wait for.

To take a sensor off the map, untick the box and save, or make the sensor private — going private clears the listing, and ticking Public again later does not re-list it until you tick the box again. Removing the coordinates also removes it from the map.

What a visitor sees about a listed sensor

USGS gauges (streamflow conditions)

The map also draws every active USGS stream gauge in the country — about twelve and a half thousand — at once, as small dots, from a catalogue of published USGS coordinates this platform keeps and refreshes once a day. The layer is on by default; Hide beside it in the layers panel switches it off, and the choice is remembered in your browser.

The colour is the gauge's condition: how the latest flow compares with the flow recorded on the same day of the year across the gauge's whole approved record. Since September 2026 it is drawn on one scale shared by every river layer on this map — USGS, the NWS forecast points and the modelled overlay — so two pins of the same colour mean the same thing whoever published them. Much above normal (above the 90th percentile) is deep blue, above normal (76th–90th) light blue, normal (25th–75th) a dark grey-green, below normal (10th–24th) amber, much below normal (under the 10th) burnt orange; a gauge at or above an official flood stage is pink; a gauge reporting zero flow is purple. The record high and low for the day are drawn with the arm they belong to and named on the card. The pin's size says the same thing again: flooding and drought are the biggest marks and normal is the smallest, so nothing on this map depends on telling two colours apart.

Pale grey means one thing, everywhere: there is no current answer — and hovering the pin says why, without a click. A gauge may have under ten years of record, or none; it may measure stage or rainfall rather than flow; its flow record may have ended years ago; it may have no reading in the last day; USGS may have flagged the latest value for ice or equipment. Those were five different greys, whites and browns until September 2026, and none of them said which it was. What the symbols mean in the layers panel is the key, and Show only narrows the map to one condition, to ranked gauges, or to the gauges that were reporting at the last daily check.

The agencies' own colours are one click away. USGS WaterWatch's percentile ramp and the NWS's flood-category colours are those agencies' own, and a reader who has spent years with them can have them back: Agency colours, in the layers panel beside the key, repaints the river layers in each publisher's palette. The choice is remembered in your browser and travels in the page's address, so a link you send shows what you were looking at. The default is the shared scale because the two agency ramps contradict each other on the same map — the same red is USGS much below normal and NWS moderate flooding — and because neither survives colour blindness: WaterWatch's normal and below-normal are indistinguishable to a deuteranopic reader, which is about one man in twelve.

How the colour is made. Once a year this platform fetches, from the USGS Water Data Statistics API, each gauge's day-of-year percentiles, minimum and maximum of daily-mean discharge over its approved record, and keeps them. Once an hour it fetches every gauge's latest discharge value from the USGS Instantaneous Values service, one request per state, compares it with the thresholds for that day, and writes the class. So the colours are at most an hour old, and the line under the map says when they were last computed. Flood stage is not shown: that is National Weather Service data, not USGS's.

At a continental zoom the map holds some pins back, and says so. Zoomed out past a regional view, a station with no reading of its own in the last day is not drawn — at two pixels a pin those were an even grey wash rather than information. A button on the map beside the zoom controls says how many are being held back and brings them all back in one click, and it disappears at the zooms where nothing is held back. A station is drawn whenever it has a recent reading, whether or not anything can be said about how the water compares with its record.

What is held here, and what is not. This is the one place the platform keeps a USGS reading: the latest value, one per gauge, overwritten every hour. No history is kept, nothing is exported from it, no alert reads it. The statistics are USGS's published numbers, not measurements.

Click a gauge and the plot window opens — the same window one of this platform's sensors opens — with its last 14 days of discharge, gage height and rainfall as charts, one per parameter the gauge reports, and under each the latest, the low, the high and the mean. Above the charts sit the name, the latest value with the time it was measured, the condition, today's normal range for that gauge, and a link to its page at USGS Water Data. The two-week series is fetched by your own browser straight from the USGS Instantaneous Values service, for that one gauge, at that moment: it does not pass through this platform, and nothing is fetched for a gauge nobody opened.

NWS river forecasts (NOAA)

The third layer is the National Weather Service's own river picture: every gauge and forecast point of NOAA's National Water Prediction Service — about 12,900 — drawn at once, from a catalogue this platform refreshes every hour. Each dot is drawn on the same shared water scale as the USGS layer. Where the forecast office has an official category it wins outright: minor, moderate or major flooding is flooding, in pink, whatever a percentile would say. Where the office has published no category — and that is most of them — the pin takes the percentile of the USGS gauge the NWS pairs it with, so about five thousand two hundred gauges this map drew as one flat green now say how the river is. A gauge at action stage reads much above normal and one at the low-water threshold much below; out of service, no current observation and “below flood stage with no flow record” are all pale grey with the reason on hover. A red ring still means the NWS forecasts flooding there — a ring is a shape, not a colour, so it survives the repaint. The forecast office's own flood-category colours come back under Agency colours.

Click a gauge and the plot window opens with two weeks of observed stage (and flow where the gauge reports it), the official NWS forecast drawn dashed beyond the observations, and the flood stages ruled across the chart, so you can see at a glance whether the forecast crosses one. Above the charts: the observed value and its category, the forecast crest and when it is expected, the flood stages, which NWS office issues the forecast, and a link to the gauge's page at water.noaa.gov. The series is fetched by your own browser straight from the NWPS API for that one gauge, at that moment; it does not pass through this platform.

The flood override crosses to the USGS layer, and the percentile crosses back. USGS WaterWatch's “above flood stage” came from NWS data, and so does ours: where the NWS pairs a forecast point with a USGS site and reports the river at or above flood stage, that USGS gauge paints as flooding whatever its percentile rank would otherwise say. The pairing now runs the other way too: an NWS gauge with no flood categories of its own is coloured by the paired USGS gauge's rank. Neither layer invents a number; each says what the other already published.

Five more networks

Switched on under Browse sources, each drawn and read the same way as the layers above — a dot per station, coloured by what it is, and the plot window on a click.

NOAA tides & water levels is the Center for Operational Oceanographic Products and Services: about 300 stations on the coasts, the Great Lakes and some rivers. A click charts two weeks of water level against the station's datum, which the card names, because a level without its datum is a number rather than a level; where NOAA publishes NWS flood levels for the station they are ruled across the chart. Fetched by your browser straight from NOAA.

NRCS snowpack (SNOTEL) is the Natural Resources Conservation Service's snow network — about 900 SNOTEL sites plus SCAN soil-climate stations and snow courses, mostly in the western mountains. A click opens on snow water equivalent where the station reports it, because that is the number western water supply turns on, with buttons for snow depth, precipitation, temperature and soil moisture. Fetched by your browser straight from NRCS.

USBR reservoirs (RISE) is the Bureau of Reclamation's data service: about 1,000 federal reservoirs, canals and gauges, with storage, elevation, inflow and release series. Reclamation refuses cross-origin requests, so unlike the two above, a RISE series is fetched by this platform's server on your click and passed straight back — not stored.

USACE water management (CWMS) is the Corps of Engineers, district by district: reservoirs, dams and river locations with their operational time series. A location name is unique only within its district, so a location here is written district/name. Internal kinds — turbines, pumps, locks — are left off the map. Fetched by your browser straight from the Corps.

The Water Quality Portal is the odd one, and the map treats it differently on purpose. It federates USGS, EPA and state sampling programmes across more than a million monitoring locations, and it is not real time: someone takes a bottle to a laboratory and the result is published weeks later. So this layer asks only for the area on screen, above about zoom 8 — below that it says to zoom in rather than drawing nothing — shows sites sampled in the last three years, and offers no chart, because sampling results are not a continuous series. A click says what kind of site it is, who samples it, how many results it holds, and links to the Portal, where those results can be downloaded. Each area's answer is cached here for six hours, so the Portal's slow query is paid once rather than per visitor.

Ocean waves — the sea state, drawn as flow

One layer on this map is not made of stations. Ocean waves, under Reference layers in Browse sources and off until you switch it on, draws the modelled sea state as a field: thousands of particles, each travelling the way the swell under it is going, each coloured by the significant wave height there — blue for a flat sea through teal and mint to white at six metres or more — and each trailing a short tail of where it has just been. It is the picture a wind map gives, applied to the sea, and it is animated in your browser.

What it is. Significant wave height and mean wave direction from Open-Meteo’s marine API, which republishes the Météo-France MFWAM, ECMWF WAM and NCEP GFS-Wave models and chooses per point. This platform’s server asks for a fixed grid of coordinates rather than an image — there is no wave tile to fetch — caches each grid tile for three hours, since the wave models run every three to six hours, and shares it with every reader. Wave direction is the direction the waves come from, so a westerly swell runs east across the picture.

The grid is sampled no finer than 0.25° however far you zoom in. The model is finer than that, but the request is charged against a shared, metered allowance and a finer sample would spend nine times as much to draw interpolation; the layer says so on its card rather than pretending to a resolution it did not buy. It is switched off on every visit and not remembered, because it animates and it spends.

CDEC (California Data Exchange Center)

The fourth layer is California's own network: every active station of the Department of Water Resources' Data Exchange Center — about 3,100 — drawn at once from a catalogue this platform refreshes daily. A dot's colour says what the station mainly is: blue a reservoir (storage, elevation, inflow, outflow and releases), green a river or stream gauge (stage or flow), teal a rain gauge, white a snow course or pillow, brown-grey weather or something else, and pale grey a station whose sensors have not been read from CDEC yet. Show only narrows the map to one kind. The layer sits under State & regional in the layers panel.

Click a station and the plot window opens with two weeks of the sensor it is mainly for — a reservoir's daily storage, a river's flow, a rain gauge's daily total, a snow pillow's water content — and a button for each of its other sensors, so a click costs one request and you choose whether to make more. Above the charts: the basin, county and elevation, the operator, and a link to the station's page at CDEC.

How it reaches CDEC, and why that is different. DWR's servers refuse connections from this platform's hosts, so every request — the daily catalogue and the series a click opens — travels through a small relay the operator runs on Cloudflare, which needs a token that must never reach a browser. A series is therefore fetched by this server on your click and passed straight back, as the Monitor My Watershed layer does: nothing is stored, cached or written down on the way through, and nothing is fetched for a station nobody opened. Times are CDEC's, in Pacific time.

Monitor My Watershed (community stations)

Beside the USGS gauges, the map draws every station on Monitor My Watershed — the EnviroDIY / WikiWatershed data portal run by Stroud Water Research Center and LimnoTech — at once, as teal dots. There are about 1,250 of them in this platform's catalogue, registered by 133 separate watershed groups, schools and volunteers. The layer is listed under Research & community in the layers panel rather than under Federal, because the panel groups by who runs a source and these are not an agency network.

Most of them are quiet, and the map says which. This is a portal people register a logger on, and loggers get taken down; at the last count about a sixth were reporting. So a station's dot carries its state, read from this platform's own daily catalogue: solid teal measured something in the last three days, solid grey has reported before but not lately, hollow is a registered site with no measurements on record. Activity in the layers panel narrows the map to the first, or to the first two; Organisation narrows it to one group's stations.

Click a station and the plot window opens with its last 14 days of the first thing it measures — an EnviroDIY logger typically publishes water depth, temperature and conductivity, with battery voltage and signal strength as housekeeping — and a button for each of the others, so a click costs one request and the reader chooses whether to make more. Nothing is fetched for a station nobody opened.

One difference from the USGS layer, and it is worth knowing. A USGS series is fetched by your own browser and never touches this platform. Monitor My Watershed does not permit that — its servers send no cross-origin headers, so a browser can ask but is not allowed to read the answer — so a Monitor My Watershed series is fetched by this platform's server on your click and passed straight back to you. It is not stored, cached or written down anywhere on the way through, and it still only happens when you open a station. But this platform does see which station you opened, where it would see nothing at all for USGS.

These are not regulatory-grade measurements. Nobody here quality-assures them, and calibration, siting and maintenance vary entirely by the group running the station. Treat a community logger as what it is: a real instrument, run by volunteers, with no promise about it.

Want one of these stations as your own sensor, with its history stored, charted, exported and alerting? That is a different thing from the map, and the Sensors page describes it — register a sensor with Data Source = Monitor My Watershed and search for the station by name.

Flood maps: where it is mapped, and where water went

Two layers, added together on purpose, because they answer complementary questions and neither is a substitute for the other. FEMA’s National Flood Hazard Layer is where the hazard is mapped to be. The USGS high-water marks are where water was observed to go. Where the two disagree is usually the most interesting thing on the screen, so nothing here tries to reconcile them — both draw what their source published, and the difference is yours to read.

Both ship switched off. Turn them on under Browse sources, in the Federal group.

USGS high-water marks (STN)

A high-water mark is a physical observation: after a flood, a survey crew finds where the water actually reached — a seed line, a debris line, a mud or stain line on a wall — and surveys its elevation. It is the only direct measurement of a flood’s extent on this map: every other answer to “how far did the water go” here is modelled or regulatory, and this is somebody standing at the line with a GNSS receiver. (Eleven other layers on this map do measure the water — they measure its level or its quality at a point, not where a flood reached.) One of the first marks in California is recorded as “high water mark on side of the T-2986 transformer box close to the bridge”, at 11.44 ft NAVD88.

Three things travel with every mark, and none of them is decoration.

• The datum. An elevation with no datum is not an elevation. NAVD88 and NGVD29 differ by roughly a foot in California and are not interchangeable. Every card names the datum, and a mark surveyed against something that is not a geodetic datum at all — a local control point, say — is labelled as not comparable with the NAVD88 numbers around it, rather than quietly plotted beside them.
• The survey quality. USGS grades each mark with a tolerance in feet: Excellent is ± 0.05 ft, Good ± 0.10, Fair ± 0.20, Poor ± 0.40, and Very Poor is worse than that. About one Californian mark in seven is Fair or worse. Those are drawn smaller and paler deliberately, and Survey quality narrows the map to the sharp ones. A half-foot observation plotted identically to a half-inch one is how somebody ends up calibrating a model against the wrong number.
• The storm. A mark belongs to one named flood event on one date, and is never shown without it. It does not say “the water gets this high”; it says “the water reached this, on this day, in this storm”. Flood event narrows the map to a single storm.

An empty map here means one of several different things, and the line under the map says which. No marks in a state means no crew has been sent there to survey after a flood — that is a statement about where USGS has surveyed, not about where floods happen. That is kept strictly apart from “USGS did not answer” and from “you have not switched the layer on”.

FEMA flood hazard (NFHL)

The regulatory floodplain: Special Flood Hazard Areas, the floodway, base flood elevation lines, the FIRM panel grid. This is the map a lender means by “in a flood zone” and the one a community enforces.

Three map states, and the difference matters. FEMA publishes the NFHL as three separate services and Map state switches between them — one at a time, because a picture blending two of them would be true of neither.

• Effective — the map in force today. This is what a community enforces and what a determination cites.
• Preliminary — issued for review and comment. Not in force. It cannot be enforced and does not change anyone’s insurance requirement, but it is what FEMA currently proposes, so it is what the effective map is likely to become.
• Pending — past the appeal period, awaiting a Letter of Final Determination and an effective date. Not in force yet, and closer to becoming so.

The two non-effective states carry a banner saying so, which cannot be dismissed. A preliminary flood map is not the effective one, and the difference decides what a floodplain manager can enforce today.

The tiles come from FEMA to your browser directly, not through this platform, so no request of yours is relayed or recorded here. The consequence is that if FEMA’s service does not answer, the failure happens at your end — and an unanswered request draws exactly the same nothing as an area with no mapped flood hazard. Those are opposite facts, so the layer says which one it is in words rather than leaving you to guess at a blank.

Geology — bedrock and mineral records

The two newest layers are not networks, and the map files them apart from the rest: under Reference layers in Browse sources, folded away and off until you switch them on. Everything else on this map is an instrument that reports. These two are maps of the ground, and nothing about them changes from one day to the next — so instead of an age, the layers panel says static reference — no freshness beside each. That is not a missing timestamp. A bedrock map does not have one.

Why geology is on a water map. The rock sets the background chemistry before anybody samples anything. Carbonate country gives hard, well-buffered water; granite gives soft water with almost no buffering, and so a stream that is sensitive to acid; volcanic terrain is where arsenic and fluoride come from; marine shales carry selenium; serpentinite carries chromium and nickel; and sulfide ore bodies are where acid drainage and dissolved metals come from. A reading without its geology is a number with no baseline. Locally the case is sharper still: the Coast Range that Putah Creek drains is mercury mining country, and mercury from those workings is a live TMDL matter downstream in the Delta.

Bedrock geology (USGS) draws nothing until you click. Switch it on, click anywhere, and it names the geologic unit mapped at that point — its generalized lithology, its formation name and the state map the polygon came from — and outlines the polygon the answer came from, so you can see the actual shape, and how coarse it is. The source is the USGS State Geologic Map Compilation, stitched from fifty state maps at differing scales and vintages, so the vocabulary and the detail change at state lines. It is bedrock: much of what a creek actually touches is the soil and alluvium on top, which the compilation maps as “unconsolidated” where it maps them at all.

Mines & prospects (USGS) is the Mineral Resources Data System: mines, prospects and recorded mineral occurrences, each with the commodity it was worked for. A dot's shade says how far the ground was actually developed — palest is a recorded occurrence, which is most of the catalogue and means only that somebody wrote mineralisation down, and darkest is a producer or a plant. Commodity narrows the map to mercury, to the other metals that matter in water chemistry, to precious metals, to industrial and construction minerals, or to energy and geothermal; Development status narrows it to the places where ground was actually opened. A record worked for several commodities is filed under one of them, and its full list is on its card. The USGS stopped maintaining this compilation in 2011.

Soil and land cover (NRCS, MRLC) also draws nothing until you click, and it answers the question in the middle of the rainfall-runoff chain: how much of the rain falling here the ground can absorb before the rest of it runs off. Click, and it reports the available water storage in the top 150 cm of soil — a depth of water in centimetres, not a percentage — and the hydrologic soil group, A to D, which runs from sand that infiltrates freely to clay that sheds almost everything. Beside them it reports the impervious fraction: the share of the ground that absorbs nothing at all, which is what turns a storm into a flashy hydrograph in a built catchment. The soil comes from the USDA-NRCS soil survey via Soil Data Access and the surface from the National Land Cover Database. Either can be down while the other answers, and a half that failed says so rather than reporting bare ground.

It stops short of one obvious step, deliberately. Hydrologic soil group and land cover together are exactly the pair that indexes a curve number, and a curve number plus a rainfall depth gives a runoff depth — but that number is a model, not a reading. It commits to how wet the ground already was and to a treatment class the soil survey does not publish, and printing it beside the mapped facts would make a modelled guess look like a measurement. So this map gives you the inputs and names them as inputs; the model is yours.

Hot springs (NOAA NCEI) is NOAA’s 1980 Thermal Springs List for the United States: 1,661 springs, pools, geysers, fumaroles and mud pots at 20 °C or hotter, each with the hottest surface temperature ever published for it and the USGS and NOAA references it was compiled from. Dots are coloured by that temperature — warm, hot, very hot, boiling — and a filter picks out the ones hot enough to scald. NCEI decommissioned the database in May 2025 and never updated it after 1981, so this platform serves an archived copy and the layer says “static reference” rather than an age. Positions are from 1:250,000 overlays, to about a kilometre; test before you touch, and a dot is not permission to go there.

Geothermal systems (USGS) and Geothermal favorability (USGS) are the two halves of the USGS’s 2008 Assessment of Moderate- and High-Temperature Geothermal Resources of the United States. The first is the 253 systems the assessment identified — each with its estimated reservoir temperature, reservoir volume and mean electric power potential, coloured by reservoir temperature; these are estimates for a reservoir, not measurements at the surface, and a power potential is not a plant. The second is the assessment’s relative-favorability surface for undiscovered systems across the western United States, ten classes drawn as a translucent wash under the pins, with the USGS’s own caution on its card: it is meant to highlight areas of elevated potential and “not meant to be used to locate exact areas for exploration.” Both are fetched from USGS once a week and served from this platform’s copy.

Terrain — ground elevation, and which watershed you are in

Two more Reference layers, in the same fold and off until you switch them on, and they answer the two questions that come before every other layer here: how high is the ground, and which delineated watershed is this? Water goes downhill; a stage is a height above a datum, a rating curve is a statement about a channel’s geometry, and a hydraulic model is a terrain surface before it is anything else. Both say static reference — no freshness, for the same reason the geology layers do.

Ground elevation (3DEP)

Click anywhere and it reads the ground elevation there from the USGS 3D Elevation Program, in metres and feet, together with the resolution of the model it came from and the date that data was acquired. Resolution varies a great deal: 1 m where lidar has been flown, 5 m across much of Alaska, and about 10 m or 30 m from the seamless national models elsewhere. Every answer says which it used, because a number from a 30 m model and a number from a 1 m model are not the same kind of number.

Bare earth also means a reading in a forest is the ground under the trees rather than the canopy, and a reading on a bridge is usually the water or the ground beneath it.

Watershed boundaries — HUC12 (USGS WBD)

Click anywhere and it outlines the published 12-digit hydrologic unit containing that point, with its code, its name, its area, the unit it drains to, and the six nested units its code names — region, subregion, basin, subbasin, watershed, subwatershed. This is how water programmes in the United States are indexed: 303(d) listings, nutrient budgets, TMDL geography, state planning areas.

The boundary drawn here is the national display version. USGS publishes this geometry for regional and national analysis at 1:250,000 to 1:2,000,000, so a line on screen can sit a long way from the divide on the ground. It is not a property line and not a jurisdictional determination. And a subwatershed is not always a watershed in the ordinary sense: the type on each card says which. Standard units drain to one outlet, Frontal units drain directly to a coast or a large water body, Closed Basin units drain to no outlet at all, and Water units are the water body itself.

Asking about a place, rather than about a source

Every layer above answers from one source. Two controls, top left, ask all of them at once about one place and lay the answers out together.

Watershed answer takes a click and works out the land draining to it, from the USGS river index, then asks eight sources about that basin. Draw a shape is for the places that are not a watershed — a parcel, a project footprint, a study reach, a fire scar, a stretch of coast. Click corners on the map, press Finish shape (or double-click, or press Enter; Backspace removes the last corner and Escape starts again), and the same eight sources answer about the outline you drew.

Each section carries what kind of true it is — observed, modelled, regulatory, reference — how old it is, and which source it came from. A source that fails says so rather than disappearing: a missing section would read as “there is none of that here”.

The endangered species screen

One of the eight is an Endangered Species Act screen, and it lives here rather than in the signed-in platform because it is a question about a place, not about a reading. (The old page at /species-screen/ redirects here.) It consults four federal map services and reports each separately, with its own state: designated USFWS critical habitat, proposed USFWS critical habitat, NOAA Fisheries critical habitat, and the EPA compilation of listed-species range envelopes. Both agencies are asked on purpose — USFWS covers most listed species, but NOAA Fisheries administers the Act for anadromous and marine species, so a screen wired only to USFWS answers a Central Valley creek with confident silence about the salmon.

No occurrence points, ever. Every source here is a polygon or a line drawn at regulatory scale. Precise locations of rare species are withheld deliberately — California’s Natural Diversity Database and the other state Natural Heritage programmes restrict them, because a published point is how a rare plant gets collected and a raptor nest gets disturbed.

Finding your way around

The Layers panel, top right, is deliberately quiet at rest: it lists only the sources that are switched on, each with its symbol and how many of its stations are in view. Browse sources opens the full list, grouped by who runs each one — this platform, federal agencies, state and regional agencies, and research or community networks. Grouping is by operator rather than by what a source measures because most sources measure several things: a single agency network may carry stream gauges, reservoirs, snow and precipitation, and listing it four times would be worse than useless. What a station measures is a filter instead. One group is not an operator: Reference layers holds the maps of the ground itself, which report nothing and have no freshness, so the questions the operator grouping answers cannot be asked of them.

The key is behind What the symbols mean: shape says how a station is doing — whether it is reporting, and how recently.

Show only narrows what is drawn without asking this platform for anything new — filtering to streams, or to lakes and reservoirs, is done in your browser on the pins already there, and the panel says immediately how many of the stations in view match. Filtering by station type needs individual pins, so while the map is showing clustered counts the control says so rather than sitting there doing nothing. Filters are not remembered between visits, on purpose: a filter is a question you asked once, and a map that quietly hid half its stations weeks later would be a map you could not trust.

Search, top left, is the way in to this map: there is no list of sensor cards beside it, because a column naming this platform's sensors spends height the map wants. Type at least two characters and it answers with the matching sensors. Sensors on this platform are matched in your browser out of the rows the page already carries, so they appear the moment you stop typing, without asking this platform for anything. Pick one to fly to it, flag it and open its charts.

Whether the panel is folded away and which groups you left open are remembered in your browser, along with which layers are switched on. Nothing about that leaves your device; the Cookie Policy lists every key.