Changelog

For paths gauged in Wales we don't run the Environment Agency's warnings feed — it covers England only, and near the border it returns the wrong country's areas — so instead we point you to Natural Resources Wales, who issue the warnings there.

That link used to land on NRW's guidance page — a page about flood warnings, one click short of the warnings themselves. It now goes straight to their live Flood warnings and alerts service, the Welsh counterpart to the map England paths already link to: the warnings actually in force for the area, searchable, with a live map — so "check the live warnings for this path" now lands on exactly that.

If you use the site with a screen reader, searching for a path is now clearer. As you type in the search box, it announces how many paths match — "3 paths found", or "No matching paths" — so you know the list under the box has changed without having to go hunting for it.

Under the hood we also tidied up how the search box describes itself to assistive technology, so it no longer points at a results list that isn't there when nothing matches.

No change for anyone using a mouse or touch — the search looks and works exactly as before. This one's about making the quick "is my path flooded?" check just as quick however you get around the web.

In a dry spell almost every path we track reads a reassuring green Passable — which is honest about today, but says nothing about the winter. A towpath that goes under most Novembers looks exactly the same as one that never floods.

Now, when a path is passable (or we've no live reading) but its own river gauge's year of record shows it properly floods on some days, the answer card adds one plain line beneath the verdict:

Marston Cycle Path is currently passable. The river's 1.2 m below the level where parts of this path can start to flood. Over the past year, though, the river topped the level where this path usually floods on about 17 days, mostly in November to February.

It's read straight from the gauge operator's own archive — no forecast, no alarm, just the seasonal picture the live badge can't give — so you can judge whether a route is one to count on through the winter. The fuller month-by-month breakdown is still there under Through the year below.

If you've been down a path that was wet or flooded, you can now tell the next person how deep the water was — ankle-deep, calf-deep, knee-deep, or over the knee. It's an optional tap on the "Been down today?" report, and it only shows up once you've said it was wet or flooded.

We added it because that's the fact the whole site is really about: not just "is it flooded?" but "is this a wellies job, or should I go round?" A wellington comes up to just below the knee, so "calf-deep" and "over the knee" are two very different mornings — and until now the only way to say which was a free-text note that most people (understandably) skipped.

Depths people report show up in the From people on the ground list, right next to how they were travelling and the river level at the time. It's local knowledge, plainly attributed to the person who was actually there — we don't fold it into our own flood-line estimates, and we don't dress it up as anything more than what someone saw on the day.

The quickest way to check a route is to go straight to it — floodpath.co.uk/marstoncyclepath. But it's easy to add a stray hyphen (marston-cycle-path), half-remember the name, or slip a typo, and until now that landed on a plain "there's no page here".

Now, when the address is close to a path we track, the not-found page offers the ones you most likely meant — by name and place — as a single tap through:

Did you mean Marston Cycle Path — Marston, Oxford

If nothing's close, it stays out of the way and points you to search as before. No wrong guesses, just a shorter way back to the path you came for.

On a path that floods in winter, the Through the year section tells you how often the river got high enough to flood it last year, and when. On the routes that flood the most, that used to arrive as one long sentence carrying two different flood lines and two day-counts at once — a lot to take in at a glance.

Now the headline is the part that matters most, on its own:

Over the past year the river topped the level where this path is usually flooded on about 32 days, mostly in November to March.

and the near-miss days — when parts could start to flood but it wasn't properly under water — sit just below as a second, quieter line:

It also reached the lower level where parts of this path can start to flood on about 32 more days.

Same facts, read in two easy passes instead of one dense one.

In a dry summer every path looks safe — which tells you nothing about whether your route floods each winter. So path pages now carry a "Through the year" read: over the last year, on how many days did the river reach the level where this path starts to flood, and when in the year did it happen.

For a route that tends to hold up you'll see just that — the highest the river got, and how far short of the flooding line it stayed. For one that floods each winter you'll see the days it topped the line and the months they fell in, so you know what to expect before the season turns.

It's read straight from the gauge's own record against the level we estimate this path floods at — a guide to the season, not a forecast. Where a gauge is tidal, held in a weir pool, or simply doesn't keep a long enough record to say anything honest, we leave the section off rather than guess.

You check a path, and it's flooded. Now what? Until now you'd head back to the full list and hunt for something close.

Every path page now ends with the closest paths we track — nearest first, each with today's status and how far it is, so if your usual route is under water you can eye up an alternative (or the next stretch of the same towpath) without typing a thing. It's the same honest read as everywhere else: real status only, "No recent data" when a gauge is quiet, never a dressed-up guess.

If a path is off on its own with nothing else close by, we just don't show the section — no padding it out with somewhere half an hour's drive away.

Some rivers are held to a steady level by a weir or lock — the gauge barely moves until a flood actually tops the weir. On a couple of these the level we started flagging from sat only a few centimetres above where the river normally rests, so an ordinary summer ripple could read as partly flooded on a path that was completely dry.

The path on the Thames near Runnymede was the clearest case: its gauge sits around 0.26 m and we were flagging from 0.30 m. We now set that line above where the river actually rests day to day, so the badge reflects a real rise, not the gauge's normal breathing. It only ever moves the line up, never down — so this can only remove a false flag, never hide a genuine one — and community reports still pull it to wherever a path really floods.

The instructions we hand search engines (robots.txt) now steer them away from the site's internal endpoints — the form-submission and behind-the-scenes routes under /api/ that have no page to show — so crawlers spend their time on the path pages people actually search for. It also points them at our sitemap using the same canonical web address as the rest of the site, so the two can never drift apart.

No change to what you see — this is housekeeping that helps the site show up correctly in search.

On a path that hadn't been reported in the last day, the "From people on the ground" section used to say "No reports in the last 24 hours — if you've been down, you'd be the first" — even when there was an older report sitting right below it. So it could tell you you'd be the first at the same time as showing you someone who clearly wasn't.

Now, when there's an earlier report, it names how long ago it was — "the most recent was 1 month ago" — so you can weigh how fresh that local knowledge is, and invites you to add yours if you've been down since. A path that genuinely has never been reported still reads "No reports yet — if you've been down, you'd be the first."

When you tell us how a path was, we say thanks and promise it'll show up below. It just wasn't quite keeping that promise: your report — and the "N reports in the last 24 hours" tally next to it — only appeared after the page happened to refresh itself, up to a minute later. If you scrolled down to check straight away, as you would, it looked like nothing had happened.

Now the page pulls in your report the moment it's saved. Send it, and it's there in "From people on the ground" below, counted in the tally, without touching anything. And if you reported a flood that nudges the path's flood line, the status at the top updates in the same breath.

Small thing, but it's the moment that matters most: your bit of local knowledge is what keeps this honest for the next person, and you should get to see it land.

River gauges wobble a centimetre or two even when nothing is really changing — small eddies, wind, a passing tide. When a path's gauge happens to sit right on the level where we start flagging it, that ordinary wobble used to tip the badge back and forth: partly flooded in the morning, passable by lunch, partly flooded again by evening — all on noise, not on any real change to the path.

Now the status holds steady. A reading has to move a clear step past the line — not just brush it — before the badge changes, and it has to fall back a clear step before it changes back. A genuine rise still shows up straight away; only the jitter is filtered out.

One path on the Thames near Runnymede had been doing exactly this, sitting on its line in the recent dry spell. It now reads passable — with the Running high flag it already carried, which is the honest signal to keep an eye on it.

A path can be passable and still worth watching. Every gauge has an everyday range it sits in most of the year, and when the river climbs into the top of that range it's primed — a bit more rain could start to reach the path.

Now you can spot those at a glance, without opening each one:

• A Running high flag on the path card when the river is high in its gauge's usual range (but still below the level where parts start to flood)
• On the home page, when nothing is flooded or partly flooded, the paths running high lead the "Right now" section — the ones to keep an eye on

It's read straight off the Environment Agency's own published range for each gauge — no guesswork. Paths sitting low or in the middle of their range stay quiet, so the flag only shows when there's genuinely something to notice.

Each path page has a small chart of the river's recent level — it's there to show the shape of the last two days, because a river that spiked overnight and is already dropping is a different call from one that's been climbing steadily since yesterday.

On gauges that report every 15 minutes, that chart was quietly stopping short — it showed the most recent day and a half or less, even though the label was honest about it. It now covers the full two days on every gauge we track, still drawn only from real readings, never smoothed or filled in.

Two paths — Longrun Meadow in Taunton and the Barton Farm Riverside Walk in Bradford-on-Avon — had been showing "No recent data" since mid-June. Their gauges were fine the whole time: the Environment Agency had moved those readings to a new feed address, and the old address kept answering with the last reading it ever took, so our checks quietly went stale.

Both paths are back to live readings now, with the missed weeks filled in so the level chart and trend read normally again. The same slip had also cost a few other paths their backup gauges — those are restored too.

More importantly, the site now spots this for itself. If a gauge we track goes quiet for a few days but its station is still publishing the same series under a new address, we follow it there automatically — and we only carry a path's flood lines across after checking the old and new readings actually agree where they meet, so a genuine change in how a gauge measures can never smuggle in wrong flood lines.

If you've added Flood Path to your home screen, your phone resumes it exactly as you left it — check on Monday evening, open it Tuesday morning, and until now you'd be looking at Monday's answer. The "Updated 11 h ago" line was honest about that, but on an installed app there's no reload button to do anything about it.

Now the site quietly re-checks for you. Come back to it after more than a few minutes away — reopening the app, switching back to the tab, or your signal returning on the towpath — and it fetches the latest statuses on its own. A page left open on a shelf re-checks itself every half hour too.

It's deliberately silent: no spinner, no jump — your scroll position and anything you've typed into a report stay put. And it only ever makes the screen fresher; nothing is served from an offline cache. The morning glance now answers today's question, not yesterday's.

Most people don't file a report standing on the towpath — they cross at half seven and remember the site at their desk. Until now, a report filed at lunchtime was paired with the river level at lunchtime, even if the wade through the flood happened five hours earlier on a river that's been falling since.

The report form now asks one more thing: when was this? Just now, an hour or two ago, or this morning. "Just now" is preselected, so if you're reporting from the path itself nothing changes — no extra taps.

If you backdate it, we pair your report with the gauge reading closest to the time you were actually there (and if we don't have a reading within a few hours of that moment, we honestly store no level at all rather than a wrong one). The report's time on the page now shows when you crossed, too.

This matters because your reports are what tune each path's flood lines. A "wet but passable" matched to the wrong level teaches the model the wrong number — matched to the right one, it sharpens the exact boundary between a wellies day and a skip-it day.

When a path is partly flooded, the page tells you how much further the river would have to rise before it floods properly — the number that decides whether today's a wellies day or a skip-it day.

That line used to start by repeating that the river was already past the point where parts of the path can flood. But the status right above it — Partly flooded — already says exactly that, so you were reading the same thing twice before getting to the bit you came for.

Now it leads with the number:

The river's 20 cm below the level where it usually floods properly.

Same information, one fewer thing to wade through — and it reads straight on into the "when could it get there?" line just below it.

When a river is climbing toward the level where a path floods, the path page gives you a forward read — roughly when it could get there at the current rate. That line used to open by repeating the rise ("rising about 1 cm an hour") right above the trend tag that already says the same thing ("rising — up 4 cm over the last 6 h"). Same fact, twice, on the one card you're reading in a hurry.

Now it just tells you the part you can't get anywhere else:

At this rate, it could reach the full-flood line in about 5 hours.

The "how fast" stays in the trend tag where it belongs, and the forward line reads straight on from the headroom above it. Shorter card, same information — on the page you most want to be quick.

When a path shows up amber or red, that verdict comes from one of two places, and they don’t deserve the same confidence. Either people have actually been down and told us where it gets wet — so the flood line sits where they found it — or we’ve only ever had the gauge’s own history to go on, and “partly flooded” really means the river is running high for this gauge. The second kind is a fair heads-up, but it isn’t the same as someone standing in it telling you to bring wellies, and we shouldn’t have been wording it as if it were.

So on a path whose flood line hasn’t been tuned by reports yet, an amber or red verdict now carries a plain line under it saying exactly that — the river is high against this gauge’s record, no one has confirmed it on the path itself, and a quick report from anyone who goes down is what turns the guess into the real thing. A path whose line has been calibrated from reports gets no such caveat: there, the badge means what it says.

This matters most in a dry spell, when a gauge can top its yearly-high mark for a day without the path actually flooding. Rather than cry wolf, we’d rather tell you plainly what we know and what we don’t — and ask for the one report that settles it.

A while back the path page started leading with the margin — not just passable / flooded, but how close: how much room there is before the river reaches the level where the path starts to go under. That's the call this site is really for (dry today, or one wet night away? wellies, or skip it?).

Now the same number rides on every card — the home page, the all-paths list, and your saved paths — right under the status:

Riverside Valley Park · Passable 94 cm of room

Thames Path at Runnymede · Partly flooded 34 cm below fully flooded

So a quick scan down your usual routes tells you not just which are clear, but which are getting close — a path with a metre of room reads very differently from one sitting a few centimetres under the line, even though both say "passable". Room to spare shows calm green; getting wet, amber; over the line, red.

The honesty rule carries over unchanged: the margin only appears off a current reading with a known flood line. If the nearest gauge has gone quiet, the card says "No recent data" and shows no margin — never an old number dressed up as now. No new data was fetched for this; the flood lines were already there behind each card's status, just not shown.

The decision this site is for is rarely a flat yes/no — it's how close. Dry today, or one wet night away? Wellies, or skip it altogether? That call runs on the margin: how much room there is before the path starts to flood.

So the path page's headline answer card now carries that margin right under the status, beside which way the river's heading and when it last updated — the whole decision in one glance, without scrolling:

Marston Cycle Path is currently passable. The river's 67 cm below the level where parts of this path can start to flood.

When a river is genuinely climbing toward that line, the same card adds the forward read — "if it keeps up, it could reach the level where parts of this path can start to flood in about 5 hours" — so a rise you'd want to plan around is the first thing you see, not something you find further down.

Nothing was added twice over: the margin used to sit lower down in the gauge read-out, and it's simply moved up to where the answer is. The gauge card below still has the full picture — the exact reading, the meter, the last two days' trace, and a link to the Environment Agency's own page for the gauge. And the honesty rule is unchanged: if the nearest gauge has gone quiet, there's no margin and no projection — the page says it can't tell you either way rather than dressing an old reading up as current.

The bottom of the home page used to carry a "Recently added" row. In practice the whole catalogue was added in one go when the site launched, so that row showed the same handful of paths to everyone, every visit — dressed up as fresh news when nothing had actually changed.

It now does something genuinely useful instead: a Browse by area grid of every region we cover, with the live count of paths in each — and, when a river is up, the worst flood flag in that area right there on the tile, so you can jump straight to the part of the country you care about.

"Recently added" hasn't gone away — it comes back automatically, and only when it's true: whenever a path is genuinely added in the previous week, the newest ones show up there for a few days, then the area grid takes the slot again. No more standing in for growth that isn't happening.

The "rising / falling / steady" line on each path — and the "if it keeps up, it could reach the flood line in about N hours" note that goes with it — now reads the river more carefully.

Before, the direction was worked out from just two readings: the latest one and one from about six hours back. That's fine when the river is moving cleanly, but two readings is a thin thread to hang a forecast on — if either one is a blip, the whole "rising" call (and the hours-to-flooding estimate) goes with it. We saw exactly that on a gauge that had spiked and already dropped back: read off two points it looked like a steady climb toward the flood line, when in fact the water had just fallen.

Now we fit a line through all the recent readings, not two of them, so a single odd reading no longer swings the verdict. And — this is the important part — when the readings are bouncing around too much to draw an honest line through (a gauge glitching, or one on a tidal stretch swinging up and down with the tide), we simply say nothing about the direction rather than guess. The river-level number and the two-day shape strip are still right there; we just don't put a confident "rising" or a countdown next to a reading that can't support one.

A calm, flat river still reads "steady" as before. The change only ever makes the trend quieter and more trustworthy — never louder. </content>

The path page already tells you how much room is left — "12 cm below the level where parts of this path can start to flood". That answers right now. The question on a wet morning is the next one: is it creeping, or about to close the gap before I'm back this evening?

So when the gauge is genuinely rising toward a flood line, the page now adds a plain forward line under the level: "Rising about 5 cm an hour. If it keeps up, it could reach the level where parts of this path can start to flood in about 4 hours." Enough to judge whether to leave now, wait it out, or take wellies.

We've kept it honest:

• It only appears when the river is actually rising — never to reassure you off a falling one, and never on a gauge that's gone quiet.
• It only looks a little way ahead. A river isn't a ruler, so if the line is more than a few hours off at the current rate, we say nothing rather than guess at tomorrow.
• The wording is deliberately conditional — "if it keeps up" — because that's exactly the assumption. The river can speed up or ease off; the official gauge page (linked right there) has the EA's own forecast where there is one.

On a dry day, you won't see it at all — which is the point.

If you check the same towpath or cycle path most mornings, the quickest possible answer should be one tap away — not a scroll down the home page.

Your saved paths now live at their own address: floodpath.co.uk/saved. It shows every path you've starred on this device, most urgent first (anything flooded or partly flooded leads), with a one-line summary across the lot. It's a clean URL you can bookmark or add to your home screen, so the morning check goes straight to your routes.

Two more ways to get there:

• Once you've saved a path, a Saved link appears in the top menu — with a little count — on every page, so your routes are one tap away from anywhere on the site.
• If you've added Flood Path to your home screen, press and hold the icon and you'll get a Saved shortcut that jumps straight in.

Nothing about saving changes: your starred paths still live only on your device, no account and no tracking. The new page just gives them somewhere proper to live. Haven't saved anything yet? The page explains how — open a path you use and tap Save path.

The whole point of this site is a single everyday call: can I take my usual path today, do I need wellies, or should I go round? So the reports people leave now match that decision exactly.

When you tell us how a path was, there are three choices instead of two: got through fine, wet but passable (wellies, or you rode through standing water), or flooded — the same three states we already use to describe a path. "How did you cross?" is now just on foot or by bike, kept separate from how wet it was, so a cyclist who splashed through isn't logged as wearing wellies.

It changes what the next person sees, too. The tally used to read "2 passable" — which hid the thing you actually want to know. Now it reads "2 got through (1 wet)", so at a glance you can tell whether people strolled it dry or waded it in boots.

That middle "wet but passable" report is also the one that teaches us where a path starts to flood, not just where it's impassable — so the more people pick it, the sharper both flood lines get, in good time for the winter. Reports left before today still show and count exactly as before.

When you tell us a path was flooded or passable, we line that up against the river level the nearby gauge was showing at that moment — that pairing is what slowly teaches each path the level at which it actually floods.

But gauges sometimes drop offline, and that tends to happen during the worst weather, exactly when people are reporting. If a gauge had gone quiet, the most recent reading we held could be a day or two old — not the level that was really there when you crossed. Pairing a fresh "flooded today" against a stale reading like that would nudge the flood line toward the wrong number.

From now on, if the backing reading was already more than 24 hours old when a report came in, we still keep and show your report, but we leave it out of the threshold maths rather than learn the wrong lesson from it. It's the same 24-hour "is this still current?" rule we already use to decide whether to show a river level as live. No change to anything you'll see day to day — just a more careful model underneath, in time for the winter.

The official flood-warning banner we show on English path pages comes from the Environment Agency — and the Environment Agency only covers England. On our seven Welsh paths that feed was the wrong tool for the job: near the border it hands back warning areas across the estuary or downstream in England (a Somerset coast area turns up within a few miles of Cardiff, a Gloucestershire stretch of the Wye near Monmouth) that have nothing to do with the Welsh path you're looking at. At best it showed nothing; at worst it could have flagged a warning for the wrong place.

Flood warnings in Wales are issued by Natural Resources Wales, on their own service. So Welsh path pages no longer run the English feed at all — instead they carry a plain line telling you where the live warnings for that area actually live, with a link straight to NRW's "Check flood warnings" service.

Your river level and our flood lines on these paths were always from the right place — they come from the nearest NRW gauge. This only fixes the warnings layer, so an Abergavenny or Brecon page never points you at an English warning that doesn't apply, and always points you at the Welsh one that does.

The "Your saved paths" row on the home page is the quick check for people who come back every day — your own routes, in one place. It's now ordered the way you'd actually want it: anything flooded comes first, then partly flooded, then the ones that are passable, with "no recent data" last. If you've saved a handful of paths and one of them has gone under, it's the first card you see — not buried below the ones that are fine.

There's also a one-line summary at the top of the row — "3 paths · 1 flooded · 2 passable" — so you get the whole answer at a glance before scrolling through the cards. Same honest counts as the rest of the site: a path with no recent river reading shows as "awaiting data", never dressed up as passable.

Nothing changes about how you save paths — star any path and it shows up here, and your starred list still lives only on your own device.

River gauges report their level in metres, but against different reference points — and the codes the feeds use for that ("mASD", "mAOD") are hydrology jargon. On a dry, perfectly passable day a path could read 1.74 mASD, which is easy to misread as "1.74 m of water on the path" when it actually means the river's height at a gauge a few hundred metres away, measured from that gauge's own zero point.

Two small fixes:

• The prominent reading is now labelled "River level at the gauge" (not "Water level"), so the number reads as what it is — the river's height at the gauge, not how deep the water is on your path.
• The unit codes are kept exactly as the gauge operator publishes them (so they still match the official page you can open from "see the EA's own readings"), but each now carries a plain-English explanation: mAOD is metres above sea level, mASD is metres above the gauge's local zero. Hover or long-press the unit to see it.

The number that actually matters for the "wellies or skip it?" call — how much room is left before the path floods — has always been spelled out in plain centimetres, and still is.

When you share a path — with the new Share button or just by pasting the link — the preview that pops up in WhatsApp, Messages, Slack or X now shows that path: its name, where it is, and the river it runs along, on its own card.

Before, every shared link unfurled to the same generic site card, so your mates got a preview with no idea which path you meant until they tapped through. Now the card names it straight away.

The card itself doesn't show a flood verdict, on purpose. A link preview gets saved by the chat app and can sit in the conversation for days, so a "flooded" stamped onto it could be long out of date by the time someone scrolls back to it. Instead the card invites you to check, and the link takes you to the live page where the status is always current. (The shared message still carries the right-now status — that's written fresh each time you share.)

Half the reason this site exists is so you can warn other people. If you get to the towpath and it's under water, the useful thing is to tell everyone who walks or rides it before they set off too.

Every path page now has a Share button next to Save. On a phone it opens your usual share sheet — WhatsApp, Messages, wherever your group chat lives — with a one-line note on how the path is doing right now and a link to the live page. Tap it while you're standing at the flooded bit and "Marston Cycle Path is flooded right now" lands in the chat in two taps.

On a computer, where there's no share sheet, it copies the link instead and tells you it's done.

The note only ever states a status we can stand behind on a current reading — if the gauge has gone quiet we just say "check whether it's flooded" rather than claim either way. And the link always points at the live page, so whoever you send it to sees the real, up-to-the-minute state, not a frozen snapshot. No accounts, no tracking, no share counter — the sharing is your browser's, we just hand it the words.

The all-paths page groups every path we track into ten areas, with a row of jump links at the top. On a dry day, when everything reads passable, that's all you need. But when one or two paths flood — the day the site is actually for — you'd have had to scan every card in every area to find them.

Now the directory does that scan for you. When an area has a path that's flooded or partly flooded, its jump link gets a small coloured dot (red for flooded, amber for partly flooded), and the area's heading spells out the count — "1 flooded · 2 partly flooded" — so you can jump straight to the section that matters instead of reading past the ones that don't.

Nothing changes on a day when every path is passable: no dots, no extra wording, the same clean list as before. The flags only appear when there's genuinely something to flag, and they always match the live status shown on the path cards themselves.

The flood lines on each path tell you how close the river is to the level where the path starts to flood. But on an ordinary day, well below that line, there's a question they can't answer: is the river sitting low for this gauge — lots of room, a wet night won't reach the path — or already running high in its everyday band, where a bit more rain could close the gap?

Every Environment Agency gauge publishes its own "typical range" — the band the level sits in the bulk of the year. Today each path's water-level card reads your reading against that band and says it plainly:

• "The river's running low for this gauge — below its usual range."
• "The river's around the middle of this gauge's usual range."
• "The river's sitting high in this gauge's usual range."

It's the gauge operator's own published range, not a guess of ours — so a level like 0.07 m that sounds low can honestly read as "high in its usual range" when that's where it sits for this particular gauge. That's the extra bit of foresight for the "wellies tomorrow, or skip it?" call.

We only show it where we can prove the published range sits on the same scale as the live reading; where we can't, we stay quiet rather than risk a misleading comparison.

The whole point of this site is the call every British commuter makes at the door in winter: wellies, or find another way round? A path page gives you two things to make it with — the live river level at the top, and below it the reports from people who've actually been down ("passable, in wellies, 3 hours ago"). Until now those two halves never spoke to each other. You could see the river is at 1.65 m right now, and that someone got through in wellies this morning, but not how high the river was when they did — so you couldn't tell whether their morning was wetter or drier than yours.

It turns out we already knew. Every report quietly records the gauge's reading at the moment it's filed — that's how the system learns each path's flood line over time — we just never showed it. Now each report carries the level it was made at: "Passable · in wellies · river was 1.79 m · 3 hours ago." Read it against the live number at the top and you have a real yardstick — they waded it in wellies at 1.79 m, it's 1.65 m and falling now, so you're likely fine on foot.

We only show the level when it was read on the same gauge the page's live level and flood lines use, so the numbers are always on the same scale and a comparison is honest — if a backup gauge happened to be standing in at the time, that report simply shows without a level rather than print a figure you can't fairly compare. No new data collection: this is a fact we already held, finally made visible.

Nothing changes on the page, but search engines now get a machine-readable summary of what each page is — so a result for a path is more likely to show its place in the site, and a search for "Flood Path" is better understood as this site.

Every page previously handed crawlers only its title and description; there was no structured data at all. Now the home page identifies the site itself (its name and the floodpath.co.uk address), and each path page carries two things: a breadcrumb trail (Flood Path → All paths → this path), which is what lets Google show that trail under a result, and a plain description of the path as a real place — its name, the route description we already show, its locality and its coordinates. Only facts the page already states out loud; no invented ratings, reviews or opening hours, in keeping with the rest of the site.

Under the hood it's a small, pure, unit-tested builder (src/lib/structured-data.ts) rendered as a standard application/ld+json script. The serialiser escapes the data so a path description — authored by us today, but arriving from a community suggestion tomorrow — can never break out of the script element, closing off a cross-site-scripting route before it could open.

A path's flood line is only as good as the gauge reading it's anchored to. For most of our gauges that anchor comes from the Environment Agency's published "typical range" — built from years of observations — or, for Welsh gauges, from a full year of the gauge's own readings. But a handful of gauges publish no such long-term figures, and those paths had to start from a weaker guess: the river's level over the last couple of weeks. Seed that in a dry June and the line sits just above normal summer flow, ready to cry wolf at the first ordinary autumn rise.

Those gauges turn out to have a deeper archive available after all — the EA's Hydrology service holds years of their readings, lagging real-time by only a few weeks. We now pull a full year of that history (with a winter in it) and set the warn line where the river only sits about 5% of the time, the same honest statistic the published figures approximate. Eleven gauges moved across, including the main gauge behind the Chippenham, Monkton Park path, whose warn line had been sitting almost exactly at the river's median level — roughly half the year above it. It's now set where high water genuinely begins.

Two safeguards keep this honest. We never overwrite a line that community reports have already tuned. And because the archive is a separate record, we only trust it once we've checked it reads on the same datum as the live gauge — matching their recent readings before believing a single number; the few gauges that don't match, or that aren't in the archive at all (locks, sluices and pump stations mostly), keep their existing estimate rather than borrow a wrong one. As always, every path page still says plainly where its flood line came from.

Each path page already told you which way the river was heading and how fast — "rising, up 9 cm over the last 6 hours". Now it shows you the shape of the last couple of days, as a small line under the water-level reading.

That shape is the bit a single rise-or-fall number can't tell you. A river that spiked overnight and is already on its way back down is a very different wellies-or-skip call from one that has climbed steadily for two days to reach the same level — even though both might read "rising" right now. The strip lays that out in one look.

It's drawn against the same colour bands as the level meter: green below the level where parts of the path start to flood, amber between there and a proper flood, red above. The flood lines only appear on the strip when the river is actually trading near them — so on a dry day it's just a calm line low in the green, and when the water climbs toward the flood line you'll see the line nose up into the amber, in context.

It plots only the real readings from the gauge — no smoothing, no guessing where the line goes next. When the gauge has gone quiet, the strip steps aside, the same as the rest of the live read-out, rather than implying a trajectory we can't stand behind.

The search box is the first thing most people touch here, so today it grew the manners a search box should have had all along — on a phone and on a keyboard alike.

• The "Go" key now goes. Type a path on your phone, hit the keyboard's Search/Go button, and you land on the top match — no reaching back up to tap the row. Before, that button did nothing and you had to tap; now the obvious gesture just works.

• Arrow keys, Enter and Escape. On a keyboard you can run down and up the results, open the highlighted one with Enter, and dismiss the list with Escape — without losing what you typed. Tapping anywhere off the box closes it too.

• Properly announced to screen readers. The box is now a real combobox: assistive tech is told there's a list of matching paths, how the active one is highlighted, and which one you're on — so this works the same whether you read the screen or listen to it.

Same results, same live status on each row — just reachable however you get around.

Type a path into the search box on the home page or the all-paths page and each result now carries its live status right there — Passable, Partly flooded, Flooded, or No recent data — so the usual "is my route OK this morning?" is answered in the dropdown, without a click through to the path page.

The status shown is the same live reading that powers the rest of the page, and it follows the same honesty rule as everywhere else on the site: if the nearest gauge has gone quiet, the result reads "No recent data" rather than a stale verdict dressed up as current. Open the path itself for the full picture — the exact water level, how much room is left before it floods, which way the river is heading, and what people on the ground have reported.

Until now the search results showed only a name and a chevron; the live status was sitting in the page the whole time, just not surfaced where you'd look first.

The home page used to open with a "Your saved paths" panel. Because your starred paths live only on your own device (we keep no accounts), the page couldn't know about them until it had loaded — so the first thing most people saw, right under the headline, was an empty "No saved paths yet" card where the live national picture should have been. Even if you did have saved paths, you'd catch a flicker of that empty card before yours appeared.

Now that panel only shows up once your device confirms you've actually saved something. If you have saved paths, they're still the first thing you see — no more flicker. If you haven't, the page goes straight to "Right now across Britain": how many tracked paths are passable, and any that are currently flooded. Either way you reach the useful part faster.

Saving paths still works exactly the same — star any path on its page, and it turns up here for a one-glance check before you head out.

Every path's status comes from a real river gauge, and now you can go straight to that gauge's own official page to see the full picture for yourself.

Under the water level on each path, there's a new link — See the Environment Agency's own readings for this gauge, or Natural Resources Wales's in Wales. It opens the gauge's authoritative page, where you'll find the official five-day graph (and, for Environment Agency gauges, their own short forecast).

We've always named the gauge behind each path and shown how far away it is; this just closes the loop, so nothing we tell you is something you can't trace back to the source. A handful of older gauges don't have a public page to link to — there we still show the gauge name plainly, same as before.

A site you glance at from the saddle lives and dies by its timestamps, so today's pass made every one of them trustworthy and made the path page's answer read in one line.

• British Summer Time, fixed. Absolute times — "Last reading Thu 11 Jun, 14:30" and the like — were rendered in UTC, which reads an hour early all summer. Every absolute time on the site is now pinned to UK clock time, whatever the season.

• "12 min ago" now keeps itself honest. Pages are cached for up to half an hour, so a freshly loaded page could claim a reading was "12 min ago" when it was really 40. Relative times now re-check themselves against your device's clock the moment the page loads and every minute you keep it open — and hovering (or long-pressing) one shows the exact UK time it stands for.

• The answer card now completes the glance. A path page's status card used to give you the verdict and a date-and-time you had to do mental arithmetic on, with the river's direction of travel living further down the page. It now reads in one line: Passable · Updated 12 min ago · Rising — up 9 cm over the last 6 h. The water-level panel below keeps the reading, the meter and the flood lines, without repeating the trend.

Nothing about today's work changes what a path page tells you — it's the plumbing a site should have before winter traffic arrives.

• Security headers on every response. The site now tells browsers it must never be embedded in another site's frame (which blocks clickjacking tricks), that files are exactly the type they claim to be, and that the only device capability it will ever ask for is your location — and only for the "near me" sort you trigger yourself. Camera and microphone are explicitly switched off at the browser level.

• One true address. floodpath.co.uk and www.floodpath.co.uk both work, but everything we hand to search engines — the sitemap, social-share links, and a new canonical tag on every page — now agrees on one of them. That stops search engines treating the two as rival copies and keeps the path you search for ranking as a single page.

• A wrong address now gets a useful page. Mistype a path's link (or follow a stale one) and instead of a bare error you get a proper page saying we don't track a path at that address, with a button straight into the path directory and its search.

• Small wins for keyboards and slow connections. Keyboard and screen-reader users get a "Skip to content" link as the first tab stop on every page, and the text font starts downloading earlier so pages settle faster on a weak signal.

This site began life on a commercial page template, and a few bits of that costume were still showing. They're gone now.

The footer's social-media icons linked to accounts that don't exist, and its Contact, Terms and Privacy links went nowhere at all — for a site that asks you to trust how it handles your reports, that wasn't good enough. There's now a real privacy & terms page that says plainly what we store (your reports and suggestions), what never leaves your device (your starred paths and your location), and what we don't do (no accounts, no advertising, no tracking cookies). Contact now reaches a real address: hello@floodpath.co.uk.

We also cleared out the template's leftover stock photos and third-party logos that were still quietly being served from this domain, pointed the footer's "report a path" link at the actual suggest a path flow, and corrected the about page to credit both of the agencies whose gauges power the site — the Environment Agency in England and Natural Resources Wales in Wales.

One wording fix in the same spirit: a path in good shape now reads "currently passable" rather than "open and passable". River gauges tell us about water, not about locked gates or closures — we won't claim knowledge we don't have.

When we anchored flood lines to gauge history on Tuesday, Wales got left behind: Natural Resources Wales gauges don't publish the long-term statistics the Environment Agency does, so all nineteen Welsh flood lines stayed pinned to a few weeks of June readings — centimetres above quiet summer flow, primed to cry wolf at the first proper autumn rise on the Taff, Usk or Wye.

NRW does serve the readings themselves, though — a full year of them, measured every 15 minutes. So now we fetch that year and work out the same statistics ourselves: the warning line sits at the level the river only topped about 18 days in the past year, and the flood line a third of the way up to the year's peak. It's exactly the formula English gauges use, computed from the same river the path sits beside.

What that did to the lines, concretely:

• Monmouth's Wye riverside: warning line 1.12 m → 3.47 m (the Wye at Monmouth routinely runs over 2 m in an ordinary winter without the path caring)
• Abergavenny Castle Meadows on the Usk: 1.01 m → 2.31 m
• Pontypridd's Taff Trail: 0.93 m → 1.43 m
• All nineteen Welsh gauge thresholds re-anchored; every Welsh path page now says its flood line starts from the gauge's measured history

The honest details, as ever:

• A year is less history than the EA's five, so these anchors are first guesses with the same modest confidence as the English ones — reports from the ground still outrank them and tune each path's line directly
• We refuse to anchor when the year is too thin to contain a real winter (long outages, or a series that doesn't span the seasons) — those gauges keep the cautious estimate and the path page says so
• A single glitched sensor reading can't distort the line: the "year's peak" ignores lone spikes, since a real flood lasts hours, not one reading
• We also unlinked a backup gauge in Pontypridd that hadn't produced a reading in over a year — it could never have served as a fallback

Nothing changes about how reports work. If you're on the Taff Trail in wellies this winter, thirty seconds of "still passable" or "flooded" makes your path's line sharper than any statistic can.

Until today we tracked eight paths, all around Oxford. Now we cover 56, from Cricklade to Carmarthen.

• England: the Severn's flooding hotspots (Worcester, Shrewsbury, Bewdley, Ironbridge, Tewkesbury, Gloucester), the Thames from North Meadow to Hurst Park, York's riverside walks, the Calder Valley, the Trent, the Cam and Great Ouse, the Bristol Avon, the South West, and flood-prone canal towpaths from the Kennet & Avon to the Weaver
• Wales, live for the first time: the Taff Trail at Pontypridd and Cardiff, Brecon Promenade, Carmarthen Quay, The Groe at Builth Wells, Castle Meadows at Abergavenny and the Wye at Monmouth — powered by Natural Resources Wales river gauges, joining the Environment Agency network we already read
• Every path earned its place: documented flooding history (council closure notices, flood-warning areas, local news), a named route people actually use, and a verified river gauge nearby on the right river — we dropped more than twenty candidates that couldn't meet that bar
• Several paths link to the exact gauge their authorities watch: Shrewsbury's towpath reads the Welsh Bridge gauge the council uses to close it, and Brecon's gauge stands on the promenade itself

New paths start with cautious estimates and sharpen as you report — if you walk one of these and the status looks wrong, hit "Been down today?" and tell us. That's how the thresholds learn.

Every path's flood line — the river level where we flag it as flooding — has to start somewhere before anyone's reported from the ground. Until now that starting guess came from the last few weeks of gauge readings. Seeded in a dry June, those lines sat a few centimetres above normal summer flow: two paths were showing partly flooded this week on rivers behaving completely normally, and come the first ordinary autumn rise, most of the rest would have cried wolf too.

Now the starting estimate is anchored to each gauge's published long-term record — the top of its typical range (the level the river only tops a few days a year, from roughly five years of data) and its highest level on record. On the Severn at Bewdley, for example, the warning line moved from 0.70 m (June-calm plus a whisker) to 3.30 m (where the river genuinely leaves its usual range).

The honest details, as ever:

• 95 of 128 gauge thresholds were re-anchored. The rest keep the old recent-readings estimate because their gauge doesn't publish a usable long-term record — each path page says plainly which kind of estimate it has
• Community reports still outrank everything: a flood line tuned by your reports is never touched by this, and one-sided evidence ("still passable at 1.79 m") still pushes the line clear of levels people have actually walked
• A sanity check rejects gauge records that don't match what the gauge actually reports (it happens — one York gauge publishes a "typical high" above its all-time record), falling back to the cautious estimate instead

Nothing changes about how reports work — every flooded-or-passable report from the ground keeps tuning the line for its own path. This just makes the starting point honest about winter, not calibrated to a quiet June.

River gauges occasionally stop reporting — maintenance, telemetry faults, the usual. Until now that meant the path it watched showed No recent data until the gauge came back, even when another live gauge sat a few hundred metres away.

Every path now has up to two backup gauges linked alongside its usual one, nearest first. If the usual gauge goes quiet for more than a day, the status switches to the nearest live backup automatically — and switches straight back when the usual gauge returns.

Full transparency, as always:

• The path page tells you it's a backup, which gauge it is, how far away it sits, and when the usual gauge last reported.
• Levels at different gauges sit on different scales, so each backup gets its own flood line, learned separately from your reports. A report you send while a backup is active calibrates that gauge — nothing gets mixed up.

This took Osney Thames Path straight from No recent data back to a live status: its usual gauge has been silent since 8 June, but the downstream gauge at the same lock is alive and well, 180 m from the path.

Until today, the paths we track were the ones we picked. Now you can add yours.

• Hit Suggest a path in the menu (or the button on the paths page) and you'll get a map of Great Britain — zoom in to your path and tap along it to trace the stretch that floods
• Tell us in your own words how it floods — where the water sits, how often, how deep. That description is exactly what the reviewer reads, so landmarks help ("calf-deep by the footbridge" beats "gets wet")
• Suggest a name if you like, or leave it to us
• Searched for a path and found nothing? The search now offers to turn that search straight into a suggestion

Every suggestion gets reviewed before going live: we check the route against the river, make sure there's a level gauge near enough to read, and settle on a name. Once it's in, it works like every other path — live river readings, a starting flood estimate, and your reports tuning it from day one.

No account needed, as ever. An email is optional and only used if we have a question about your path.

Two improvements to how much you can trust what you see:

• Fresher data, automatically. If our stored river levels are more than two hours old when you load a page, we now top them up from the Environment Agency on the spot. Previously a scheduling quirk meant levels could be a day old by the evening commute — that's gone.
• "No recent data" instead of a guess. River gauges occasionally go quiet. When the latest reading is more than a day old we now say No recent data and show you the last reading and when it was taken, rather than presenting an old verdict as current. The same applies anywhere we simply don't have a reading — we'll never invent a status.

Each path's flood line is tuned by your reports — and from today, not every report counts the same.

• Recent reports carry more weight than old ones. Paths change — drainage gets fixed, banks erode — so if a report from this morning disagrees with one from two winters ago, this morning wins
• Old reports fade gradually (half their weight after about six months) but never vanish entirely: on a path nobody has reported in a while, an old report is still the best evidence there is
• A report sent while an official Environment Agency flood warning was in force counts extra — the warning independently backs up what you saw
• The confidence we show is earned from that weighted evidence, so it honestly drifts down as the reports behind it age, instead of staying frozen at a number from last winter

Same deal as always: no accounts, no tracking — just river readings and your reports, kept honest.

Our river readings tell you how high the water is. But when the Environment Agency itself has a Flood Alert or Flood Warning in force for your area, that's the most authoritative word there is — so now we lead with it.

When there's an active EA warning near a path, the path page shows it up top:

• The level — Flood Alert, Flood Warning or Severe Flood Warning — in the Agency's own plain wording, with when it was issued
• A direct link to the official EA page for that area, so you can check it at source — we're not asking you to take our word for it
• It sits above our own estimate, because an official warning outranks a gauge reading every time

And it works both ways: if you report a path as flooded while an EA warning is in force, we note that your report lines up with the official picture — an extra bit of confidence behind what people on the ground are telling each other.

Warnings come straight from the Environment Agency under the Open Government Licence and refresh through the day.

The whole point is a fast check before you head out, so we've moved the two most useful facts up onto the cards — on your saved paths and the full list — where you see them without tapping through:

• Which way it's heading — a quiet ↑ Rising or ↓ Falling cue right on the card, so a flooded-but-falling path reads differently from one that's still climbing
• How fresh it is — a plain "Updated 2 h ago" so you know how much to trust the status at a glance
• We only show the arrow when the river is actually moving, and we built it so the whole grid still loads in a single trip to the database — no extra waiting

Same readings we already pull every two hours; nothing new to set up.

The level right now is only half the story. A path sitting at its flood line and falling is a very different shout from one rising fast — wellies-and-go versus leave-it-today.

Every path page now shows the recent direction of its river, in plain terms:

• Rising, falling or holding steady, with the actual change — e.g. "up 4 cm over the last 6 h" — so you can judge it yourself
• Worked out from the readings we've already been logging every two hours, so there's nothing new to wait for
• We only show a direction when there's enough recent data to mean something — no reading tea-leaves

Until now, the line between "passable" and "flooded" was a starting guess based on recent river levels. From today, your reports move that line.

• When you report a path, we line up the river level at that moment against what you saw on the ground — flooded, passable, or passable-in-wellies
• Enough of those and the path's flood line shifts to where it really floods, instead of where we first guessed
• The "wellies" reports are doing real work: they mark the in-between level where a path turns boggy but is still rideable, not just the point it goes fully under
• Each path page now says plainly whether its flood line is still a starting estimate or has been tuned by people who've been down

No accounts, no tracking — just the river readings and your reports, kept honest.

A river gauge a mile away only tells you so much. Now every path has a quick Been down today? report, right on its page — no account, no fuss.

• Tap Passable or Flooded, add how you crossed (on foot, by bike, in wellies) and an optional note
• Your report shows up straight away for the next person checking the path
• We record the river level at the moment you report, so over time the reports teach us the level where this path really floods
• Each page now shows the live river reading it's based on, in plain terms

See how readings and reports fit together

Path statuses now reflect live river-level readings from the Environment Agency rather than a static seed value. We match each path to its nearest gauging station and refresh every couple of hours.

• Statuses update automatically — no manual editing needed
• Each path page regenerates on a schedule so it stays current
• See how statuses work

The new path checker lets you search by name, place or waterway and see at a glance whether a route is passable, partly flooded or flooded — no account, no fuss.

• Search across every path on the map
• Clear status badges on every path page
• Browse all paths

Flood Path is live. It's a simple, free way to check whether a local cycle path, towpath or footpath is under water before you head out — starting with the flood-prone routes around Oxford.

Got a path that floods? Tell us about it and we'll add it to the map.