Smith Lake

Calibrated against a NIST-traceable certified thermometer. All five probes match that reference to within 0.1°F and each other to within 0.03°F.

Every reported temperature is an average of about 60 readings taken over a 5-minute window. A single instantaneous reading catches whatever the water is doing at that exact moment — a passing boat wake, a pocket of sun-warmed water drifting by. Averaging smooths that out, and it lets the sensors resolve finer detail than any one reading can.

This isn’t off-the-shelf. I designed and built the whole system — the probe array, the electronics, the firmware, and this website. Five marine-grade digital probes on a custom array, reporting to a small computer in a waterproof enclosure that handles the sampling, averaging, local storage and upload. Nothing about it came out of a box.

Specifics, for anyone who wants them: five DS18B20 digital probes on a one-wire bus, read by an ESP32 with a real-time clock, hardwired Ethernet and battery backup.

A lake doesn’t have one temperature. Sunshine heats the top few feet faster than anything below, and because warm water is lighter it stays there — so on a bright afternoon the 2-foot reading can run several degrees above the 10-foot one.

Overnight, or as the seasons turn, the surface cools, gets heavier and sinks, pulling the whole column into circulation until the depths even out. Wind does the same thing.

So the gap between the top and bottom readings tells you what the lake is doing. Wide means the surface is soaking up heat faster than it can mix downward. Narrow means the column has mixed. Over a day you will usually see both.

No. This is an independent project — not a measurement or forecast from the Army Corps of Engineers, Alabama Power, or any lake-management agency. It’s a continuous temperature record from one dock.

It’s meant for observation and curiosity, not as a scientific instrument. Sensors fail, networks go down, equipment needs maintenance. The goal is simply a long-term record: the same water, measured the same way, at the same depths, every day.

Boat sensors sit right at the surface, often near a hot engine, and read the top inch or two. On a sunny afternoon that can be several degrees warmer than the water two feet down.

The color follows published cold-water thresholds, not a judgment about whether to swim:

• Orange — 77°F and up. Below 77°F, breathing starts to be affected by the water temperature.
• Gold — above 70°F up to 77°F. Above the National Weather Service’s cold-water line, but cool enough to affect breathing.
• Light blue — 60°F to 70°F. The National Weather Service classifies water at 70°F or below as cold water, where sudden immersion can cause cold shock.
• Deep blue — below 60°F. The 50–60°F range is where cold shock is most intense.

This site reports the temperature. It doesn’t say whether it’s safe to swim.

For more on cold water and cold shock, see the National Center for Cold Water Safety.

Hover anywhere to read all five depths at that moment, with each one’s low and high for the period shown.

On a computer, zoom in by dragging across the chart, or scrolling. On a phone, drag across the chart to read the temperatures at any moment instead. The bar underneath shows where you are in the year — drag it to move, or drag its edges to widen or narrow the window. To change the range, tap any of the buttons above the chart — Today, Yesterday, 7 Days, 30 Days, or Full year for January through December.

The checkboxes turn individual depths on and off. A greyed-out depth means there’s no data at that depth for the period you’re looking at.

Compare days overlays up to five dates on one time-of-day chart, so 2 PM on one day sits directly above 2 PM on another — useful for seeing how the daily heating and cooling pattern is changing. One depth at a time there, since five depths across five days is unreadable.

On wide views, the chart averages. A single day shows every 15-minute reading; a full year shows daily averages, so the chart stays readable instead of cluttered. That means a summer peak looks lower on Full year than on the day view — the average of a whole day is never as high as its hottest moment.

Yes. The temperature at the top is the whole answer for most people, and it reads fine on any phone.

The chart works too. Drag a finger across it to read all five depths at that moment — the numbers show up just above the chart, and they stay put when you lift your finger, so your finger isn’t covering the lines while you read them. Tap anywhere else to clear them.

To change what you’re looking at, use the buttons above the chart — Today, Yesterday, 7 Days, 30 Days or Full year — or drag the bar underneath. Zooming into a narrow slice of time is still easier with a mouse, so for really digging into the detail, a computer is the better tool.

Deliberately. All times are Central, and the day runs midnight to midnight — including the part that hasn’t happened yet.

The empty space is there for scale. You can see exactly where you are in the day, and watch the temperatures fill in as it goes on.

Every reading works the same way, around the clock. Each one is an average over a five-minute window, labeled with the middle of that window rather than the end.

For example, the 10:00 reading starts sampling at 9:57:30 and stops at 10:02:30, so it can’t be sent to the website until that window closes. It reaches the database within seconds, and an open page picks it up a minute or two later — so if you’re watching, expect the 10:00 temperature to appear around 10:03, sometimes 10:04.

Alabama Power starts lowering Smith Lake on July 1, from its summer pool near 510 ft toward a winter pool near 496 ft by December 1. It stays there through the winter, then refills toward summer pool in the spring.

I have year round water at my dock, so the dock never touches bottom — but the water does get too shallow for a 10-foot sensor array. When the lake gets near 499 ft I raise the array 2 ft, and near 497 ft I raise it another 2 ft to keep the array from hitting bottom. The site then shows 2 to 8 ft, and 2 to 6 ft respectively.

As soon as the water comes back up enough for the full range, I drop it back down and the 8 and 10 ft readings resume.

The system stays running, but the website will lag behind and you’ll see a banner notice that tells you how long it’s been since the last temperature reading was received.

When power fails, the entire system switches to battery instantly and keeps running normally for about 36 hours.

The internet usually goes down with the power, so temperature readings are stored on the device until the connection returns. It holds about 10 days of readings locally before it runs out of storage.

For a longer outage, a generator keeps it running and I upload temperature readings over a hotspot or Starlink every day or two, so I never hit the limits of the battery or storage. The goal is to never lose a reading — the worst that should happen is a delay in posting them here.

Because it mostly is. The record starts July 22, 2026, so there’s only summer on the chart — the temperature scale reaches down to the 60s to leave room for winter, and right now everything sits bunched near the top.

That fixes itself. As fall and winter fill in, the wide views become the interesting ones: how fast the lake cools, when it turns over, how cold it gets, and when it starts back up in spring.

That’s the intention. This is built to run for years — not a temporary summer project that’s going to disappear. The site should get more useful the longer it runs: one summer’s worth of data is neat, but watching the lake move through the seasons, and eventually comparing one year against the next, is where the real value is.

The data is protected against loss. Every reading is stored in three separate places — the online database, a nightly backup on a machine at my house, and an off-site cloud copy of that backup. The backup only ever adds to the record; it’s built to refuse any change that would make the history smaller, so a failure upstream can’t quietly erase what came before.

Maybe. It isn’t something I sell or install — but this site could grow into more than one dock’s readings, and if you’re set up for it, we might work something out.

What you’d need at your dock:

• At least 25 feet of water depth at your dock at summer pool (510 ft). More is better. That’s the minimum that lets a 2–10 ft array ride out a full 14-foot winter drawdown without repositioning — at 496 ft there’s still a foot of water under the deepest probe. Deeper is genuinely useful though: with 40 feet I could run probes down to 25 feet instead of 10.
• Reliable wired internet at the dock. Wi-Fi from the house isn’t enough — everything at my dock is hardwired, and yours would need to be too. A dedicated point-to-point wireless bridge to feed a hardwired network switch on the dock is acceptable, but you’d be setting up the bridge, a static IP and the switch yourself, so some networking knowledge is required.
• Reliable power at the dock. Simplest is a proper 120 V circuit from shore — not an extension cord. Solar-only docks can work too, with some extra planning and different equipment than I’m running. I can’t make recommendations there, but whoever installed your system could. Solar isn’t an automatic dealbreaker.
• A weatherproof enclosure and the gear to go in it, ideally installed under the dock’s roof to protect it from the elements.
• Willingness to install it and keep it running. You’d handle maintenance — cleaning the probes, replacing a sensor if one fails, and troubleshooting your own network. I’d be available for questions and support.

I’d build the array and help you set it up. The rest is yours.

I’m working on a new design that uses the same equipment but makes the build a lot faster and simpler. I’m probably a couple of months from being able to put one together, so realistically this is a winter project, not a next-week one.

If you’re interested and comfortable installing and maintaining it yourself, get in touch — contact@smithlaketemp.com. If your installation proves reliable, I would publish your dock’s data here alongside mine — and maybe this website grows into a repository for a dozen docks or more, publishing temperature data from several different parts of Smith Lake.