A single test result cannot tell you your lake is getting worse. Three years can. This is the whole argument.

Everything else on this blog is a detail. This is the thesis.

Lake data has almost no value as a snapshot and enormous value as a series. Understanding why changes how you spend, what you measure, and whether anyone believes you when it matters.

Why one reading tells you so little

Suppose your lake tests at 32 parts per billion total phosphorus in July. Is that bad?

You cannot answer. It depends on what your lake usually runs, how deep the sample was taken, whether it had rained that week, how the summer compared to normal, and what part of the lake it came from. Published thresholds give you a rough category — this is a moderately productive lake — but they cannot tell you the thing you actually want to know, which is whether it is changing.

Lakes vary enormously between years for reasons that have nothing to do with management. A cool windy summer suppresses blooms in a lake with plenty of nutrients. A hot still one produces a bloom in a lake that is basically fine. Measure once and you have measured the weather as much as the lake.

What each additional year buys you

One year is a data point. It tells you roughly where you sit. It cannot support any claim about direction.

Two years is a comparison, and a dangerous one. Two points always make a line, and that line is usually weather. This is where communities most often reach a confident wrong conclusion.

Three years is where a trend becomes arguable. Not proven — arguable. Enough to justify a closer look.

Five years is where a trend becomes defensible to a sceptic. This is roughly the threshold at which agencies, grant reviewers, and consultants start treating a dataset as real evidence.

Ten years and beyond is where a lake community stops asking questions and starts answering them. You can see the effect of an intervention. You can separate weather from change. You can say what happened, when, and with what.

The asymmetry

Here is what makes this urgent rather than merely interesting.

The cost of collecting data is roughly constant per year. The value of that dataset grows faster than linearly, because every new year is comparable to every previous year. A ten-year record is not ten times as useful as one year; it is far more.

And you cannot buy the past. Whatever you did not measure in 2019 is gone. No budget, no instrument, no consultant recovers it. The only decision available is whether 2027 will exist in your record.

This is why the most expensive year of monitoring is always the one you skipped.

Consistency beats sophistication

Given a choice between a sophisticated measurement taken occasionally and a simple one taken reliably, take the simple one.

Same place. Same time of year. Same method. Written down with the date and the conditions. A modest dataset with no gaps is worth more than a precise one with holes, because gaps break exactly the comparison that gives the data its value.

This is also why changing your method is costly. A lake that switched labs, or moved its sampling location, or changed from surface to depth sampling, often cannot compare across that boundary. If you must change something, run both for a season.

What this is really about

Lake decisions get made in rooms where someone has to be believed. A budget meeting. A county hearing. A grant application. A conversation with a neighbour who thinks the association is overreacting.

Without a record, those rooms run on conviction, and conviction loses to competing priorities. With one, they run on evidence, and the argument becomes about what to do rather than whether anything is happening.

That is the entire reason to measure. Not the numbers. The ability to be believed.