Does cleaning solar panels actually pay?

The dry-climate answer: rain is the cleaner

The conventional model of soiling is a sawtooth. Dust accumulates, output falls, it rains, output recovers, and the cycle repeats. In dry climates that model appears to hold, and the laboratory work supports both halves of it.

On accumulation, the finding is specific: examining a number of sites in southern California, researchers found that during long soiling periods humidity is typically 55% or less, dew cycles are not typical, and "periods of three or more months without rain can easily result in accumulated soiling losses of 10-20%".

On recovery, chamber testing soiled a module to 20% loss with a standardised road dust, then sprayed it with water "until dust or particles were no longer visible", three times at moderate humidity and three more at higher humidity. The cycles reset each time. The report treats the water spray as rain recovery and re-soils from the recovered baseline, which is only a defensible method if recovery is close to complete.

So in a dry climate the honest framing is not "should I clean" but "how long until it rains". If the answer is next week, paying someone to climb on your roof buys you a week of output you were going to get back anyway. If the answer is three more months of a dry season, the accumulated loss over that period is real and the arithmetic changes.

Field measurement gives a sense of pace: at a utility plant in central California, two strong soiling periods produced measured divergence at rates of about 0.086% and 0.066% per day. Those are plant figures rather than rooftop ones, and they suggest soiling as a slow accumulation rather than a sudden problem.

The humid-climate answer, and it overturns the model

A 2026 laboratory collaboration went looking in exactly the region the earlier work excluded, and what it found does not fit the sawtooth at all.

The starting hypothesis was pollen, and it was wrong. Verbatim: "The original hypothesis was that locations with a heavy spring pollen season would be correlated with the worst photovoltaic bio-soiling losses… While soiling losses did increase during pollen season the losses were not sustained through rain events and into the summer.… Careful analysis of these images showed that pollen was present in the spring months but not at other times of year. Furthermore, regardless of the time of the year, fungus of varying degrees of coverage was found in the majority of images."

What they found instead were sustained losses. The partner organisation "provided thousands of microscopic images of the surface of solar panels that had sustained losses in the 5% to 15% levels".

And the organism is not what most people picture as dirt. The report identifies "heat- and radiation-adapted epiphytic fungi that form complex sub-aerial biofilms" as the primary bio-soiling problem, describing them as "black microcolonial fungi originally adapted to thrive on rock surfaces in various harsh climates" that now use the same adaptations on solar glass. Electron microscopy confirmed that black lumps on the surface "are actually primarily spherical fungal material with a few shards of mineral material intermixed".

The conclusion for the models is stated directly: "current PV sawtooth soiling models have been proven to be incorrect for the Southeast United States. Whereas previous models predict negligible soiling losses in the Southeast, the authors have demonstrated a number of sites in the region with sustained soiling losses greater than 10%."

And on remedy: the hypothesis upheld at the conclusion was that "bio-soiling losses from fungus and bacteria build up over months and years to create sustained soiling losses. Fungus and bacteria build up requires mechanical cleaning to remove."

What this does to the number in your proposal

There is a consequence of the southeastern finding that reaches beyond cleaning, and it is worth following because it affects people who never intend to clean anything.

The modelling tool behind a great many residential estimates applies a default soiling loss of 2%. A single figure, applied nationally, for every climate.

Set that beside the two findings on this page. In a dry climate with a long dry season, accumulated losses reaching 10 to 20% before rain clears them are not well described by a flat 2% annual figure, though averaged across a year with rain events the number may not be far out. In the southeast, sites were found with sustained losses of 5 to 15% that rain does not clear, and the report states the models predicting negligible losses in that region have been proven incorrect.

So a proposal for a house in the southeast, built on a 2% soiling default, may be assuming away several times the loss that is actually there. Not because anyone was dishonest, but because the default encodes a model the research now says does not describe that region.

This is one of the clearer illustrations of a point we make elsewhere: the assumptions inside an estimate are where most of the disappointment lives, and they are rarely on the page you are shown. If you are in a humid region, ask what soiling loss the estimate used, and whether it was the default.

What to actually do about it

Work out which problem you have before spending anything. The dry-climate question is about waiting for rain; the humid-climate question is about whether a biofilm has established. They are different problems with different answers and the same word attached.

Look at your production data across a full year rather than at your panels. Soiling in a dry climate shows as a gradual decline through the dry season with a step recovery after significant rain. Bio-soiling shows as a decline that does not recover, which is exactly what you would otherwise mistake for degradation. That distinction is visible in monitoring data and invisible from the ground.

If you do decide to clean, the risk is the roof, not the glass. Falls are the serious hazard in this activity and they are entirely avoidable by not going up. Ground-level methods with a soft brush and a hose reach many residential arrays. Where they do not, that is a job for someone equipped for it.

Timing matters more than technique. Cleaning glass that is hot in direct sun risks thermal stress and leaves deposits as water evaporates before it can be removed. Early morning or an overcast day is the sensible window, which is also when a roof is most likely to be wet and slippery. Those two considerations pull against each other, which is a further argument for doing it from the ground or not at all.

Do not use anything abrasive or high-pressure. The front surface is glass with an optical treatment on many modules, and damaging it is not recoverable. Manufacturer instructions govern here and several warranties exclude damage from improper maintenance, which makes an aggressive cleaning attempt a way to create a much larger problem than soiling.

And check your own warranty before hiring anyone. Where a warranty imposes maintenance obligations, doing nothing may matter; where it excludes damage from maintenance, doing the wrong thing certainly does.

One thing that genuinely does change the calculation is roof access. A single-storey array reachable from the ground with a hose and a soft brush is a different proposition from a second-storey pitch requiring equipment and someone insured to use it. The loss may be identical and the economics are not, and that is a fact about your house rather than about solar. If your array is not reachable safely from the ground, the honest comparison is between the value of the recovered output and a contractor's visit, not between the output and an afternoon of your time.