One Number, Fifty Climates: The Case Against Generic Soiling Assumptions in Solar Analysis

Open almost any solar analysis platform used in the United States today and navigate to the loss configuration screen. Somewhere in that interface, typically listed alongside wiring losses and inverter efficiency, you will find a soiling loss input. In most tools, that field is pre-populated with a value between 2 and 3 percent. In many workflows, it stays there.

That number did not emerge from a rigorous national measurement campaign. It did not originate from satellite monitoring data or from a statistically representative sample of US installation environments. It is, at its core, a consensus placeholder—an industry-wide approximation that has persisted through years of platform development because it was good enough to avoid obvious errors and specific enough to feel credible.

For a rooftop installation in suburban Minneapolis, it may be a reasonable approximation. For a ground-mount system in California's Central Valley during almond bloom season, or a coastal commercial array in the Carolinas exposed to salt-laden marine air, it is not an approximation. It is a systematic error embedded at the foundation of the financial model.

What Soiling Actually Looks Like Across US Markets

Soiling is the accumulation of particulate matter—dust, pollen, agricultural residue, industrial emissions, biological material, sea salt aerosol—on module surfaces. It reduces transmittance through the cover glass, attenuating irradiance before it reaches the photovoltaic cell. The energy loss is a direct function of accumulation rate, particle type, particle size distribution, and cleaning frequency.

None of those variables are geographically uniform, and none of them are captured by a single national default.

In California's Central Valley, agricultural operations generate dust and organic particulate at densities that are orders of magnitude higher than national averages. Systems in Fresno, Bakersfield, and the surrounding agricultural belt routinely experience soiling losses of 10 to 15 percent during peak agricultural activity periods—losses that do not distribute evenly across the year but concentrate in specific seasonal windows tied to planting, harvesting, and irrigation cycles.

In coastal markets from the Outer Banks of North Carolina through the Gulf Coast and into the Pacific Northwest, salt spray aerosol creates a qualitatively different soiling mechanism. Salt deposits are hygroscopic—they absorb atmospheric moisture and form a conductive film that not only reduces transmittance but can accelerate corrosion of frame components and junction box seals. The optical loss from salt soiling is typically lower than agricultural dust on a per-event basis but is more persistent between cleaning cycles because salt deposits do not respond to natural precipitation the way loose dust does.

In the desert Southwest—Arizona, Nevada, New Mexico—soiling losses are driven by fine mineral dust with high silica content. These particles are highly abrasive, create strong electrostatic adhesion to module surfaces, and are partially resuspended by wind without being fully removed. Systems in this region commonly show soiling losses in the 5 to 8 percent range on an annualized basis, with significant year-to-year variance tied to regional drought cycles and dust storm frequency.

In the industrial Midwest and Mid-Atlantic, combustion-related particulate and secondary aerosol from regional air quality events add a soiling component that is temporally correlated with atmospheric inversion events rather than seasonal agricultural patterns.

The Financial Modeling Consequences Are Not Marginal

The difference between a 2.5 percent soiling assumption and a 10 percent soiling assumption on a 500 kW commercial system is not a rounding error. At a blended electricity rate of $0.12 per kilowatt-hour, that variance translates to tens of thousands of dollars in annual revenue difference—and compounds over a 25-year project life into a material discrepancy between the financial model that was used to secure financing and the actual cash flows the system produces.

For developers and engineering firms operating in markets where soiling losses are systematically underestimated, this creates a recurring pattern: systems that underperform relative to the production guarantee, performance ratio disputes with asset owners, and O&M contracts that did not price in the cleaning frequency required to keep actual performance aligned with modeled performance.

The problem is not that soiling is unpredictable. It is that the tools engineers rely on to model it are not asking for—or incorporating—the information that would make the prediction accurate.

The Satellite Monitoring Alternative Is Available Now

The technological infrastructure to support location-specific soiling inputs in solar analysis tools already exists. Several commercial providers aggregate satellite-derived aerosol optical depth data, surface reflectance measurements, and ground-truth calibration from reference systems to produce spatially resolved soiling rate estimates at resolutions relevant to project-level analysis.

These datasets can be queried by geographic coordinate and time period, providing engineers with soiling rate estimates that reflect the specific site's historical accumulation patterns rather than a national average. Some platforms have begun integrating these data sources directly into their loss configuration workflows; others support manual input of location-specific soiling rates when engineers supply them independently.

The gap is not technological. It is procedural. Most engineering workflows do not include a step for soiling rate verification, because most platforms do not prompt for it. The default persists not because engineers are unaware of regional variance but because the software does not create friction around a number that is demonstrably wrong for a large fraction of US installations.

Raising the Standard for Soiling Inputs

The appropriate professional standard for soiling loss inputs in US solar financial models should include, at minimum, a site-specific soiling rate derived from regional measurement data or satellite-based aerosol monitoring, seasonal disaggregation for markets where accumulation rates vary significantly by time of year, and sensitivity analysis that shows the financial impact of soiling rates at the 25th, 50th, and 75th percentile of the site's historical range.

Platforms that continue to offer a single pre-populated national default without prompting users to verify its applicability to the specific project location are providing a tool that is, in a meaningful sense, miscalibrated for a majority of US markets.

Engineers who accept that default without verification are not making a minor methodological compromise. They are allowing a number with no site-specific basis to anchor the financial case for a project that will be evaluated against real measured performance for the next 25 years.

The soiling black box is not technically difficult to open. It requires only the will to treat location-specific data as a professional standard rather than an optional refinement.