QA/QC of piranometers: identification of shading, dirt and calibration drift

A pyranometer that seems to work may be producing systematically wrong data. Accumulated dirt in the dome, shadow of a sun-rise support tube or a sensitivity that has degraded 4% over three years without recalibration — none of these problems triggers an equipment alarm. The data arrive, seem reasonable and form the basis that will estimate the long-term irradiance and the P50 of the project.

This note presents the three most common systematic errors in solar measurement campaigns, such as identifying them by their temporal patterns and documenting the treatment for compliance with IEC 61724-1.

The three modes of silent failure

– Structural shade, dirt and calibration drift share a characteristic: they reduce the irradiance measured without interrupting the measurement. Each one has a distinct mechanism and therefore a different temporal signature — and it is this signature that allows them to be identified in QA/QC.

Structural shading It occurs when a physical element of the monitoring system—fixing rod, ventilation tube, support arm, ill-positioned cable—projects shadow over the dome of the pyranometer at certain solar angles. As the geometry is fixed, the shading occurs in the same hours and directions throughout the year, with magnitude varying with the solar decline. The resulting signal is acute and recurrent depressions in specific positions of the solar diagram.

Dirtyness It is the gradual deposition of particles over the glass dome — dust, sand, soot, pollen, birds' waste. The dirty dome attenuates the radiation in a diffuse way, with no preference for an incidence angle, but with a characteristic daytime asymmetry: the depression is more intense in the early hours of the day, when the night dew cemented the deposited particles, and is slightly attenuated during the morning by drying and convective dispersion. In arid or coastal regions with high salinity, the accumulation rate can be significantly faster than in wet areas with frequent rain.

Calibration derivative It is the slow and monotonous change of the sensor calibration constant — the relationship between the tension generated by the thermopile and the incident irradiance. The degradation occurs by aging the thermopile materials, prolonged UV exposure and cyclic thermal stress. Unlike dirt and shading, the drift has no temporal signature within the day — it produces a uniform offset that is only detectable by comparison with calibrated reference or by analysis of long-term trend.

Identification by Kt analysis and clearsky ratio

The clearness index Kt = GHI / GHI_Alien and the ratio clearsky Kcs = GHI / GHI_{clearsky_modelado} In days of clean skies independently identified, systematic deviations of Kcs reveal the type of problem:

• Kcs < 1 uniform throughout the day → dirt or calibration drift (indifferent only by day profile — requires temporal analysis)
• Kcs < 1 concentrated in periods of the day with recurrent pattern → Structural shading
• Kcs > 1 (values above 1.0) → possible cloud reflection (cloud edge effect), alignment error or wrong parameter in the clearsky model
• Kcs in gradual decline over weeks → progressive accumulation of dirt
• Kcs in monotonic decline over months to years → Calibration drift

The distinction between dirt and drift is critically important because the treatments are different: dirt requires cleaning and, if the dirty period is short, it can be possible to reconstruct the data with clearsky modeling. Calibration derivative requires recalibration and can retroactively invalidate the entire affected series, depending on the magnitude.

Calibration derivative: typical rates and normative limits

ISO 9060:2018 classifies piranometers in three classes — A (secondary standard), B (first class) and C (second class) — based on spectral uncertainty, temperature response and linearity criteria. IEC 61724-1:2021 requires Class A or B instrumentation for bankable resource campaigns; Class C is not accepted in technical reports for funding.

The typical drift rate observed in the field varies with exposure class and regimen:

• Class A well maintained: <0.5% per year, detectable only on annual recalibration
• Class B without recalibration: 1.0–2.5% per year, accumulating significant drift in campaigns of 2–3 years
• Any class in aggressive environment (extreme temperature, coastal salinity, high altitude): up to 3–4% per year

The IEC 61724-1 limits the permissible accumulated drift to ±2% In relation to the last valid calibration. Piranometers outside this limit should be removed from operation, recalibrated and — depending on the magnitude and period affected — their data retroactively signaled as suspects in the report.

Separation of populations by clearsky dispersion analysis

In addition to time series analysis, the dispersion between measured GHI and modelled GHI under light sky conditions is a powerful diagnostic tool. When plotting pairs (GHI_clearsky, GHI_measured) only for periods identified as clear sky, three distinct populations appear:

Maintenance frequency: what IEC prescribes

The IEC 61724-1:2021 does not define a fixed cleaning interval — it recognizes that the required frequency depends on the site environment. What the normalized PRescribes is that the interval is determined from data, not from assumptions:

1. During the first 2–3 months of the campaign, clean up Daily - and record the Kcs ratio before and after each cleaning;
2. Based on the observed accumulation rate, define the cleaning interval that keeps the salting loss below 1% - between consecutive cleanings;
3. Document all cleanings with date, time and condition of the dome before cleaning;
4. Required recalibration if the accumulated drift exceeds ±2%, or when the sensitivity varies more than 0.5% over two consecutive verification cycles.

In high deposition environments — semi-arid regions of the Northeast, coastal areas with high salinity, areas close to dirt roads — weekly cleaning is often insufficient to keep the salting loss below 1%. Campaigns that adopt biannual or monthly cleaning in these environments accumulate systematic errors that cannot be corrected retroactively with precision.

Documentation for bankability

A technical solar resource report for financing shall document, for each campaign piranometer:

• Class ISO 9060:2018 instrument and date of last traceable calibration to BIPM;
• Registration of all cleanings performed with date, time, operator and condition of the dome;
• – Time series of daily Kcs with identification of clean days used for drift monitoring;
• – Kcs trend analysis throughout the campaign with Mann-Kendall test or equivalent;
• Soiling ratio estimated by period (pre/after cleaning) with associated uncertainty;
• For replaced or recalibrated piranometers during the campaign: analysis of homogeneity of the series before and after the replacement.

Campaigns that did not systematically record clean-up — or that only recorded the date without the dome condition — face difficulties in demonstrating that the salting loss is within the tolerable limits. In rigorous due diligence processes, the absence of this record may lead to the application of an additional 1–3% uncertainty factor on solar P50, with direct impact on P90 and debt service coverage.