A soiled or out-of-calibration pyranometer under-reads irradiance, and because performance ratio (PR) divides measured energy by the energy the irradiance says you should have produced, a low irradiance reading silently inflates PR. Every downstream metric, from PR to availability claims to investor reporting, inherits the error. That is why solar sensor data quality belongs on the asset manager’s risk register, not only on the O&M checklist.
Short answer: data quality is not a maintenance issue; it is a financial risk layer. Platforms that do not validate sensor accuracy continuously are reporting on data nobody has checked.
ClearSpot.ai solves this by acting as a solar data intelligence platform, where every data point is validated, cross-checked and auditable.
The Real Problem – Solar Platforms Were Built for Connectivity, Not Accuracy
Most platforms focus on collecting data—not verifying it.
That’s why modern asset managers are shifting to an independent performance data layer with sensor validation that separates “data availability” from “data reliability.”
What IEC 61724-1 Expects From Your Monitoring Sensors
The reference document is IEC 61724-1:2021, Photovoltaic system performance – Part 1: Monitoring. It defines three monitoring accuracy classes: Class A (high accuracy, the normal expectation for utility-scale plants), Class B (medium accuracy) and Class C (basic). For each class it specifies which irradiance, temperature and meteorological sensors are required, how many are needed as plant capacity grows, how often they are sampled and recorded, and how they must be cleaned, maintained and recalibrated.
Pyranometers themselves are graded under ISO 9060:2018 (Class A, B and C, with a “spectrally flat” designation for PV use). A Class A monitoring system with lower-grade sensors, missed recalibrations or no cleaning log is not delivering Class A data, whatever the specification sheet says. Lenders’ engineers and buyers in technical due diligence increasingly ask to see calibration certificates and cleaning records for exactly this reason.
Q1 – When Was Your Pyranometer Last Calibrated?
A pyranometer or reference cell that has drifted, or that has not been cleaned, introduces a systematic bias into every irradiance-normalised KPI. Unlike random noise, that bias does not average out over a month or a year.
ClearSpot solves this using AI agents for data validation and anomaly detection that:
- Cross-check ground sensors against satellite-derived irradiance
- Detect gradual sensor drift against neighbouring sensors and modelled clear-sky values
- Flag calibration certificates that are overdue
- Separate soiling-related under-reading from genuine calibration drift
Q2 – Is One Weather Station Per 100 MW Creating Blind Spots?
A single met station represents conditions at one point. On a large plant, passing cloud, terrain and local soiling can make that point unrepresentative of the rest of the array, which is why IEC 61724-1 scales the number of irradiance sensors with plant size.
ClearSpot supplements on-site stations with satellite fallback and distributed validation, so a single unrepresentative sensor does not define the whole plant’s expected yield.
Q3 – Is Your Soiling Sensor Lying?
Soiling sensors compare a cleaned and an uncleaned reference device. If the “clean” reference is not cleaned on schedule, or the two devices drift apart, the measured soiling ratio is wrong, and cleaning is either triggered too late (lost energy) or too often (wasted cost).
ClearSpot links SCADA data, satellite inputs and sensor outputs, triggering corrective actions through predictive O&M built on trusted data.
Q4 – Are You Cross-Checking Ground Sensors Against Satellite Data?
Best practice combines ground sensors with satellite validation. Ground sensors are precise at their own location but vulnerable to soiling, shading and drift; satellite irradiance is spatially consistent but coarser in time and space. Each checks the other.
ClearSpot cross-validates both automatically, so a divergence between the two is flagged as a probable sensor issue before it reaches a monthly report.
Q5 – How Do You Handle SCADA Data Dropouts?
Improper gap handling corrupts analytics and reporting. A communications dropout, a plant outage and a failed sensor all look like missing data, but they must be treated differently: a plant outage is lost energy, while a logger dropout is not.
ClearSpot classifies every data gap, applies the correct imputation method and tags estimated data—ensuring transparency in investor-ready reports from auditable data sources.
Q6 – Is Your Module Temperature Sensor Accurate?
PR calculations are usually temperature-corrected using the module’s power temperature coefficient, so any error in the measured module temperature flows straight into the corrected PR. Common causes are back-of-module sensors that have detached, sensors placed on unrepresentative modules at the edge of a row, and cabling faults.
ClearSpot validates temperature readings against thermal models driven by irradiance, ambient temperature and wind speed, and flags sensors whose readings no longer track the model.
Q7 – What Happens When POA Sensors Fail?
When plane-of-array (POA) sensors fail, ClearSpot converts global horizontal irradiance (GHI) to POA using a transposition model, applies an uncertainty flag to the derived values and keeps reporting continuous, so the period is visible as estimated rather than silently mixed with measured data.
Q8 – Are Your Timestamps Misaligned?
Timestamp drift breaks root-cause analysis. If inverter, met-station and SCADA clocks disagree, or if some devices label intervals by their start and others by their end, alarms and production dips appear to happen at different times.
ClearSpot normalises all data streams to UTC, monitors NTP drift and ensures accurate event correlation.
Q9 – Can You Upgrade SCADA Communications Without Losing Data?
Communication upgrades are a common point of data loss: tag names change, scaling factors are re-entered by hand and historical series are split in two.
ClearSpot keeps continuity through validation layers, transition tagging and automated quality checks, working as an AI layer over your existing SCADA rather than replacing it.
Q10 – What Happens When the Weather Station Goes Down?
Without redundancy, analytics stop when the weather station stops. ClearSpot maintains continuous operation by switching to satellite and neighbouring-sensor fallback, with every substituted value tagged.
A Solar Sensor Data Quality Checklist
- Confirm the monitoring class (A, B or C) your contracts and reports assume, and check that installed sensors match it.
- Keep calibration certificates for every pyranometer and reference cell, with the next due date tracked.
- Log every sensor cleaning, and compare readings before and after cleaning to estimate soiling bias.
- Cross-check on-site irradiance against satellite data and against neighbouring sensors every day.
- Inspect back-of-module temperature sensors for detachment during routine site visits.
- Classify every data gap (communications, sensor or plant) and tag every estimated value in reports.
- Synchronise all loggers to UTC via NTP and document interval-labelling conventions.
Conclusion
A small data error does not stay a data error; it becomes a financial error in PR guarantees, availability claims and asset valuations.
ClearSpot.ai ensures:
- Every sensor is validated
- Every data gap is handled correctly
- Every report is defensible
Frequently Asked Questions
It is the accuracy, completeness and traceability of the measurements a PV plant uses to judge its own performance: irradiance, module and ambient temperature, wind and soiling sensors, plus the timestamps and gap handling around them.
A soiled pyranometer under-reads irradiance, so the expected energy is understated and the performance ratio appears higher than it really is. The plant looks healthy while real losses go unnoticed.
IEC 61724-1:2021 defines Class A, B and C monitoring systems and sets requirements for sensor type, number, maintenance and recalibration. Pyranometer accuracy classes are defined separately in ISO 9060:2018.
Missing irradiance should be replaced with satellite or neighbouring-sensor data, transposed to plane of array where needed, and every substituted value should be tagged as estimated so reports stay transparent.