Weather-normalized production comparison

This method compares measured plant output with the output you would expect for the measured weather and irradiance, using a performance model the stakeholders agree on. Line up SCADA production with irradiance, rainfall, cleaning logs and, in winter, snowfall, snow depth and module temperature, and check how the normalized output changes around these events. If normalized output falls slowly during dry periods and comes back after rain or washing, the cause is likely granular dust soiling. If the shortfall stays after rain but output steps up after manual cleaning, the cause is more likely widespread fungal soiling, which you can confirm by close inspection or handheld microscopy. Residuals that line up with snowfall and recover slowly during melt point to snow accumulation losses, and you can compare them with the snow-loss model. Before you blame soiling or snow, remove periods affected by outages, clipping, curtailment, inverter behaviour in cold weather, sensor icing or irradiance sensor problems. Never rely on one dirty irradiance sensor, because sensor drift or local obstruction can produce what looks like a soiling signature. Small losses such as minor shading may not show clearly in yield or performance-ratio comparisons, so you may need other signals as well.

<!-- review notes

• sentence 6: support check: partial - Passages support filtering clipping, curtailment, availability, sensor issues and cold-weather inverter behavior, but outages for snow are only implied via availability and 'never/before blaming' is a stronger framing. -->