Saharan dust can influence photovoltaic generation in two different ways. Airborne particles can change the radiation reaching the plant, while deposited particles can leave modules soiled after the atmospheric event has passed. These effects have different time scales and require different evidence. A single percentage loss should not be assumed for every event, location or PV installation.
Distinguish the atmosphere from the module surface
An atmospheric dust event can alter the solar resource before light reaches the panels. Its effect depends on the dust concentration, cloud conditions and the event's location and timing. Surface deposits reduce the light reaching cells until weather or cleaning changes the situation. A forecast that follows atmospheric conditions alone may not capture persistent module soiling. Conversely, an output reduction during a dusty sky does not by itself prove that the modules need cleaning.
Use plant measurements with weather context
Compare measured generation with an appropriate reference, accounting for irradiance, temperature, plant availability and clipping. A lower output than expected can have several causes, including cloud error, an outage or curtailment. Check operational records before attributing the deviation to dust. Where available, suitable irradiance and soiling measurements can help separate atmospheric and surface effects. Document the sensor location and measurement quality.
Watch the recovery period
A short atmospheric event may leave a longer operational signature. Rain, cleaning and wind can change surface deposits, but their effect depends on the installation and event. Record maintenance and cleaning timestamps so that historical data is interpreted correctly. A correction trained during a soiling period should not continue unchanged after the modules have recovered.
Evaluate forecasts around complete events
Inspect the event's arrival, peak and recovery rather than selecting only its worst interval. Compare the issued forecasts, weather inputs, measured production and relevant operating flags. Keep the issue time fixed to avoid using information that became available later. Review both the expected generation level and the timing of changes. The practical question is whether the forecast gives a useful schedule while the event evolves.
What the historical article established
The original 2018 report described enercast's investigation of dust effects on PV generation and forecasts. Its loss estimates were contextual claims from that report, not a current performance guarantee. For a new evaluation, agree which atmospheric inputs and plant observations are available and test the proposed handling on the actual portfolio. A general dust alert is useful context, but it does not determine the output loss at a specific plant.
Do not infer cleaning value from one deviation
The economic case for cleaning depends on the installation, access arrangements, expected recovery and the cost of the work. Forecast deviations alone are insufficient to decide it. Separate the question of expected production from maintenance action. A forecasting review can identify an unusual pattern and support discussion, but the asset operator should use appropriate inspection and operating information when deciding what to do.
Build an event record
For a pilot, retain the weather inputs used, issued power forecasts, generation observations and timestamps for rain or cleaning where known. Add operational flags and record which observations were missing. Compare pre-event behaviour, the event itself and recovery using a consistent capacity reference. Sites without useful telemetry should be distinguished from those with complete records. A portfolio-wide dust episode may still produce different effects at individual assets, so inspect site results before applying a uniform correction. This evidence makes a later discussion about model changes more useful than a general claim that dust always causes a particular percentage loss.