The array ran at 110% of normal for the sunlight.
Each 15-minute interval's output is divided by its sunlight and the array's kilowatts, and the day is compared with the median of the 30 days before it. Expected output is the sunlight times the array's kilowatts times that median, so a cloudy day expects little and a clear day a lot.
The sensor lies flat and the panels are tilted, so morning and evening intervals read a little differently from midday on every day alike. That is why the comparison is with this array's own recent days rather than a theoretical figure.