Hi Chris,
The Pxx values are based on total annual values. For example, to calculate the P90 value for annual energy, SAM uses the following method (described in detail in Dobos (2012) "P50/P90 Analysis for Solar Energy Systems Using the System Advisor Model" available here
sam.nrel.gov/weather-data/weather-data-publications.html):
1. Run an 8760 simulation for each year to calculate the total annual energy production for each year.
2. Sort the resulting annual energy values in increasing value, and assign each value a probability equal to 1 divided by the number of values. For example, the table below shows results for 18 years, so each value is assigned a probability of 1 / 18 = 0.056.
3. Determine the Pxx values. For these results, the P50 value for annual energy is 57,430,500 kWh, meaning that 50% of the annual energy values are greater than that number, and 50% are less. The P90 value is between 52,685,900 kWh and 53,637,500 kWh. SAM uses linear interpolation over all the values to estimate the P90 value, in this case 53,447,200 kWh, so 90% of the energy values are greater than that value.
That means there isn't really a set of hourly 8760 (or subhourly) results for a given Pxx value. In this case, the P50 value happens to correspond to the results for Year 2000 because there is an even number of years, but the P90 value does not correspond to results for a particular year. You could choose 1998 or 2011 to represent the P90 year, as long as you keep in mind that its the total annual energy value that is the P90 value, not the hourly generation values.
Best regards,
Paul.