Why does the PV Output graph show odd data for certain months?

Occasionally, customers run across anomalous PV production patterns in an analysis, where one or two months show significantly lower production than expected. For example, the output for March might be lower than April. In most of the cases of this that we've seen, this is due to a peculiarity of the weather data set used in the production estimate. 

SolarNexus uses PVWatts to estimate production, which uses historical weather data to estimate the amount of available sunlight at the location. By default, we specify to use weather data from the nearest TMY3 weather station to the project site. However, certain suspect weather stations appear to result in peculiar production patterns.

To independently confirm whether an anomaly you are seeing is a result of suspect weather data, you can go to pvwatts.nrel.gov, enter the customer address, and then select the nearest weather station to the customer site. On the results page, if you see the same anomalous production pattern that you see on the SolarNexus analysis screen, then try going back and selecting a different nearby weather station for the PVWatts analysis. If the production pattern looks more normal, it means the data for the nearest weather station is either bad or the actual historical weather patterns for that weather station are unusual and perhaps not representative of future weather patterns for your project site.

Within SolarNexus, you can override the default weather station by going to the Systems page, clicking the PV Output Modeling button toward the bottom, and selecting a different weather station. To compare production estimates using different weather stations, you can create a new solution cloned from an existing solution, specify to use a different weather station on the PV Output Modeling screen, and re-analyze the solution.