Contradictory NLDN monthly metadata

I am trying to determine what is exactly contained within the monthly NLDN gridded data. The website where you can access the data (CMAS: Community Modeling and Analysis System) via a Google Drive folder says it is flash data with units of km^-2 day^-1. However, the metadata stored in the actual .ioapi files say it is stroke data with units of m^-2 day^-1. Which is correct?

I checked my archived data for these monthly NLDN datasets, in the files that I checked, the metadata shows that the unit is km^-2 day^-1 consistent with the Google Drive description. Some particular files may have been mistakenly changed to m^-2 day^-1. As my experience using these datasets many years ago, lightning flash data should always be in the km^-2 (either by per day or per hour) conforming to the CMAQ code . By the way, which specific data files that you checked have the unit of m^-2 day^-1? In addition, we stopped supporting these monthly NLDN datasets.

All of the files in that Google Drive folder have units of m^-2 day^-1, and a description of ‘NLDN strikes from NLDN data source.’ Filename structure of NLDN.YYYY.MM.ioapi.

Yes, I checked the files in Google Drive, and it was marked as 2015version. I downloaded the 2008 files and compared with what I have in my archive. The numbers are comparable, but apparently the unit was wrong. I don’t know who created these files, but I’m pretty sure that the script that was used to create these files mistakenly left out the letter “k” before “m”. By any means, the unit should be km^-2 day^-1 in these files.

Ok, but are the data flashes or strokes? The metadata in the files clearly says it is strokes (“strike”), but the website says flashes.

The terms are confusing in many ways. Generally the NLDN dataset provides the CG flashes which defined as strokes/pulses within 10/20 km from the first stroke and 500 ms of a previously added stroke into a flash. Some flashes may contain just one strike. In the earlier days, it was called strikes or strokes, but later on, for consistence with the NLDN dataset, we changed to flashes.