The highest annual frequency of thunderstorms in Alaska is over the Tanana-Yukon uplands. Many weather-aware Fairbanksians know that there are thunderstorms, say, out Chena Hot Springs Road than at the airport, but the airport is where there are long terms records. Here's a graphic showing the number of days each year (April-September) when thunder was reported. I have no explanation for the jump in 1988 beyond that it's thundering more frequently nowadays. It's not from automation, since thunder reporting in aviation observations has never been automated in Alaska. When I first noted this some years ago, I asked a former long time, now retired, NWS Fairbanks observer, and he could not recall any changes in procedure or instructions on when to report thunder in the late 1980s, or at any time during his career (coding, of course, changed with the advent of METAR code as the standard in 1996).
So far (through July 11) this summer there have been eleven days with thunder at the airport, already above the recent year average. The typical drop-off in thunderstorm frequency is coming soon, so can we come up with a climatologically based estimate of what will be the total number of days for 2020? To do that, I looked at 1988-2019 (32 years) and asked how many thunder days occurred from July 12 through the end of September. That ranged from zero (three years) to 10 (two years), with an average of 3.5 days. But of course, thunderstorms days only come as whole days (technically, this is an example of "count data"), i.e. there can never be a year with 14.5 thunderstorm days. So to create an outlook for the 2020 seasonal total, I fitted a Poisson distribution (good for count data since it is a discrete probability distribution) to the 32 years of data (the era of more thunderstorms). The primary assumption of a Poisson distribution is that the values are independent. In this case, the number of additional days with thunder this year is not conditioned by the days in 2019, or 1988, etc. That seems like a reasonable assumption in this case. So, putting all the pieces together yields this outlook with is based on nothing but history (no meterology): better than 50-50 chance that there will be 2 to 4 four additional thunder days by the time the snow flies.