Lightning in the Arctic

World Wide Lightning Location Network (WWLLN) data on global lightning are used to investigate the increase of total lightning strokes at Arctic latitudes. We use the summertime data from June, July, and August (JJA) which average >200,000 strokes each year above 65°N for the years 2010–2020. We minimize the possible influence of WWLLN network detection efficiency increases by normalizing our results to the total global strokes during northern summer each year. The ratio of strokes occurring above a given latitude, compared to total global strokes, increases with time, indicating that the Arctic is becoming more influenced by lightning. We compare the increasing fraction of strokes with the NOAA global temperature anomaly, and find that the fraction of strokes above 65°N to total global strokes increases linearly with the temperature anomaly and grew by a factor of 3 as the anomaly increased from 0.65°C to 0.95°C.

though WWLLN has five stations in Antarctica. As such we consider only northern hemisphere summer data.
In this study, we discuss the temperature and lightning data sets to be used and motivate the work with a global look at both the Earth's temperature anomaly as well as northern latitude lightning distributions. We then present an analysis using latitudinal and annual lightning variations to investigate the increasing fraction of global lightning which occurs at high latitudes. These lightning data are then compared to the three-month global temperature anomaly for northern hemisphere summer to arrive at a linear relationship between the fraction of global lightning occurring above 65°N, with the temperature increase. If this trend continues and the Earth has another 0.5°C global temperature increase, then the lightning stroke rate in the Arctic could increase by 100% from the 2020 stroke level.

Data Sets
Here we introduce the global temperature data set available from NOAA's National Centers for Environmental Information (NCEI) and the lightning data from WWLLN. NCEI tracks the Earth's temperature variations and has global temperature change data from 1880 to today (NOAA, 2020a). Figure 1 is a plot of the full globe, yearly temperature anomaly for JJA, showing the, now well known, steady increase of global temperature. Data used for these plots comes from over 1200 stations around the globe, and ocean data comes mostly from an extensive network of buoys (NCDC, 2020). The units are given in degrees C above the average temperature for the 20th century (0° line in Figure 1). We will concentrate on the yearly temperature anomaly changes corresponding to the 11 year time period of the WWLLN lightning data (2010-2020, toward the end of the plot in Figure 1) and use the global temperature data from the June-July-August (JJA) period each year corresponding to the summer time of the lightning data (NOAA, 2020b).
The World Wide Lightning Location Network (WWLLN) has been locating lightning strokes globally since 2004. The WWLLN uses the radio energy emitted by lightning (in the VLF, Very Low Frequency-range) and detected at receivers all over the world to locate lightning using the time of group arrival (Dowden et al., 2002;Hutchins et al., 2012;Rodger et al., 2006;Virts et al., 2013). Lightning produces a strong, narrow impulse during each return stroke which results in the emission of radio frequency energy which peaks in the range of 10-15 kHz (e.g., Dowden et al., 2002;Malan, 1963). This narrow impulse, which can be recognized as a transient even at AM or FM radio band frequencies, produces a wave packet which propagates around the world in the Earth Ionosphere Wave Guide. During propagation the wave packet spreads out in frequency by a process called dispersion, requiring a careful analysis of the wave packet to find the time of group arrival (TOGA) (Dowden et al., 2002). When the TOGA is measured with 100 ns absolute accuracy by several widely spaced receivers, it is possible to locate lightning to within < 5 km and within a few microseconds. Currently WWLLN locates 600,000-800,000 strokes globally every day, while in the past the detection efficiency was about 10% of what it is today. Here we use data from 2010 to 2020 to analyze the increase in high latitude lightning and concentrate on the northern summer months of JJA. We can see an increase in global strokes located with six or more WWLLN stations during these months, from 3.21 × 10 7 strokes in the northern summer in 2010 to about twice that beginning in 2014, and more or less steady after that. The red curve counts the average stations operating each year during the 92 days of the JJA months. Comparing the red to the blue curves one can see that the increase in strokes (blue) is closely associated with an increasing detection efficiency due to the increasing number of stations (red). However, after 2014 the total global stroke count in Figure 2 varies by less than 10%, as does the station count. There is little or no lightning in the Arctic outside of northern summer time, hence we focus here on just those three months each year.
The distribution of high latitude lightning found during these 11 years is shown in Figure 3, which is a plot of just the strokes poleward of 75°N latitude. In Figure 3 we can see that the stroke distribution is dominated by lightning in the eastern hemisphere from about 70°E to 170°E, with relatively little lightning north of Canada/Alaska by comparison. This is probably due to the fact that mainland Canada is mostly south of 70°N, while mainland Russia reaches up to over 77°N, with substantial mainland Russia north of 70°N latitude.
HOLZWORTH ET AL.    Table S1 for the actual strokes, and Figure S1 to see the WWLLN stations which located these strokes). This paper does not address the meteorology associated with this northern intrusion close to the pole, but it is clear that it is associated with an energetic, well organized event which lasted for hours and will be examined in a future study.  Figure 2). The red line in Figure 4 is the result of this adjustment due to the increasing number of WWLLN stations: with the primary effect being only a few tens of percent increases in the first few years. Here one can see that in Figure 4 the adjusted histogram (red line) still indicates a great increase in the number of WWLLN strokes north of 65° latitude over this 11 year period. So, there is no evidence from our growing station locations that the relative number of strokes detected in the Arctic would be favored in any way. In fact, one might have expected a reduced ratio in the Arctic, due to increasing total global WWLLN stations, which are all outside the Arctic.

Analysis of High Latitude Lightning
We plot in Figure 5 (blue line) the fraction of total global strokes during JJA each year so that the increasing detection efficiency effect is minimized. In this Figure 5, the blue plot refers to the total well located strokes above 65°N normalized by the total number of WWLLN-observed global strokes in that summer time period. Comparing the blue line to the histogram or red line in Figure 4, one can see that the plot strongly reflects the increasing total strokes above 65°N, including the relative dips in 2015 and 2018. Thus, Figure 5 is evidence that the fraction of global lightning occurring north of 65°N has increased by over a factor of three during this time period (from 0.002 to over 0.006 or 0.2%-0.6%).
Another point to make is that the increase is evident even just looking at the 7 year period from 2014 to 2020 when the WWLLN detection efficiency did not vary by more than 10% (as discussed above). We looked at three 10° latitude bands starting at 45°N and found the same trend in each separate band as shown for all strokes north of 65°N in Figure 5 blue line (see Figure S2 for the latitude bands detail).
Figure 5 also includes the three month (JJA) global temperature anomaly in degrees Celsius reported by NOAA (NOAA report 2020 and Table 1 below). These temperature anomaly data are from the same NOAA data set used in Figure 1. This Figure 5 demonstrates the strong similarity between the fraction of strokes above 65°N and the 3 month average global summer temperature anomaly for JJA for the 11 year period of the WWLLN stroke data.
There is obviously a correlation between the blue and red plots in Figure 5, which we quantify in Figure 6. At the very least, the two linear trends are consistent. In Figure 6 we see that the linear correlation coefficient is R = 0.802 and R 2 = 0.644. In this figure, we can clearly see the increase in the fraction of total strokes occurring at high latitudes has increased by a factor of three during the temperature increase of 0.3°C in the global 3-months average global temperature anomaly (from 0.65°C to 0.95°C). As discussed in the introduction, the evidence and modeling regarding any possible global lightning increase with global temperature is mixed at best. The global WWLLN data in Figure 2 may show a slight upward trend from 2014 to 2020 when there was no clear trend in WWLLN detection efficiency. So, it is possible that total global strokes may indeed increase, but not by much compared to the large increase of Arctic lightning from 2010 to 2020 (as seen in Figure 4).
To put this in raw terms we could say that if one thinks there are, say, 44 lightning strokes per second globally (e.g., Christian et al., 2003), which would be 0.35 × 10 9 strokes globally during three months of summer, then we can expect 0.006 × 350 million strokes = 2.1 million strokes to occur in the Arctic (all in the summer) or about 23,000 every day of JJA. We note that the total global strokes per second is not known from ac-HOLZWORTH ET AL. tual measurements from any network or spacecraft data set, but rather is projected from existing global lightning measurements (e.g., Christian et al., 2003). WWLLN has a detection efficiency variously identified as between 10 and 15 percent of all global strokes and currently up to 80% of all strong global lightning (Holzworth et al., 2019), so we would expect to see 210,000 to 315,000 WWLLN strokes above 65°N every summer. In fact we directly measured 380,000 strokes in 2020 during JJA the subset of those strokes above 75°N were plotted above in Figure 3.
We can ask what these numbers suggest for future Arctic lightning? We can use the regression line in Figure 6 to look out to a time in the future beginning from the current values in 2019-2020 (where the ratio is 0.0059) and the Temperature anomaly is 0.93°C. Then, when the global temperature anomaly has increased by just 0.5°C (to 1.45°C), the fraction of strokes in the Arctic (vertical axis Figure 6) would increase to 0.011 or about 100% of the current total global lightning. This assumes the total global stroke rate does not change, but if the total global stroke rate increases, while the fraction in the Arctic also increases, then the total net increase in the Arctic could be much more.

Data Availability Statement
These data are available at nominal cost from http://wwlln.net.