3.2 Lightning climatology
The ISS LIS lightning climatology has been completed for the first three years of data (March 2017 - Feb 2020; Fig. 6a). Within ± 38° latitude, these results are broadly similar to the more than 13-year (after orbit boost; September 2001 - December 2014) TRMM LIS climatology (Fig. 6b). During this period, TRMM had a nominal orbit altitude of 402.5 km - very similar to the ISS orbit altitude range of 400-405 km. (The TRMM pre-boost orbit altitude was substantially lower, at 350 km, and thus is not considered here to keep the comparison more direct.)
Because ISS LIS has a shorter period of record (3 years vs 13 years), the lightning density maps are not as smooth despite the 0.5° gridding (Fig. 6a). There are also sampling limitations over low flash rate regions, such as the global oceans. However, notable hotspots [Albrecht et al. , 2016] from the TRMM LIS climatology (Fig. 6b), such as central Africa, Paraguay/northern Argentina/Rio Grande do Sul (Brazil), Lake Maracaibo (Venezuela), the Himalayas/Indian Subcontinent, and the Maritime Continent stand out, and feature comparable flash rate densities between ISS and TRMM LIS. For example, peak flash rate densities over central Africa exceed 80 km-2 yr-1 in both datasets. The well-known stark contrast between land and ocean lightning flash rate densities also stands out in both plots, as do the coastal enhancements in lightning over the Gulf Stream, near west Africa, the Caribbean and near Central America, near southeastern Brazil, the Bay of Bengal, etc. Despite the more limited sampling, ISS LIS also has been able to observe the small enhancements in lightning over the open Pacific, between ± 30° latitude, west of -120° (south Pacific) and -150° (north Pacific) longitude (including the Intertropical Convergence Zone, ITCZ). This is similar to TRMM LIS (Fig. 6b), but the patterns are more diffuse due to fewer samples. Note that portions of this more active ocean region are now under continuous observation by GLM-17. Additional notable lightning features that weren’t fully observed by TRMM LIS include enhancement over the Tien Shan mountain range near northwest China, and a slight enhancement over New Zealand (especially the north island).
In the global aggregate, lightning flash rates (between ± 38° latitude) are comparable between ISS LIS and TRMM LIS (Fig. 7). Both datasets show globally averaged flash rate ranging between 25 and 55 s-1. In addition to significant annual and interannual variability, both datasets also appear to show semiannual variability in the global flash rate, which is consistent with Williams[1994]. This, along with Fig. 6, demonstrates that ISS LIS is making fundamentally similar observations to TRMM LIS, and thus is capable of extending the TRMM LIS dataset over the tropics and subtropics for a longer time period.
ISS LIS enables coverage of higher latitudes (± 55°) compared to TRMM LIS (± 38°), providing renewed viewing of regions not observed by spaceborne global lightning sensors since the OTD mission ended in 2000. For example, ISS LIS reenables more complete viewing of the Great Plains of the United States (US), which features flash rate densities ~30 km-2 yr-1extending as far north as the border with Canada (Fig. 6a). The improved coverage of the continental US is a particularly important advantage of ISS LIS, because this coverage allows for a more robust examination of lightning/climate relationships within ongoing National Climate Assessment (NCA) studies [Koshak , 2017]. Another mid-latitude hot spot over Manchuria is also observed by ISS LIS, and the coastal enhancement of lightning near eastern South America is seen to extend further south. ISS LIS also provides coverage of most of Europe, including the lightning enhancement near the Alps. Lightning enhancement over Turkey is observed by ISS LIS.
The combined TRMM LIS and OTD dataset [Cecil et al. , 2014] provides a useful point of comparison for ISS LIS. OTD was in LEO orbit at 70° inclination and 740-km altitude [Christian et al. , 2003], so it provided coverage at higher latitudes than ISS LIS, but with reduced spatial resolution and geolocation accuracy. Table 1 shows a comparison of globally averaged flash rates from ISS LIS relative to the OTD and TRMM LIS climatology published by Cecil et al.[2014]. ISS LIS is measuring slightly lower flash rates, but the numbers are generally within 5-10% of the previous climatology, which is well within the magnitude of expected offset from the reduced effective detection efficiency in version 1 ISS LIS data (Section 3.1), the sampling limitations of the 3-year ISS LIS record, as well as interannual variability (e.g., Fig. 7). Even the relatively larger discrepancies seen between TRMM LIS/OTD and ISS LIS during December-February are reasonably attributed to the above causes as well, since the differences are still within ~15%. Note also that the ISS LIS global flash rates in Table 1 are not smoothed, unlike the TRMM LIS/OTD values.
Relative to Cecil et al. [2014], ISS LIS has observed potentially higher flash rate densities in notable mid-latitude areas - such as Turkey and the Middle East, southern Canada, Manchuria, Europe and Northern Africa (Fig. 6a). However, caution in interpreting the 3-year ISS LIS dataset is required, since the relative impacts of individual storms may be influencing these differences. Integration of ISS LIS observations into the full LIS/OTD gridded dataset, which will enable detailed quantitative comparisons for individual regions, is planned for a future study. This planned analysis should be able to determine if, and to what extent, lightning has increased globally at higher latitudes as a result of climate change, relative to the OTD era (1995-2000) [e.g., Veraverbeke et al. , 2017; Williams , 2020].
The seasonal distribution of lightning from ISS LIS also follows expectations established by previous global climatologies (Fig. 8). Globally, lightning is maximized during June-August (Fig. 8b); however, both March-May and September-November also have significant activity (Fig. 8a, c). Notably, in boreal autumn the northern Great Plains of the US can remain active, even as similar latitude locations in Europe, for example, see a substantial decrease from the summertime peak (Fig. 8c). The Manchuria lightning peak is primarily a boreal summertime phenomenon, with a significant decrease in both spring and fall. Lightning in the Middle East is most prevalent during boreal spring and fall, providing evidence for a semiannual signal in lightning in certain regions of the globe [Williams , 1994]. Turkey reaches its maximum in summer. Boreal winter leads to a significant reduction in northern hemisphere lightning (Fig. 8d); however, there are noticeable hot spots remaining in the US Gulf Coast. The lightning peak near Paraguay is most distinctive during austral spring (Fig. 8c). These basic seasonal patterns are also observed in the TRMM LIS/OTD dataset [Cecil et al. , 2014].
Globally averaged diurnal variability of lightning (Fig. 9) follows the typical patterns observed in previous climatologies [Virts et al. , 2013; Blakeslee et al. , 2014; Cecil et al. , 2014]. Namely, the diurnal cycle over land drives the overall global diurnal variability in lightning, with the ocean flash rate essentially flat throughout the day and night. On average, lightning peaks in the local afternoon (3-4 pm Local Solar Time, LST), and reaches a minimum near 10 am LST (Fig. 9a). The timing and the approximate dynamic range in the LST reference frame (15-100 s-1) are comparable to the analysis of Blakeslee et al. [2014] for TRMM LIS/OTD. Viewed in UTC time coordinates (Fig. 9b), lightning follows the classic Carnegie-like curve structure [Mach et al., 2011], peaking during 18-19 UTC. The timing and the 30-60 s-1 dynamic range are very close to Blakeslee et al. [2014] as well. Note that the approximately 2x larger global diurnal variability in flash rate, versus the diurnal variability of electric field in the Carnegie curve, is explained by the effect of higher currents in oceanic thunderstorms, as well as the influence of electrified shower clouds [Mach et al. , 2009; 2010; 2011].