Figure 4. (a) Maximum CATS IWP near a cluster of ISS LIS flashes vs. number of flashes in the cluster. The best-fit line is shown in red. (b) Maximum CATS cloud optical depth near a cluster of ISS LIS flashes vs. number of flashes in the cluster. The best-fit line is shown in red. (c) Maximum CATS cloud-top height near a cluster of ISS LIS flashes vs. number of flashes in the cluster. The best-fit line is shown in red.
Next, LIS-detected flashes were clustered following the methodology described in Section 2.3 and compared to the maximum CATS-inferred cloud properties (IWP, cloud-top height, and optical depth) within 50 km of the cluster along the CATS ground track, and the results are shown in Fig. 4. At first glance, the results are not as good as radar-lightning analyses (e.g., Wiens et al., 2005), but in all subpanels the best-fit lines have Spearman correlations ranging between +0.38-0.42, with significance values > 99%. That is, the analysis found that linear correlations between a proxy for lightning flash rate and IWP (Fig. 4a), cloud optical depth (Fig. 4b), and cloud-top height (Fig. 4c) were all positive (which is expected and physically meaningful) and highly statistically significant. These metrics demonstrate the utility of comparing lightning measurements with lidar-inferred cloud properties, satisfying the success criteria for this analysis. Higher-power polynomial fitting (e.g., 5th power relationships like Price and Rind, 1992) was not attempted due to the relatively small number of samples available in this study. Instead, the focus was simply to determine whether simple yet statistically significant relationships could be found.
3.3 Using lidar to explore LIS false-alarm rate
Blakeslee et al. (2020) reported an ISS LIS false alarm rate (FAR) under 5%, based on comparisons with other reference lightning datasets. This FAR likely results from unfiltered radiation-induced noise as well as a combination of unfiltered cloud-based and surface-based solar glint. The latter is nominally controlled by a surface glint filter already contained within the ISS LIS processing code. This filter checks for the continuous occurrence of transients over many successive frames (not just intermittently like lightning) in a location that stays fixed within the instrument’s frame of reference (similar to how solar glint appears to an observer when viewing from, e.g., an airplane flying over the sunlit ocean). It would be useful to use CATS “no cloud” null cases (Fig. 3) to evaluate the performance of this surface glint filter.