Figure 6. (a) Raw, un-geolocated ISS LIS background image from
approximately 0953 UTC on 9 August 2017. (b) The same ISS LIS background
image but geolocated, showing ISS LIS events detected in the domain of
the image during the overpass. The suspect LIS flash not seen near
CATS-inferred cloud also is indicated, and LIS and CATS boresight ground
tracks are shown.
Further analysis found only 6 instances where there were no apparent
clouds in the LIS backgrounds. An example is shown in Fig. 6, which
appears to show glint from a water surface (in this case, the Persian
Gulf) made it through LIS data filters. Other cases (not shown) suggest
that snow fields in mountainous regions also may have contributed to
these false alarms. Overall, this analysis demonstrated that the surface
glint-based FAR for ISS LIS is ~0.1% during daytime (6
confirmed false alarms out of 65 candidates, which themselves only
represented ~1% of the daytime matchup dataset). Since
these surface glint filters were inherited from the long-term TRMM LIS
mission, this excellent performance is a highly encouraging sign of the
quality of the long-term combined LIS dataset (1997-present).
4. Discussion and Conclusions
This was a pathfinder study that made use of a short
(~8-month) overlapping dataset to explore the viability
of combining spaceborne lidar and lightning observations to study
thunderstorms. The results demonstrate that lidar observations can
provide insight into the global thunderstorm climatology, are
realistically correlated with lightning characteristics such as flash
rate despite the lack of deep penetration into thunderclouds, and enable
new methods for quality control and validation of spaceborne optical
lightning observations.
Specifically, this study found that lidar-inferred cloud tops near
lightning behave realistically given how the tropopause slopes downward
from the tropics to the poles. In addition, these cloud-tops average
approximately 2-km higher in altitude compared to radar-based
climatologies using 20-dBZ echo tops. This demonstrated the advantage of
using lidar as a more accurate measure of cloud-top height compared to
less-sensitive Ku-band radar (e.g., TRMM, GPM), but at the same time
also points toward a way to use lidar to help scale radar-based
climatologies to more accurately convert echo-top heights to cloud-top
heights.
Moreover, this study found that a proxy for lightning flash rate shows
reasonable behavior relative to lidar-retrieved cloud properties, such
as cloud optical depth, cloud-top height, and IWP. The results also
demonstrated that CATS’ cloud-top phase retrieval worked with likely
90% or better accuracy.
Finally, this study determined that the false alarm rate for
LIS-identified flashes associated with no nearby cloud (e.g., solar
glint off water) is ~0.1%. This demonstrated that the
surface glint filters in the ISS LIS processing code (inherited from the
much longer TRMM dataset) are working extremely well and that surface
glint likely is a nearly negligible contributor to LIS false alarms
(which have been estimated at less than 5% of the LIS dataset;
Blakeslee et al., 2020). This is in contrast to, e.g., GLM where surface
glint has been a major contributor to false alarms, particularly before
implementation of the blooming filter in GLM processing (Rudlosky et
al., 2019).
Based on this pathfinder study, fruitful scientific insights are
expected from larger combined lidar/lightning datasets. These could
include analysis combining GLM with the Cloud-Aerosol Lidar with
Orthogonal Polarization (CALIOP; Winker, 2022). In addition to building
on the analyses performed in the present study, such an analysis could
use CALIOP to help validate stereo retrievals of cloud-top height from
dual GLM observations (Mach & Virts, 2021). Moreover, potential future
datasets, including a lidar (or lidars) hosted on the AOS mission, are
also exploitable using lessons learned from this study. These larger
datasets potentially could mitigate some of the limitations of this
study, specifically the short time period used here, which didn’t even
cover one full annual cycle. Moreover, a larger dataset also would
enable the use of tighter time/distance thresholds for defining
instrument conjunctions, potentially leading to more accurate
comparisons.
In addition, a larger number of samples may allow for detailed
investigation of aerosol interactions with thunderstorms. Since lidar
can determine aerosol type (e.g., Omar et al., 2009), lightning
enhancement (or suppression) by smoke, dust, and other aerosol types
could be studied.