Initial Assessment of GLM Performance Using Vaisala’s GLD360 Dataset

In mid-December 2017, the GOES-16 satellite reached its operational GOES-East location. At this time, a focused period of assessment of the Level 2 (L2) product from the onboard Geostationary Lightning mapper (GLM) was initiated, in preparation for the Peer Stakeholder Product Validation Review in January, 2018. Because of the efforts of several research group’s assessments at various spatial and temporal scales, the GLM L2 product achieved “Provisional Validation Maturity” on January 19, 2018.

We report here on various performance metrics used to evaluate GLM relative to Vaisala’s GLD360 during late December, 2017. GLM Group-level data were used to evaluate timing offsets, location offsets, and outlier events/groups as a function of time-of-day. Group data correlated extremely well with lightning processes reported by ground-based VLF/LF lightning locating systems. Group centroid locations (lat/lon) were generally within one GLM pixel (~8 km) of the GLD360 location, but location differences grew large with increasing distance from the satellite nadir “location” (lat/lon). Misplaced (false/outlier) groups were localized to three regions at certain times of the day. GLM flash detection efficiency (DE) relative to GLD360 varied about 20% over the diurnal cycle and varied somewhat by region. GLM detected 77% (and GLD360 reported 74%) of the combined flashes reported by GLM and GLD360. These results lead to further assessment of GLD360 suggesting that more than half of the discharges reported by GLD360 are associated with in-cloud processes.

Bounds on the uncertainty of DE estimates are difficult to establish unless there is a perfect reference system. Although a few Lightning Mapping Array (LMA) systems can provide this reference in localized regions, no such reference is available throughout the GLM domain. We present work-to-date on characterizing and bounding these uncertainties.