290 likes | 440 Views
GOES-R3 Annual Meeting GLM Algorithm Development. June 10-11, 2010 Madison, Wisconsin Dr. William Koshak, Dr. Richard Blakeslee NASA/MSFC/VP61. Project Participants, Purpose, Interactions. PI and Other Participants : Dr. William Koshak (NASA-MSFC)
E N D
GOES-R3 Annual MeetingGLM Algorithm Development June 10-11, 2010 Madison, Wisconsin Dr. William Koshak, Dr. Richard Blakeslee NASA/MSFC/VP61
Project Participants, Purpose, Interactions • PI and Other Participants: • Dr. William Koshak (NASA-MSFC) • Dr. Richard Solakiewicz (Chicago State Univ.) • Purpose: Retrieve ground flash fraction in a set of N flashes observed by GLM. • Interactions with AWG: Collaborate with AWG (D. Mach) to make sure that optical group product maintains attributes required for ground flash fraction retrieval. Koshak/NASA/MSFC/VP61
(Somewhat) Unexpected Result Maximum Group Area (MGA) much bigger on average for ground flashes than cloud flashes Cloud Flashes: Ground Flashes: Mean = 493.0 km2 Mean = 215.6 km2 Koshak/NASA/MSFC/VP61
Major Accomplishments • Significant progress has been made in defining, testing, and applying novel methods for retrieving the fraction of ground flashes in a set of N flashes observed from a satellite-based lightning imager (e.g., GLM, LIS/OTD). • 3 full-length journal articles have been written, submitted, and accepted. Papers discuss the optical data, three retrieval methods, numerical simulations, and preliminary application results. • In particular, Koshak developed a bayesian retrieval method using a mixed exponential distribution model: • This is the best retrieval method to date. • Maximum Group Area (MGA) data is inverted by the retrieval method. Here, several “optical groups” compose a “flash”, and MGA is a type of “return stroke detector”. • Retrieval errors from simulations are very reasonable when N > 2000 flashes. • Preliminary global-scale retrievals have been completed using 5yr OTD dataset. Koshak/NASA/MSFC/VP61
3 Completed Journal Papers • Koshak, W. J., Optical Characteristics of OTD Flashes and the Implications for Flash-Type Discrimination, accepted , JTECH, May 2010. • Provides statistical distributions of lightning cloud top optical emissions. • Provides foundation for retrieving ground flash fractionαbased on mean optical data & the Central Limit Theorem. • Koshak, W. J., R. J. Solakiewicz, Retrieving the Fraction of Ground Flashes from Satellite Lightning Imager Data Using CONUS-Based Optical Statistics, accepted pending (presentation) revisions, JTECH, May 2010. • Provides two methods for retrieving α based on CONUS mean data • Koshak, W. J., A Mixed Exponential Distribution Model for Retrieving Ground Flash Fraction from Satellite Lightning Imager Data, accepted pending minor revisions, JTECH, May 2010. • Provides bayesian retrieval of α • Idea is to maximize LHS … is called the Maximum A-Posteriori (MAP) solution (Bayes’ Law)
Retrieval Errors from Simulations[an example of mean (std dev) ground flash fraction error from Bayesian algorithm] Mean ground flash fraction (alpha) error Retrieval errors are shown as a function of other modeling parameters (Ug,Uc) in the problem. Koshak/NASA/MSFC/VP61
Preliminary Global-Scale Retrieval • Based on the 5yr • OTD dataset. • Using Bayesian Algorithm • Only lat/lon bins with • 2000 or more flashes • were analyzed. • Lat/lon bins are • 4 degrees x 4 degrees (ground flash fraction) Koshak/NASA/MSFC/VP61
Additional Activities Leveraged • NASA Post Doc Program: was used to support Solakiewicz for 1 year in his co-authoring efforts of journal paper #2 described above. • NASA LIS Project: LIS dollars used to fund some of this research. • NASA Applied Science Activities: Beneficial synergy exists between funding from NASA AS Air Quality Program and this R3 work (i.e., both improve understanding of lightning NOx production important to air quality & climate modeling). Koshak/NASA/MSFC/VP61
Recommendations for Follow-On Research • Perform additional rigorous tests of existing bayesian retrieval method • Build additional confidence • More cases to improve robustness of retrieval error analyses • Apply to LIS data • Examine connections with Gröbner Basis solution theory, and exploit if beneficial. Koshak/NASA/MSFC/VP61
Lightning Cell Tracking - example Top: Reflectivity (L) and small scale cells (R) Bottom: Lightning (L) and large scale cells (R) w2segmotionll -i /usbdisk2/GLM/proxy/netcdf/frd_vil/code_index.fam -o /usbdisk2/ GLM/proxy/netcdf/frd_vil/segmo -T "frd_vil" -d "1 50 5 -1" -p "1,50,100:1:1,0.2, 0.2" -S -O 5 -k none (click through sequence)