Lightning Imaging Sensor (LIS) sampling error due to TRMM orbit

This study examines the impact of the low earth orbit of the Tropical Rainfall Measuring Mission (TRMM) satellite and the sparse observation time of the Lightning Imaging Sensor (LIS) on the monthly, seasonal and yearly total lightning climatology derived from LIS observations. The LIS was launched onboard the TRMM satellite into a 35° inclination, 350 km altitude, precessing orbit in November 1997. In August 2001 the orbital altitude was raised to 402.5 km to reduce atmospheric drag and extend mission life. Pre-orbit boost, the LIS sampled a field of view of about 580 by 580 km for 83 s as it passed overhead. Post-boost, the field of view increased to about 680 by 680 km and the sampling time increased to 97 s.

Total lightning climatology studies have binned the LIS data into various bin sizes (e.g., 0.5°, 2.5°). Due to the precessing orbit, a bin may only be seen once or twice a day at the equator (once on the ascending portion of its orbit and once on the descending portion) or about 164 s per day of observation time. Near the northern and southern extents of the TRMM orbit, the satellite may observe a bin 3-4 times per day or up to 388 s per day. In addition, to sample the entire diurnal cycle at a grid point takes 23.2 days (equator) and 46.4 days (at northern/southern extents). These times increase to 23.8 days and 47.5 days post-boost. Since lightning activity is strongly correlated to the local time, it is important to obtain samples from the entire diurnal cycle to get good estimates of monthly, seasonal, and yearly lightning activity.

In this study, we examine the errors in the monthly, seasonal, and yearly lightning within various bin sizes over the portions of the U.S. observed by LIS. As a proxy ground truth dataset for LIS observations, we use lightning flash data from the National Lightning Detection Network (NLDN). Although the NLDN and LIS have different detection techniques and detection efficiencies, the NLDN provides a dataset with continuous and well characterized temporal and spatial characteristics. Using the actual TRMM orbit, we “fly” the TRMM satellite over the NLDN observations and determine which NLDN flashes occurred within the LIS field of view. Then, the same analysis is done but with an offset in the TRMM orbit time to simulate changes in sampling that occur as a result of TRMM's orbit precession. This is repeated a number of times to obtain simulated LIS datasets at different sampling times. Then for various bin sizes, we compare the number of simulated LIS observations to the actual number of NLN flashes that occurred at monthly, seasonal, and yearly time scales. From this analysis, an estimate of the sampling error due to the TRMM orbit and lightning temporal characteristics are determined.