by Professor Steven Businger
Abstract
The waveguide between the Earth's surface and the ionosphere allows very low frequency emissions generated by lightning, called sferics, to propagate over long distances. The advent of GPS and microprocessors recently opened the door for the development of specialized long-range lightning detectors to utilize this natural waveguide to monitor lightning activity over the open ocean.
Quantitative applications of data streams from long-range lightning detection networks (LLDN) require that the lightning flash detection efficiency (DE) and location accuracy (LA) of the network be taken into account. These depend on the geometry of the network, sensitivity of the sensors, and the nature of the wave guide. The approach that we took was to carefully construct DE and LA models by comparing accurate data from the U.S. National Lightning Detection Network (NLDN) with data collected by a long-range lightning sensor located in New Mexico. Since the conductivity of land surfaces are very different than that of an ocean surface, the DE model was calibrated for an ocean environment by comparing lightning data collected by a local network in Puerto Rico with LLDN data (to derive the salt-water peak current distribution and over-water space constants for the DE model). The space constants for open water turned out to be surprisingly large when compared to continental space constants, expanding the reach of the sensors over the ocean. Finally to test the accuracy of the DE and LA models, model predictions were compared with LLDN lightning observations over the Pacific Ocean and with NASA's Lightning Imaging Sensor (LIS).
The success of the models in predicting observed DE and LA over the Pacific Ocean helps facilitate quantitative applications of LLDN data streams over ocean areas. However, lightning rate data cannot easily be assimilated directly in the NWP models. To find relationships between lightning and storm properties that can more directly be assimilated, lightning data from the LLDN and LIS were compared to data from TRMM^Ò's precipitation radar and microwave imager. Three years of data over the North Pacific Ocean were analyzed. The data comparisons showed consistent logarithmic increases in convective rainfall rate, radar reflectivity, and ice water path, with increasing lightning rates. These relationships open the door not only to the possibility of assimilation of lightning data, but also allow for innovative ways to display lightning derived products in real time for nowcasting purposes. For example pseudo-reflectivity maps could be produced for data sparse ocean regions. In this way lightning data could be used to extend and refine the Divorak method for estimation of tropical cyclone strength.
Finally, a lightning data assimilation approach was developed to nudge the WRF model's latent heating profiles according to rainfall derived from DE-corrected lightning observations. Two case studies were undertaken. A rapidly deepening extratropical cyclone approaching the west coast of the U.S. was poorly forecast initially. The assimilation of lightning data improved the storm central-pressure forecast significantly. In the second case, a squall line associated with a Kona low moved over Hawaii. The location and timing of the squall line was improved using lightning data assimilation.
Summary and Future Work
Two results from our PacNet research have facilitated a breakthrough in long-range lightning detection performance. (i) Documentation of the slow signal attenuation over water and increased sensor sensitivity has resulted in much greater network range. (ii) Improved signal processing that ingests the whole waveform allows separation of ground wave from 1st and 2nd ionospheric hops, greatly reducing location errors.
These innovations are being implemented by Vaisala in a new network that promises have global reach. Development of nowcasting tools and visualization techniques are in their infancy. With the growing reach of the LLDN, there is an interesting opportunity to create tools that have tangible impact on decision making by airlines, maritime, and military interests.
The continuous nature of the LLDN data stream is ideal for data assimilation. Much work remains to be done in this area. Initially we implemented a nudging approach that adjusts the latent heating profile in the cumulus parameterization in areas experiencing lightning. However, a superior balanced approach, such as 4D-VAR, still needs to be developed. It is possible that other proxies for lightning (e.g., water-vapor profiles, stability, surface fluxes, etc.) could provide better simulation results under certain circumstances, such as cases of isolated convection, and these remain to be investigated. Tropical cyclone simulations require high resolution to resolve eye wall dynamics, thus we are developing lightning-data assimilation methods that adjust the explicit cumulus dynamics.
References
1. Pessi, A. T. et al., 2008: J. Atmos. and Ocean. Tech., 26, 145-166.
2. Pessi, A. T., and S. Businger, 2009: J. Appl. Meteor., 48, 833-848.
3. Squires, K. and S. Businger, 2008: Mon. Wea. Rev., 136, 1706-172.
4. Pessi, A. T., and S. Businger, 2009: Mon. Wea. Rev., In press.