92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Wednesday, 25 January 2012: 4:15 PM
Probabilistic Nowcasting of Severe Convection Using the Temporal Evolution of Satellite-Derived Deep Convective Cloud Properties
Room 256 (New Orleans Convention Center )
Justin Sieglaff, CIMSS/Univ. of Wisconsin, Madison, WI; and M. J. Pavolonis and D. C. Hartung

Geostationary satellite observations are an effective tool for quantifying the macro- and microphysical properties of deep convective clouds as a function of time. Various cloud properties can be derived from geostationary weather satellites, such as the Geostationary Operational Environmental Satellite (GOES), including: cloud top temperature (day and night), cloud top height (day and night), cloud top phase (day and night), visible cloud optical depth (day), and cloud emissivity (day and night) at high temporal resolution (five to fifteen minute). These cloud properties, some of which can be fully retrieved throughout the diurnal cycle, can then be tracked in time to provide insight into the micro- and mesoscale characteristics of different types of convection. By treating individual clouds as objects and tracking them over time through space, we are able to capture and study the unique temporal trends in the above satellite-derived cloud properties of individual objects. Testing various combinations of the temporal trends of these cloud properties as predictors in a na´ve Bayesian framework has suggested that such a probabilistic approach is capable of providing at least 20-40 minutes of lead-time prior to radar-indicated severe criteria for a developing severe thunderstorm. Therefore, in an operational setting, forecasters would be able to potentially use such a tool in issuing a warn-on forecast or severe weather statement prior to the observation of radar-indicated severe criteria on a cell depending on its assigned probability. This work lends new insight into the use of real-time satellite data in the predictability of severe thunderstorms and the subsequent addition of lead-time to public warning of approaching severe weather, ultimately increasing public safety. Specifically we will present a description of the predictors used in the na´ve Bayesian model and show both severe thunderstorm and non-severe thunderstorm independent test cases of the na´ve Bayesian probabilistic severe thunderstorm algorithm. Additionally we will present progress on fusion of NEXRAD and NWP data into the Bayesian model.

Supplementary URL: