Evaluation of the GOES-R Proving Ground Convective Initiation Products during Different Convective Situations Through the Use of Case Studies in the Plains

To prepare a forecaster with use of the GOES-R convective initiation (CI) products in their forecast routines and to provide assessment of the University of Alabama at Huntsville (UAH) and the University of Wisconsin (UW) forecasting CI products, a COMET/NWS Cooperative project was developed between the NWS Valley/Omaha WFO and the University of Nebraska-Lincoln. For this project, case studies were archived for different CI situations during the summer of 2012. Examples of these case studies include cases of “boundary less” nocturnal CI on June 23rd, Nocturnal CI events from July 20th to the 25th, and CI associated with cold frontal passage on June 14th and July 27th. Our goal in case selection is to provide a wide variety of convection types in difficult to nowcast situations. From these case studies, general evaluations of the CI products and corresponding atmospheric conditions were documented regarding the convection situation to illustrate the usefulness of the CI products. The research focuses on the conditions present that act to help or inhibit algorithm performance, not on which algorithm works better. For example, both algorithms tend to have increased false alarm ratios at night, possibly due to higher amounts of CIN. We will also compare algorithm output to areas of moisture convergence as derived in RAP. During daytime convective situations areas of high CAPE and low CIN are sometimes not detected with the CI products when GOES east is not in rapid scan mode. We will also address UAH and UW algorithm team concerns that the CI algorithms produce higher non detection in areas of fast cloud motions. In addition, we will discuss CIN erosion as witnessed on algorithm output.

The probability of detection (POD) and false alarm ratio (FAR) of the different convective situations were also established and compared to the Rapid Refresh (RAP) hourly analysis forecast. Positive convection cases are determined by locations with a digital mosaic radar values greater than 35 dBz within two hours of a CI algorithm forecast. Both elevated and surface based convection will be considered in this study. In addition to the POD and FAR values, Briar scores and reliability diagrams will be presented for the University of Alabama's Satellite Convection Analysis and Tracking (SATCAST) algorithm, which produces a probability of convection forecast (known as Strength of Signal). These results will explore the possibility of creating “exclusion zones” where environmental factors act to inhibit algorithm performance (e.g. a CIN exclusion zone for values above a critical threshold). While it is not our goal to determine which algorithm is better, we hope that through these case studies we can achieve a better understanding of nowcasting atmospheric conditions.