Using GOES-16 to Improve Short-Term Forecasts

Imagery from the Geostationary Operational Environmental Satellite (GOES) system has been a key element of U.S. operational weather forecasting for over 40 years. However, usage of GOES imager data for numerical weather prediction has generally lagged its usage by human forecasters. Now with the launch of the first in the latest generation of satellites, GOES-16 provides three times more channels, four times better spatial resolution, and five times faster coverage than the previous generation of GOES. These advances in observation capability provided by GOES-16 are a good opportunity to revisit the value of GOES-16 for improving short-term forecasts over the contiguous U.S.

Our investigation will focus on data from GOES-16 imager (ABI). We will evaluate and discuss models to transform ABI radiances into estimates of latent heating. We find that even when restricted to just heritage channels, such as the red visible channel and long-wave infrared window channel, the higher space- and time-resolution imagery from GOES-16 provides enhanced capabilities for locating convective initiation and identifying convective updraft cores. We will examine situations for which cloud-top phase and cloud particle size information from the new channels on GOES-16 provides information where signals in the visible or infrared channels are ambiguous. We will consider use of the new cirrus band for lowering false alarms from the use of visible imagery. ABI-derived information on convective cores will be compared with features in ground-based radar reflectivity from MRMS.

New methods for using GOES-16 data in model initialization and data assimilation will be developed and tested for the operational HRRR model. Like its parent model, RAP, HRRR is built on the WRF ARW core and uses GSI for data assimilation. ESRL GSD has developed a special cloud analysis package for GSI. The digital filter initialization in WRF is used by RAP/HRRR to ingest radar information, and we will show how this can be extended to use information from GOES-16. We will consider how lead-time can be increased for situations involving convective initiation using both GOES-16 and ground-based radar together, since GOES-16 is able to detect CI before precipitation echoes are observed by radar. The build-up of hydrometeors and latent heating (temperature tendency) using GSI will be shown. The impact of using GOES-16 on the skill of short-term (1-6 hour) forecasts from HRRR will be compared against operational HRRR, and forecast fields of reflectivity and precipitation will be compared with MRMS and Stage IV observations.

We will present results from case studies that use GOES-16 to initialize convection in the HRRR model for severe storms at different points in the convective lifecycle. This will include a case of convective initiation over the Southeast U.S. and mature thunderstorms over the Central Great Plains. Additional cases capturing different convective modes of organization and highly dynamic fire weather environments will be considered as time permits. The value of GOES-16 data for high-resolution numerical weather prediction will be discussed in the context of existing observation systems, which currently rely heavily on ground-based radar observations.