Impact Assessment of Loon Stratospheric Balloon Observations Assimilated in NOAA’s Finite Volume Cubed Sphere Global Forecast System (FV3GFS)
Stratospheric balloon observations have the potential to fill gaps in the current earth observing system architecture by providing in situ observations in an otherwise in situ data-sparse region of the atmosphere. At present, stratospheric observations are derived primarily from satellite data and are difficult to validate given that few in situ measurements are available. This study aims to exploit existing in situ balloon observations provided by Loon, a subsidiary of Google’s parent company Alphabet, to supplement the observation architecture toward improving forecasts and for validating satellite measurements. In recent years, Loon has launched over a thousand large superpressure balloons that carry communication instrumentation in the stratosphere for months at a time to provide continuous interconnectivity to internet-deficient regions around the world. In addition to the communication instrumentation, the balloon payload contains a series of sensors that downlink information including 4-dimensional GPS data, solar altitude relative to the balloon, and atmospheric observations.
This study evaluates Loon stratospheric wind observations and the impact of their assimilation in the NOAA’s Gridpoint Statistical Interpolation (GSI) based Finite Volume Cubed Sphere Global Forecast System (FV3GFS). Prior to assimilation, the balloon data quality is assessed. The balloon winds are compared with NOAA Global Data Assimilation System (GDAS) and European Centre for Medium-Range Weather Forecasts (ECMWF) analyses by collocating the analyses with the balloons in space and time. Statistical diagnostics reveal small biases in the wind observations. Applications of quality control (QC) parameters that are identified are shown to reduce systematic errors likely related to the onboard sensor instrumentation. The assimilated Loon data are preprocessed using such QC metrics.
To assess the impact of Loon stratospheric wind observations on numerical weather prediction (NWP), NWP observing system experiments (OSEs) are conducted for a 2-month period: December 1, 2018 through January 31, 2019. The first experiment acts as a control and represents NOAA’s current operational forecast model setup (as of January 2020). The model consists of the FV3 dynamical core, 80 ensemble members, C384 (~25 km) and C192 (~50 km) horizontal resolutions for the analysis and forward operator respectively, and 64 vertical levels. The second experiment is identical to the control but with Loon winds assimilated. The impact of assimilating the Loon wind observations on NOAA global weather prediction is discussed.
