Assessing the Rapid Refresh Forecast System data assimilation capability to represent an Amazon coastal squall line

Clusters of convective clouds organized in bands, which develop along the coastline of northern South America and propagate across the Amazonian basin, are known as Amazon coastal squall lines. Amazon coastal squall lines have been well studied, mainly using field campaigns held in the Amazon, reanalysis, and satellite derived precipitation estimates. Studies have also used numerical models to understand the mechanisms that drive the initiation, development, and propagation of these systems, and recently some have advanced to data assimilation applications using the data collected during field campaigns. However, numerical modeling studies simulating an operational framework are challenging and therefore hard to find in the literature.
This study aims to investigate the impact of assimilating all available data in a 3-hourly cycling configuration using the prototype Rapid Refresh Forecast System on the representation of an Amazon coastal squall line case study. The prototype RRFS is based on the FV3 Limited Area Model and the Grid point Statistical Interpolation analysis system. A domain with 3 km grid spacing covering part of northern South America is used to study a squall line case that occurred on July 5, 2020. Numerical experiments are conducted testing different configurations in GSI, such as multiple ensemble background error covariance weights in hybrid analyses, super saturation removal, the planetary boundary layer pseudo-observations function, as well as varied observation types and two physics suites: one based on the Global Forecast System (GFS) version 15 physics and a suite based on the High Resolution Rapid Refresh (HRRR) physics. Results show that large scale patterns are well captured in all experiments and the forecasts are improved when using data assimilation. Although the area studied has a low density data coverage, results are promising and will be presented during the seminar.