Spatial Variations in Moisture and Precipitation Forecast Errors from Satellite and Dropsonde Data Assimilation in Northern California Atmospheric River Events

Orographic precipitation in landfalling atmospheric river events is controlled to a great degree by large scale upslope water vapor flux, however the spatial distribution of precipitation has significant variability that is difficult to forecast accurately. We attempt to quantify the dependence of the forecast error on deficiencies in the initial moisture fields. We assess the impact of assimilating spaceborne GPSRO observations in combination with other satellite and conventional data into mesoscale forecasts for two case studies off the west coast of northern California. One event was observed during the CalWater2015 field campaign and a second event was a strong AR event that occurred during the peak of the COSMIC GPS RO observational coverage in 2009. Both of these events resulted in significant precipitation along the coast and nearby mountain ranges.

While assimilation of GPSRO observations has shown a significant positive impact on global NWP forecasts, especially at the upper troposphere/lower stratosphere, their impact at lower levels in regional mesocale forecasts is still uncertain. This is partly because of the challenge of assimilating observations with a long horizontal averaging length in highly variable lower level moisture fields, such as those associated with frontal systems. Data assimilation techniques using a non-local observation operator have been developed to accommodate this, however the balance of observation and background model error has not been sufficiently well investigated. The CalWater2015 field campaign included research flights that sampled its associated AR with numerous dropsondes while it was still offshore of California, and these provide a unique data set to evaluate the effectiveness of the data assimilation method.

This study assesses the impact of assimilating the GPSRO observations in combination with SSMI, ocean vector winds, atmospheric motion vectors, conventional observations and dropsondes on high spatial resolution mesoscale forecasts of the atmospheric river events. The impact of the data assimilation on key features of the offshore atmospheric river, including the moisture and wind fields themselves, as well as the integrated vapor transport, are quantified. After landfall, the impact of the assimilation of the various observations on the forecasts of the low-level winds, moisture, and precipitation over California is quantified. In particular, GPS precipitable water vapor observations are used to investigate the spatio-temporal variations of moisture error and their relationship to precipitation errors.