Assimilation of In-Situ Observations

In-situ observations from radiosondes, surface land and marine stations, and aircraft play an important role in data assimilation for global numerical weather prediction, even in the current era when observations from satellites provide approximately 90% of the observations assimilated. In addition, these observations are widely used in verification for both model forecasts and satellite datasets, and radiosonde data serve as critical anchor observations in variational bias correction schemes used primarily for satellite radiances. This chapter examines the characteristics of these three types of in-situ observations (along with related subtypes such as dropsondes and Mode-S aircraft observations) in terms of measurement systems, errors and biases, data formats and their impact on precision, as well as how these data are processed for use in data assimilation including quality control considerations. The underappreciated need for accurate metadata, including platform characteristics and station location, is also presented. Recent changes in in-situ observations are discussed, with the decrease in aircraft observations associated with COVID-19 demonstrating how these data complement each other and compensate for temporary or even longer term deficiencies.