Estimating GPS radio occultation observation error standard deviations over China using the three‐cornered hat method

Abstract

AbstractObservation error variances are required input parameters for assimilating GPS radio occultation (RO) data in numerical weather prediction systems. A three‐cornered hat method is applied to estimate RO observation error standard deviation (SD) of both refractivity and bending angle. RO data from the Global Navigation Satellite System Receiver for Atmospheric Sounding on board the MetOp‐A/B satellites from 1 January 2016 to 31 August 2019, along with radiosonde observations, the European Centre for Medium‐Range Weather Forecasts Interim reanalysis, and the Global Forecast System forecast analysis, are employed in this study. Taking the Guam radiosonde station as an example, we first introduce a detailed procedure of estimating GPS RO fractional error SDs, which shows the significant impacts of outliers on the resulting error SDs, especially in the lower and middle troposphere. Fractional bending angle error SDs near 48 radiosonde stations in China are then estimated for the same time period of more than 3 years. The RO fractional error SDs near the 48 radiosonde stations have a significant latitudinal dependence, larger at lower than higher latitudes. This characteristic feature of the RO bending angle errors agrees with that of the bending angle uncertainty information provided by the Constellation Observing System for Meteorology, Ionosphere, and Climate Data Analysis and Archival Center.