Optimal Assimilation of Microwave Upper-Level Sounding Data in CMA-GFS

Various approaches have been proposed to minimize the upper-level systematic biases in global numerical weather prediction (NWP) models by using satellite upper-air sounding channels as anchors. However, since the China Meteorological Administration Global Forecast System (CMA-GFS) has a model top near 0.1 hPa (60 km), the upper-level temperature bias may exceed 4 K near 1 hPa and further extend to 5 hPa. In this study, channels 12–14 of the Advanced Microwave Sounding Unit A (AMSU-A) onboard five satellites of NOAA and METOP, whose weighting function peaks range from 10 to 2 hPa are all used as anchor observations in CMA-GFS. It is shown that the new “Anchor” approach can effectively reduce the biases near the model top and their downward propagation in three-month assimilation cycles. The bias growth rate of simulated upper-level channel observations is reduced to ±0.001 K d⁻¹, compared to −0.03 K d⁻¹ derived from the current dynamic correction scheme. The relatively stable bias significantly improves the upper-level analysis field and leads to better global medium-range forecasts up to 10 days with significant reductions in the temperature and geopotential forecast error above 10 hPa.