Impacts of AMSU-A, MHS and IASI data assimilation on temperature and humidity forecasts with GSI–WRF over the western United States

Y Bao,M Shao,Y Pan,J Min,J Xu,A M Powell Jr

doi:10.5194/amt-8-4231-2015

Abstract

Abstract. Using NOAA's Gridpoint Statistical Interpolation (GSI) data assimilation system and NCAR's Advanced Research WRF (Weather Research and Forecasting) (ARW-WRF) regional model, six experiments are designed by (1) a control experiment (CTRL) and five data assimilation (DA) experiments with different data sets, including (2) conventional data only (CON); (3) microwave data (AMSU-A + MHS) only (MW); (4) infrared data (IASI) only (IR); (5) a combination of microwave and infrared data (MWIR); and (6) a combination of conventional, microwave and infrared observation data (ALL). One-month experiments in July 2012 and the impacts of the DA on temperature and moisture forecasts at the surface and four vertical layers over the western United States have been investigated. The four layers include lower troposphere (LT) from 800 to 1000 hPa, middle troposphere (MT) from 400 to 800 hPa, upper troposphere (UT) from 200 to 400 hPa, and lower stratosphere (LS) from 50 to 200 hPa. The results show that the regional GSI–WRF system is underestimating the observed temperature in the LT and overestimating in the UT and LS. The MW DA reduced the forecast bias from the MT to the LS within 30 h forecasts, and the CON DA kept a smaller forecast bias in the LT for 2-day forecasts. The largest root mean square error (RMSE) is observed in the LT and at the surface (SFC). Compared to the CTRL, the MW DA produced the most positive contribution in the UT and LS, and the CON DA mainly improved the temperature forecasts at the SFC. However, the IR DA gave a negative contribution in the LT. Most of the observed humidity in the different vertical layers is overestimated in the humidity forecasts except in the UT. The smallest bias in the humidity forecast occurred at the SFC and in the UT. The DA experiments apparently reduced the bias from the LT to UT, especially for the IR DA experiment, but the RMSEs are not reduced in the humidity forecasts. Compared to the CTRL, the IR DA experiment has a larger RMSE in the moisture forecast, although the smallest bias is found in the LT and MT.

Highlights

Instead of the random distribution and heterogeneous spatial density in the traditional conventional radiosondes, satellite observations provide a large amount of data covering worldwide areas in order to improve the initialization of the weather forecast models through a data assimilation system
Compared to the control experiment (CTRL) experiment, the smaller root mean square error (RMSE) are only found in the MW experiment in the upper troposphere (UT) and lower stratosphere (LS), and the conventional data only (CON) data assimilation (DA) gave a positive contribution at the SFC and in the UT
The RMSE is not always consistent with the bias profile in the temperature forecasts: the RMSE profile shows that the largest RMSE appeared in the lower troposphere (LT) and the smallest error in the middle troposphere (MT)

Summary

Introduction

Instead of the random distribution and heterogeneous spatial density in the traditional conventional radiosondes, satellite observations provide a large amount of data covering worldwide areas in order to improve the initialization of the weather forecast models through a data assimilation system. Many studies have demonstrated that the assimilation of satellite data has significantly improved weather forecasts 1998; Zhou et al, 2011), especially over some areas with sparse conventional observations (McNally et al., 2000; Zapotocny et al, 2008; Liu et al, 2012). Range Weather Forecasts (ECMWF) system, Andersson et al (1991) pointed out that the forecast shows a negative impact of the satellite sounding data in the Northern Hemisphere, and a strong positive impact in the Southern Hemisphere.

Objectives

Methods

Results

Conclusion