Assimilation of MWHS on FY-3B over Land Keyi Chen, Niels Bormann, Stephen English, Jiang Zhu

CMA’s FY-3 Series： FY-3A: 05/27/2008-03/2014 FY-3B: 11/05/2010-present FY-3C: 09/23/2013-present

Comparison of Instrument Parameters between MHS and MWHS MWHS Characteristics Comparison of Instrument Parameters between MHS and MWHS Channel number MHS MWHS 1 2 3 4 5 Frequency (GHz) MHS MWHS 89(V) 150(V) 157(V) 150(H) 183.31±1(H) 183.31±1(V) 183.31±3(H) 183.31±3(V) 190.31(V) 183.31± (V) Nadir Res. (km) WF (hPa) MHS MWHS 15 surface 400 600 800

Evaluation and assimilation of MWHS data over sea in the ECMWF system (see also Chen et al. 2015, Weather and Forecasting)

MWHS/FY3A MWHS/FY3B MWHS observation errors MHS observation errors CH3 2.3K 2.5K 2.4K 2K

Forecast Impact of Assimilating MWHS data over sea MWHS/FY3A VS. MWHS/FY3A+B EXP period:2*3months Positive Impacts

Activate the operational use of MWHS/FY-3B over sea at ECMWF on 2014-9-24-00Z MWHS channel 3 operational used data time series MWHS channel 3 operational used data coverage

Assimilation of MWHS data over land

Considerations for adding data over land: Specification of surface emissivity and skin temperature more difficult over land. Two methods are considered here for the surface emissivity: Emissivity atlas : averaged from retrieved emissivities at 89 GHz from different satellites, evolves slowly over time (Kalman filter) Dynamic emissivity: updated instantly by retrieving emissivity from a window-channel observation of a specific instrument Channel number Frequency (GHz) MHS MWHS 1 89(V) 150(V) 2 150(H) 3 183.31+1(V) 4 183.31+3 (V) 5 190.31 (V) 183.31+7GHz (V)

MWHS/FY3B-Oct Tropics: 150GHz less sensitive to surface, emissivity retrieval is not reliable Higher Latitudes: more sensitive; large differences over snow-covered surfaces due to stronger frequency dependence of snow emissivity

Snow-covered surfaces MWHS-CH3-Used Data SkinT>=278K, Orography<=1500m ？ ？ MWHS-CH3/clear data ,SkinT<=278K, Orography<=1500m

MWHS-CH4/clear data 255K=<SkinT<=278K, Orography<=1000m

Experiments set up Control Run: Assimilating MWHS/FY-3B over sea only BasicAtlas EXP: Assimilating MWHS/FY-3B over land by using emissivity atlas without adding data over snow-covered surfaces. SnowAtlas EXP: Assimilating MWHS/FY-3B over land with adding data over snow-covered surfaces by using emissivity atlas. (SkinT >=255K) SnowDynamic EXP: Assimilating MWHS/FY-3B over land with adding data over snow-covered surfaces by using dynamic emissivity retrieved from 150GHz(V) (SkinT >=255K) EXP period:1/1/2015-31/3/2015 + 1/7/2014-30/9/2014

Increase in the number of used data MWHS-CH3 averaged used data coverage by using emissivity atlas without adding data over snow-covered surfaces. Increased data use in snow-covered surfaces in SnowAtlas EXP.

Assimilating Impacts from two season EXP Basicatlas EXP Humidity Channels SnowDynamic EXP SnowAtlas EXP Positive Impacts

Forecast Impact over two seasons Positive Impacts BasicAtlas EXP SnowAtlas EXP EXP period:2*3months SnowDynamic EXP

Results and Future Work: Assimilating MWHS/FY-3B data over land increases the used number of observations, and adding data over snow-covered surfaces with 150GHz(V) dynamic emissivity can further increase the data use. Assimilating MWHS/FY-3B with adding data over snow-covered surfaces improves the fit of ATMS and SSMIS, especially over the Northern Hemisphere. Forecast impacts are mainly neutral when using the emissivity atlas. We do see slightly positive forecast impacts when using the 150GHz(V) dynamic emissivity with adding data over snow-covered surfaces, which suggests that the use of MWHS/FY-3B data could be further extended.

The using data rate change MWHS-CH3 averaged used data coverage difference between that by SnowDynamic EXP and SnowAtlas EXP.