Use of AMVs in the JMA model

Physical characteristics

Global Model

• Spectral model
• Horizontal resolution:
  • Model: TQ959 (∼0.1250° (13 km), Reduced Gaussian Grid).
  • Analysis: Outer loop is TQ959, inner loop is TL319 (∼0.5625° (55 km), Reduced Gaussian Grid).
• Vertical resolution: Surface + 128 levels up to 0.01 hPa.
• Analysis times (T): 00, 06, 12, 18 Z

Data assimilation method

Global Model

• Hybrid LETKF/4D-Var, 6 hr time window
• Time window: T ± 3 hr
• Time constraints (model runtime):
  • Early analysis: 40 min before time window ends
  • Cycle analysis: 8 hr 50 min after time window ends (T=00,12Z) and 4 hr 50 min after time window ends (T=06,18Z)

AMV types assimilated

Global Model

• Meteosat-9,10 BUFR IR10.8, VIS0.8, cloudy WV(6.2, 7.3)
• Himawari-9 BUFR IR(3.9, 10.4), VIS(0.64), cloudy WV(6.2, 6.9, 7.3)
• GOES-16,18 BUFR IR(3.9, 11.2), VIS0.64, cloudy/clear sky WV(6.2, 6.9, 7.3)
• CIMSS MODIS Terra A/N IR, cloudy WV
• CIMSS MODIS Aqua A/N IR, cloudy WV
• NESDIS and direct broadcast CIMSS MODIS Aqua IR, cloudy WV, clear sky WV
• CIMSS and NESDIS AVHRR NOAA-15
• CIMSS AVHRR NOAA-18,-19, Metop-B
• NESDIS VIIRS Suomi-NPP,NOAA-20
• CIMSS LEOGEO
• EUMETSAT AVHRR Dual-Metop

Global quality control

Blacklisting in space

• MODIS IR winds for NH above 300 hPa or below 900 hPa
• MODIS WV and CSWV winds for NH above 300 hPa or below 550 hPa
• MODIS IR and WV winds for SH above 300 hPa or below 550 hPa
• MODIS CSWV winds for SH above 350 hPa or below 550 hPa
• MODIS IR and WV winds for SH above 300 hPa or below 550 hPa
• MODIS CSWV winds for SH above 350 hPa or below 550 hPa
• ALL AVHRR and VIIRS winds below 700 hPa
• NESDIS AVHRR winds above 400 hPa
• VIIRS winds above 300 hPa
• VIIRS winds over land below 400 hPa
• LEOGEO winds above 300 hPa or below 900 hPa
• LEOGEO winds between 600 hPa and 640 hPa
• LEOGEO winds at the extratropics poleward of 70N or 70S
• LEOGEO winds at the extratropics poleward of 60S between 640 hPa and 900 hPa
• Dual-Metop winds between 50S and 50N
• Dual-Metop winds above 300 hPa
• Dual-Metop winds over land below 700 hPa
• All polar winds at the extratropics poleward of 88N or 88S
• All geostationary winds above 175 hPa or below 975 hPa except Himawari-9 winds
• All geostationary winds between 400 hPa and 825 hPa except Himawari-9 winds
• All geostationary IR winds above 275 hPa at the extratropics poleward of 20N or 20S except Himawari-9 winds
• All geostationary WV winds above 225 hPa at the extratropics poleward of 20N or 20S except Himawari-9 winds
• Himawari-9 IR and VIS winds over land below 700 hPa
• Meteosat-9 IR and VIS winds over land below 700 hPa

Blacklisting by wind speed

• Japan and surrounding area (Area of super-observation processing) wind speed 15 m/s and over 15 m/s and large u-component wind speed compared with v-component wind speed and O-B over 3 m/s or under -2 m/s
• Reject for O-B vector difference > 14.1 m/s for GOES winds

QI thresholds

tropics: 20S-20N, extratropics: polewards of 20S/N
HL: 10-400 hPa, ML: 400-700 hPa, LL: 700-1000 hPa
NH: Northern Hemisphere, SH: Southern Hemisphere

QI1 – EUMETSAT QI with first guess check
QI2 – EUMETSAT QI without first guess check
RFF – Recursive Filter Function (CIMSS/NESDIS)

For information on how the quality indicators are formulated
see Holmlund (1998, Weather Forecasting 13 1093-1104) and Hayden and Purser (1995, Journal of Applied Meteorology 34 3-15).

Thinning

• All geostationary winds thinned in 2° by 2° by 100 hPa boxes. A minimum horizontal distance is 200 km.
• Wind selected by highest QI and proximity to analysis time and centre of thinning boxes for all geostationary winds.
• All polar winds thinned in 1.5° by 1.5° by 100 hPa boxes.
• Wind selected by proximity to analysis time and centre of thinning boxes for all polar winds.
• Only one wind selected per box in the 6 hour time window for all winds.

• The super-observation data are produced by averaging hourly Himawari-8 winds in 1° by 1° by 100 hPa boxes.
• The data are only produced over Japan and the surrounding areas (20N-45N, 120E-150E).

Background check

• Pass for |O-B| ≤ a + 1.5 * σ
• Reject for |O-B| > a + 3 * σ
• Between these criteria, the data undergoes consistency check.

where “a” is predefined criteria calculated from averaged O-B and σ is predefined observation error, dependent on wind level.
O: observation, B: first guess

Observation errors