Status of improving the use of MODIS, AVHRR, and VIIRS polar winds in the GDAS/GFS David Santek, Brett Hoover, Sharon Nebuda, James Jung Cooperative Institute for Meteorological Satellite Studies University of Wisconsin - Madison 11th JCSDA Workshop on Satellite Data Assimilation NCWCP College Park, Maryland 06 June 2013

Outline Polar Winds What happened to Expected Error (EE)? New approach Status of MODIS, AVHRR, and VIIRS

Satellite-derived Polar Winds Unlike geostationary satellites at lower latitudes, it is not be possible to obtain complete polar coverage at a snapshot in time with one or two polar-orbiters. Winds must be derived for areas that are covered by three successive orbits The gray area is the overlap between three orbits. Three overlapping Aqua MODIS passes, with WV and IR winds superimposed. The white wind barbs are above 400 hPa, cyan are 400 to 700 hPa, and yellow are below 700 hPa.

MODIS Polar Winds QC Current Thinning criteria qcU = qcV = 7 ms -1 (O-B) U > qcU OR (O-B) V > qcV Within 50 hPa of the tropopause Within 200 hPa of the surface, if over land Special case qcU = qcV = (ObsSpd + 15)/3 (IR wind within 200 hPa of surface OR WV wind below 400 hPa) AND (GuessSpd +15)/3 < qcU

QC Approach (2012) Expected Error Least square regression is used to compute the RMSE (ms -1 ) from the summed EE components as compared to co- located RAOBs. EE Components: [Terra NH cloud drift] Five QI values [-0.1 to -2.8] Wind speed [+0.1] Wind shear [+0.03] Temperature shear [-0.01] Pressure level [-0.003] Constant [+8.4]

Difference Histogram Accepted Observations (Normalized_EE – Control) More slow winds (5 ms -1) are rejected; more mid- speed winds retained. Few additional winds rejected that deviate 7 ms -1.

Southern Hemisphere 500 hPa ACC: EERAT (red) Control (blue)

What happened to EE? EE only available for GOES and MODIS winds Not for AVHRR nor VIIRS (initially) The EE is dependent on the forecast background More so for MODIS winds than GOES MODIS EE coefficients vary significantly

New Approach One method for screening all polar winds Wind speeds vary over 3 orders of magnitude (1, 10, 100 ms -1 ) Normalize departure by Log of speed Log Normalized Vector Departure (LNVD)

Polar Winds QC Candidate Thinning criteria Discards winds when Log Normalized Vector Departure (LNVD) exceeds a threshold SQRT ( (U o -U b ) 2 + ( V o – V b ) 2 ) / log(ObsSpd) > Threshold Within 50 hPa of the tropopause Within 200 hPa of the surface, if over land

LNVD Threshold Discard winds LNVD > 3 Similar to EE ratio > 1.33 Compared to control: Similar number of vectors discarded Discard more slow winds Retain more high speed winds 9 – 26 October 2012

Log Normalized Vector Departure ObsSpd Log(ObsSpd) VecDif LNVD Threshold = 3

LNVD

Experiments Running current hybrid GDAS/GFS on S4 Early October into December Control Current QC with operational data 2.MODIS LNVD => VecDiff / Log(Obs_spd) < 3 3.AVHRR (NOAA-15, 16, 18, 19, Metop-A); LNVD < 3 4.MODIS IR-only (discard water vapor winds) Preliminary O-B and O-A statistics: 9 – 26 October 2012 Very preliminary forecast impact: October 2012

MODIS OMB and OMA 9 – 26 October 2012 Control LNVD

MODIS: Northern Hemisphere Forecast Impact 500 hPa ACC: LNVD (red) Control (blue)

MODIS: Southern Hemisphere Forecast Impact 500 hPa ACC: LNVD (red) Control (blue)

O-B: AVHRR & MODIS 9 – 26 October 2012

O-A: AVHRR & MODIS 9 – 26 October 2012

MODIS IR-only MODIS winds composed of IR, cloudy water vapor, clear-sky water vapor Current Terra MODIS water vapor channel has only 2 good out of 10 detectors, which impacts: Feature tracking Cloud heights for both water vapor and IR winds CIMSS will discontinue use of Terra MODIS water vapor channel in polar winds production Investigating tracking features from hyperspectral sounder retrievals of water vapor (AIRS, CrIS, IASI, ATMS) [NASA]

VIIRS Polar Winds Spatial: Better resolution than MODIS 750 m across swath; MODIS degrades to 9 km at the edge Wider swath 3000 km vs km Spectral: No water vapor channel on VIIRS, therefore more similar to AVHRR No water vapor winds Limits cloud height options Routine production at STAR: late FY13Q4 Routine production at OSPO: October 2013 Operational: December 2013

Status and Future Plans Completing current experiments on S4 One-year no cost extension Expecting satellite-derived winds from VIIRS (Suomi NPP) fall 2013 NOAA: NA10NES