Publishers version - DTU Orbit

sea ice mask was applied and σ0 measurements falling over this mask were not included in
the WVC.
The mission was managed by the Jet Propulsion Laboratory (JPL), operational products
were produced at the National Oceanic Atmospheric Administration (NOAA) for the international
meteorological community and were released near real time (NRT), i.e. within 3 hours
of the data collection. More information can be found in http://winds.jpl.nasa.gov/
missions/quikscat/index.cfm.
The mission requirements were i) an r.m.s. of 2 m s −1 for a wind speed range between
3-20 m s −1 and 10% within the 20-30 m s −1 range and ii) an r.m.s. of 20 ◦ for the direction
within the wind speed range 3-30 m s −1 . Science products were distributed through the
Physical Oceanography Data Archive Center (PODAAC). More information on the science
data products can be found in JPL (2006).
Different scientific groups applied different GMFs and methodologiesfor the wind retrieval,
resulting in a variety of different products. Some of these products are still being reprocessed
when new, improved GMFs are released. The initial GMF was the NSCAT-2 but from 2000 to
2006JPL derived wind velocityfrom QuikSCAT using the QSCAT-1 GMF (PO-DAAC, 2001),
developed during the calibration/validation phase. From 06/2006 and afterwards, JPL used
the QSCAT-1/F13, recalibrated for wind speeds above 16 m s −1 using speeds from SSM/I
13.
The Royal Netherlands Meteorological Institute (KNMI) has processed data using the
NSCAT-2 GMF for the Ocean & Sea Ice Satellite Application Facility (OSISAF). They also
applied a rather different methodology for the ambiguity removal and selection of the “best”
solution for the WVC. Remote Sensing Systems (RSS) has been releasing swath and gridded
QuikSCAT data using a family of descendants from the NSCAT-2 GMF, named Ku-2000,Ku-
2001and recently Ku-2011.For details, see http://www.ssmi.com/qscat/qscat_browse.
html. In addition to the different GMF used to derive winds, various products exist ranging
from the σ0 observations time-ordered and earth-located (L1), ocean wind vectors in a swath
grid (L2B) and gridded ocean wind vectors (L3).
Many validationstudies exist that compare the QuikSCAT winds with in situ measurements
from buoys and research ships and investigate potential biases. Pickettet al. (2003)compared
QuikSCAT swath and gridded winds with near-shore and offshore buoy data and found a
maximum wind speed bias of 0.5 m s −1 with a root mean square (r.m.s.) error of 1.6 m s −1 .
The maximum mean bias, for the direction, was 11 ◦ with 26 ◦ r.m.s. error, while r.m.s. errors
up to 38 ◦ were also found.
The quality of QuikSCAT retrievals using different algorithms (JPL L2B, DIRTH, RSS v2)
has been discussed in Ebuchi et al. (2002). They used offshore buoy data and found wind
speed r.m.s. differences of 1 m s −1 and ∼20 ◦ for the direction, when wind speeds higher
than 3 m s −1 were used. Bourassa et al. (2003) evaluated the Ku-2000 QuikSCAT from RSS
against winds from research vessels and found a maximum wind speed bias of 0.7 m s −1 and
10 ◦ for direction.
Especiallyforgridded(L3)products,Satheesanetal.(2007)comparedthegriddedQuikSCAT
product from RSS with buoy data from the Indian Ocean; they reported a wind speed bias
of 0.37 m s −1 and r.m.s. error of 1.57 m s −1 when all wind speeds were used. For the wind
direction, the bias was 5.81 ◦ and the r.m.s. error was 44.1 ◦ . Pensieri et al. (2010) evaluated
QuikSCAT L3 from JPL against buoy data in the Lingurian Sea, reporting maximum mean
bias of 1.09 m s −1 and r.m.s. error of 1.97 m s −1 for wind speed and bias of –8.7 ◦ and r.m.s.
error of 79.5 ◦ in direction.
InparticularfortheNorthernEuropeanSeas,Winterfeldtetal.(2010)comparedQuikSCAT
L2B products with buoy wind speeds in the eastern North Atlantic Ocean and the North Sea,
addressing the sensitivity of biases to collocation criteria and algorithms for converting in situ
measurementstoENW. Average windspeed biasesrangedfrom 0.18to0.30ms −1 depending
on collocation criteria and from 0.1 to 1.6 m s −1 depending on the conversion algorithm.
Winterfeldt (2008) compared a level2B 12.5 km QuikSCAT product with in-situ observations
from buoys and rigs in the Eastern North Atlantic and the North Sea during one year
DTU Wind Energy-E-Report-0029(EN) 301