Comparison of ERS-2 and Quikscat Winds

Comparison of ERS-2 and QuikSCAT winds

Abderrahim Bentamy1, Semyon A. Grodsky2, and ???

August 23, 2012

1 Institut Francais pour la Recherche et l’Exploitation de la Mer, Plouzane, France

2Department of Atmospheric and Oceanic Science, University of Maryland, College Park, MD 20742, USA

Corresponding author: [email protected]

Abstract

Global winds from two scatterometers, ERS-2 (1996-2001) and QuikSCAT (1999-

2009) show persistent differences during their period of overlap (July-1999 to January 2001).

This study examines a set of collocated observations during this period to evaluate the causes of these differences. A difference in the operating frequency of the scatterometers leads to differences depending on rain rate, wind velocity, and SST. The impact of rainfall on the higher frequency QuikSCAT is mitigated by combined use of the standard rain flag and removing data for which the multidimensional rain probability >0.05. Global ERS-2 wind speed based on the CMODIFR2 Geophysical Model Function (GMF) is underestimated in comparison with QuikSCAT by 0.6 m/s, but wind directions are consistent. The wind speed bias is reduced to -0.2m/s after partial reprocessing of ERS-2 wind speed using the new

CMOD5.n GMF, but keeping the wind direction unchanged. An additional component of the difference in wind speed results from biases in GMFs used in processing the two data sets and is parameterized here as a function of ERS-2 wind speed and direction relative to the mid- beam azimuth. After applying the above corrections, QuikSCAT wind speed remains systematically lower (by 0.5 ms-1) than ERS-2 over regions of cold SST<10oC. This difference results from temperature-dependence in the viscous damping of surface waves which has a stronger impact on shorter waves and thus preferentially affects QuikSCAT.

1 1. Introduction

Reliable spatial and temporal coverage of winds is provided only by satellite sensors, which are limited in their life span and different in their technical parameters (Bourassa et al.,

2009). Many studies require consistent time series of winds spanning the lifetime of multiple satellite missions. Creating consistent time series requires accounting for changes in individual mission biases, most strikingly when successive scatterometers operate at different frequencies (e.g. Bentamy et al., 2002; Ebuchi et al., 2002; Bentamy et al., 2012). One of the longest scatterometer missions QuikSCAT (mid 1999 to late 2009, Ref) overlaps in time with the following ASCAT (2007-onward) and the preceding ERS/2 (1996-January 20011). This opens a possibility to adjust and combine these data.

Scatterometers are microwave radars that infer wind velocity from the normalized radar backscatter coefficients (NRBC,  0 ) measured at different azimuth and incidence angles ( ).

The NRBC depends on the small capillary/gravity sea waves, which are in local equilibrium with instantaneous near surface wind. The near surface wind is quantified in terms of the 10m neutral wind (W ). It is inferred using an empirical GMF, which relates W and wind direction relative to the radar azimuth (  ) to measured  0 (Ref).

Since 1991 a total of eight scatterometers onboard of various satellites have been operational (Bourassa et al, 2009). To date, however, a combined use of their data is limited due to the inter-instrument biases (e.g. Bentamy et al, 2002; Ebuchi et al, 2002, Bentamy et al,

2012). Numerous efforts to combined multi-satellite winds generally utilize a reference product such as passive microwave winds or reanalysis winds to which the individual scatterometer data sets are calibrated (e.g. Bentamy et al, 2007; Atlas et al, 2011). In contrast to such previous analyses, this current study focuses on directly matching scatterometer data.

1 ERS-2 didn’t provide global winds after January 2001 due to a failure of onboard data recorder.

2 We believe that removing inter-instrument bias prior to the analysis of combined data would facilitate analyses and improve accuracy of individual data sets.

Historically, the scatterometers developed by the European Space Agency (ESA) use the C-band (5.3 GHz), while NASA scatterometers use the higher frequency Ku-band

(14.1GHz). Since 1991 there have been four long-lived missions. European Remote Sensing

Satellites (ERS-1 and ERS-2) were equipped with C-band scatterometers and provided global data from July 1991 through January 2001. The NASA QuikSCAT, a Ku-band scatterometer, provided global data during July 1999 - November 2009. The latest Advanced

SCATterometer (ASCAT) onboard European Meteorological Satellite (Metop-A) uses C- band and has been operating since October 2006.

This paper follows the Bentamy et al. (2012) study of overlapping ASCAT and

QuikSCAT winds and applies the similar methodology to overlapping winds from the C-band

ERS-2 and the Ku-band QuikSCAT to quantify and correct for the systematic biases between the two. The results are expected to support the compilation of a combined scatterometer winds including ERS-2, QuikSCAT, and ASCAT.

2. Data

In this section we provide a brief description of data. More details are available in ERS and QuikSCAT user manuals (CERSAT, 1994; and JPL, 2006).

2.1 ERS-2

The active microwave instrument (AMI) on board ERS-2 is the same 5.3 GHz (C-band;

5.7 cm) scatterometer as onboard ERS-1. It operated from April 21, 1995 through September

5, 2011. However, due to the on-board recorder failure, global data are available only through early January 2001. The scatterometer has three antennae looking 45° forward (for-beam), perpendicular (mid-beam), and 45° backward (aft-beam) relative to the satellite track and illuminating a 500km wide swath right of the track. The 10 m neutral wind speed and

3 direction are inferred at 50km spatial resolution using the CERSAT algorithm (Quifen et al,

1995) based on the IFREMER version 2 GMF (CMODIFR2 of Bentamy et al., 1999).

CMODIFR2 has been derived by fitting collocated ERS-1/NDBC buoy data and applied to the following ERS-2 without any adjustments. Land, ice, and rain contaminations are excluded using the CERSAT quality flags. Although these ERS-2 winds are known for persistent wind speed underestimation and a lack of low wind data (Bentamy et al., 2002), this dataset is the only one spanning the entire mission in the global mode.

2.2 QuikSCAT

The SeaWinds Ku-band (13.4 GHz, 2.2 cm) scatterometer onboard the

NASA/QuikSCAT (referred to as QuikSCAT or QS) was launched in June 1999. QuikSCAT rotating antenna has two emitters: the H-pol inner beam at =46.25° and V-pol outer beam at

=54° with a swath of 1400km and 1800km, respectively that covers around 90% of the global ocean daily. QuikSCAT swath is binned into Wind Vector Cells (WVC) of 25´25 km.

QuikSCAT winds used in this research are Level 2b data estimated with the empirical

QSCAT-1 GMF (JPL, 2006) and the Maximum Likelihood Estimator, which selects the most probable wind solution. To improve wind direction in the middle of swath where the azimuth diversity is poor, the Direction Interval Retrieval with Threshold Nudging algorithm is applied. This retrieval technique provides approximately 1 m/s and 20deg accuracy in wind speed and direction, respectively (e.g. Bentamy et al., 2002, Bourassa et al., 2003, Ebuchi et al., 2002).

Due to shorter wavelength Ku-band is more sensitive to impacts of rain than longer wavelength C-band. Sobieski et al. (1999) have shown that rain may decrease the atmospheric transparency and thus result in wind speed underestimation. An opposite effect develops in response to the surface ripple generation by raindrops and scattering on raindrops that both lead to overestimation of winds (Weissman et al., 2002). Two rain indices, rain flag and

4 multidimensional rain probability (MRP), are provided with QuikSCAT data to mark heavy rainfall.

2.3 Collocated data

The ERS-2/QuikSCAT collocation procedure is similar to that used in (Bentamy et al.,

2012). Temporal span of the collocation extends from July 1999 to January 2001, the period when both ERS-2 and QuikSCAT provided global ocean coverage. The spatial separation between collocated ERS-2 and QuikSCAT cells is <50km. The two satellites are on quasi sun-synchronous orbits, but QuikSCAT local equator crossing time at the ascending node

(6:30 a.m.) leads by approximately 4 hours the ERS-2 local equator crossing time (10:30 a.m.). This difference implies that collocations from the two instruments are available only with a few hour lag at low latitudes. For temporal separation  <5 hours the coverage is global. There are >36 millions of collocations, but the data coverage is higher at higher latitudes due to the polar convergence of orbits (Bentamy et al., 2012).

For ground truth comparisons, both ERS-2 and QuikSCAT are collocated with moored buoys from the National Data Buoy Center (NDBC), the Tropical Atmosphere Ocean Project

(TAO), and the Pilot Research Moored Array in the Atlantic (PIRATA). Hereafter, the combined TAO and PIRATA data are indicated as Tropical buoys. Details of buoy instrumentation are provided in Meindl et al., (1992); McPhaden et al., (1998), and Bourles et al., (2008). Hourly averaged buoy wind velocity, SST, air temperature, and humidity are converted to the 10m neutral wind using the COARE3.0 algorithm of Fairall et al. (2003).

ERS-2 and QuikSCAT are separately collocated with buoy hourly winds. Satellite/buoy collocation criteria are set at 50km and 1hour for ERS-2 and 25km and 30min for QuikSCAT.

3. ERS-2 wind accuracy

ERS-2 wind speed based on the CMODIFR2 GMF is biased low in comparison with buoy winds (Fig. 1a) as has been previously found in Bentamy et al. (2002). This

5 underestimation is apparent for the most frequently observed winds <13m/s. At higher winds the satellite-buoy wind speed difference is negative, but its estimate is subject to uncertainties in this wind speed range due to relatively rare occurrences. In spite of systematic wind speed underestimation, the satellite-derived wind direction is consistent with in-situ data to within

10o (Fig. 1b).

Table 1 presents statistical indices of the satellite-buoy comparisons, which include only buoy and satellite data with valid quality control flags. In particular, QuikSCAT rain selection employs the rain flag and MRP<0.05, as explained in Bentamy et al. (2012). In comparison with buoy data, the ERS-2 wind speed is biased low by 0.6m/s while QuikSCAT bias is negligible. Wind direction from both scatterometers compares well with buoy wind direction

(see also Fig.1b). Statistical comparisons of buoy-satellite winds based on the entire period for each mission (March 1996 – January 2001 for ERS-2, and July 1999 – November 2009 for

QuikSCAT) are in line with those based on the shorter period (July 1999 – January 2001) common for the two missions. This agreement illustrates the representativeness of the common period, which is used for collocated data.

ERS-2 wind speed underestimation seen locally at the midlatitude NDBC buoys (Fig.

1a) manifests also in global satellite collocated data (Fig. 2a). But, like in NDBC buoy comparisons, the wind direction from the two missions is consistent (Fig. 2b). Time mean

QuikSCAT wind speed is higher than ERS-2 wind speed almost everywhere (Fig. 3a) except at high latitudes where the difference in wind speed is lower. At those high latitudes the

QuikSCAT winds are also biased low (that explains the weak difference) due to stronger viscous dissipation of the Bragg waves at cold SST (Bentamy et al., 2012; Grodsky et al.,

2012). In spite of the ocean wide pattern of wind speed bias (Fig. 3a), the variability of ERS-2 and QuikSCAT winds are consistent. This reflects in high temporal correlation of the two

(Fig. 3c) that exceeds 0.8 at most locations except the low latitudes. Lack of correlation in the

6 deep tropics is explained by convective variability, which may be significant on time scales comparable to the temporal lag between the two satellites, and by weak data coverage, which is limited to collocations separated by 4hr or more. The standard deviation between collocated winds (Fig. 3b) apparently increases in the mid-latitude storm track bands due to unresolved synoptic variability.

ERS-2 wind bias may have at least two sources due to uncertainties in (i) backscatter coefficient calibration and (ii) GMF parameterization. To the best of our knowledge, only a

0.165 dB bias in the calibrated backscatter coefficients has been reported and corrected by

Crapolicchio et al. (2007). We shall further discuss (i) in the Discussion section. Impact of

GMF uncertainty is expected because ERS-2 global winds based on the CMODIFR2

(Bentamy et al., 1999) employ GMF fitted to ERS-1, but applied to ERS-2 without any adjustments.

Since original processing of the ERS-2 global winds by IFREMER, a number of C-band

GMFs fitted specifically for ERS-2 backscatter have been developed. The latest, CMOD5.n has been derived by Hersbach et al. (2007) using collocated ERS-2  o triplets and ECMWF short-range forecast winds. A simple method to reduce the wind speed bias in the original

ERS-2 winds is to apply CMOD5.n assuming that wind direction based on the CMODIFR2 is bias-free. This latter assumption is supported by Figs. 1b, 2b, and Table 1. It significantly simplifies and speeds up wind speed calculations. The resulting wind speed is referred to as the new ERS, or ERS/N. It is determined by minimizing the cost function (Quilfen, 1995):

3 0 0 2 J (W , )  [ i  i CMOD5.n (W , )] , (1) i1

where W is the new wind speed,  is the wind direction relative to antenna azimuth

0 0 (known from CMOD2IFR product),  and  CMOD5.n are observed and simulated backscatter

7 coefficients. At each ERS-2 WVC, ERS-2 wind speed based on CMOD2IFR is used as the first guess for minimization of (1).

The wind speed bias is significantly reduced for the new ERS/N winds. This reduction is seen in apparently smaller difference from the NDBC wind speed, which is less than 0.1m/s

(Table1). For the ERS/N winds the global ‘red’ pattern in the wind speed difference is gone

(compare Figs. 4a and 4b). The global bias is reduced to -0.2m/s (Fig. 5b). The negative bias is stronger over cold SST (Fig. 4b) due to the difference in scatterometer measurement physics (Grodsky et al, 2012). High discrepancies are present along the north Atlantic and

Pacific westerly storm tracks, which may be related to the sampling of synoptic events.

Deviations are also noticeable in the coastal areas where the diurnal cycle of breezes may project on the time lag of collocated data (Bentamy et al., 2012).

4. Adjusting ERS/N and QuikSCAT winds

Within the global bias between ERS/N and QuikSCAT wind speed, which is -0.2m/s

(Fig. 5b), there are biases due to inconsistencies in the retrieval procedures (GMF related bias) and due to the difference in the measurement physics. The impact of rainfall on the

-1 higher frequency QuikSCAT introduces positive biases in WQS of up to 1 ms in the tropical convergence zones and along the western boundary currents even after rain flagging is applied

(Bentamy et al., 2012). This difference is reduced by some 30% to 40% when data for which

MRP >0.05 are also removed. This combination of rain selection indices is applied to all

QuikSCAT data in this study.

4.1 GMF related bias

A difference in measuring geometry and retrieval procedures leads to the difference in wind speed (W ) due to biases in GMFs used in processing the two data sets. Following

Bentamy et al. (2012) the GMF-related correction ( W1) is parameterized as a function of

8 ERS/N wind speed and direction relative to the mid-beam azimuth. The CMOD5.n GMF is parameterized by a truncated Fourier series of wind direction relative to antenna azimuth,  , with coefficients depending on wind speed and incident angle,  . This suggests that GMF- related correction function W1 (to be added to ERS/N winds to adjust them to QuikSCAT

winds) depends on W1(WERS / N , , ) . Due to the fixed, three beam observation geometry of

ERS-2, the azimuth dependence is reduced to dependence on ERS-2 wind direction relative to the mid-beam azimuth. Like in Bentamy et al. (2012) comparison of ASCAT and QuikSCAT winds, there is only a minor dependence of W1 on  (not shown).

Empirical estimate of the GMF-related correction W1 is constructed by binning

2 observed WQS WERS / N as a function of WERS / N and  . It is based on data from latitudes equatorward of 50o where the negative bias (SST-related bias considered later) is not

-1 dominant (Figs. 4b, 5a). Binned W1 has apparent positive values at WERS / N <5ms (not shown), which are artificial and result from asymmetrical distribution of wind speed sampling for winds approaching the low wind speed cutoff (Freilich, 1997). These positive values at

4 low winds are corrected as follows, W1 tanh[(WERS / N / 5) ] . To mitigate for the sampling errors, we use bins with at least 50 samples only, smooth W1 by the triangular 3x3 spatial filter, and retain only the first 5 angular harmonics (Fig. 6a). ERS/N wind speed is lower than

-1 QuikSCAT wind speed for WERS / N >15ms in the up- and down-wind direction (Fig. 6a), but the difference is opposite in the cross-wind direction. The resulting W1 is not symmetric in the azimuth direction, which is expected based on the azimuth symmetry of the CMOD5.n.

This azimuth asymmetry is suggestive of inconsistency in antenna calibration of the for- and after-beams (discussed later).

2 We found that adding incidence angle doesn’t improve the difference in scatterometer winds.

9 The time mean spatial pattern of W1 depends on the distribution of local wind speed

and direction. Adding the W1 correction to ERS/N wind speed, WERS / N  W1, results in slight strengthening of the trade winds and weakening of the midlatitude westerlies (Fig. 7a).

This correction reduces the global wind speed bias from -0.2m/s to -0.1m/s and improves the consistency of the corrected ERS/N and QuikSCAT winds at high latitudes (Figs. 4b, 4c, 5a).

4. 2 SST-related bias

After applying the GMF-related correction W1, QuikSCAT wind speed remains systematically lower (by 0.5 ms-1, Fig.4c) than corrected ERS/N wind speed over regions of cold SST<10oC. SST can alter the growth rate of centimeter-scale waves through its impact on air and density and water viscosity. But, this SST-dependency has not been included in the standard GMFs. Modeling of the SST-related bias suggests that it is weak in the C-band and has a greater impact on shorter waves and thus preferentially impacts QuikSCAT (Grodsky et al., 2012). For QuikSCAT the major impact is due to the viscous effect that results in wind underestimation at low SST.

Here we use the Bentamy et al. (2012) estimate of the SST-related bias ( W 2 , Fig. 6b)

and subtract it from the QuikSCAT wind speed, WQS  W 2 . This correction increases WQS over regions of cold SST<10oC (Fig. 7b), eliminates much of the negative wind speed bias between the corrected ERS/N and QuikSCAT winds (compare Figs. 4c and 4d), and results in very low global mean bias between the two (Fig. 5b).

5. Discussion

Bentamy et al. (2012) have shown that the overestimation of C-band scatterometer winds for crosswind directions is related to the accuracy of CMOD5.n. However, the

difference pattern (WQS WERS / N , Fig. 6a) is not symmetric as it is expected. ERS/N wind

speed overestimation (WQS WERS / N <0) is more pronounced, up to 1m/s, for the wind

10 direction of -90° (clockwise from the mid-beam) than that for +90° where the difference is quite low. Similar angular behavior is found for NDBC buoy minus ERS/N wind speed binned as a function of wind direction (not shown). Although the asymmetry explanation is still not clear, it may be a consequence of inconsistency in the ERS-2 beam inter-calibration.

0 0 Next we compare observed ( ) and simulated ( CMOD5.n ) NRBCs for each ERS-2 beam.

0 Simulated  CMOD5.n is based on CMOD5.n forced by the corrected collocated QuikSCAT

wind speed (WQS  W 2 ) and direction. The difference between observed and simulated

NRBCs for outer beams (Fig. 8) may lead to the azimuth asymmetry seen in Fig. 6a. For aft-

0 0 beam and for-beam the same   CMOD5.n is expected. Indeed, they have same incidence angles, and differences are evaluated for same surface wind using the same GMF. These results agree with De Charia et al (2009) and suggest the need for complete reprocessing of

ERS-2 scatterometer winds.

6. Conclusion

Extending consistent scatterometer time series into pre-QuikSCAT period (before July

1999) requires the compatibility of the Ku-band QuikSCAT and the previous C-band ERS-2.

Examination of collocated observations during the 18 months of missions overlap July 1999- eraly January 2001 reveals systematic differences in the 10m neutral winds from the two. This study is an attempt to identify and model these differences in order to produce a consistent scatterometer-based wind record spanning the aggregate ERS-2 and QuikSCAT period.

The basic data set for identifying the differences and developing corrections is the set of space-time (less than 5hr and 50km) collocated winds after applying all quality control flags.

The impact of rainfall on the higher frequency QuikSCAT is reduced by applying a combination of the standard rain flag along with removing data for which the multidimensional rain probability >0.05. Due to the 4 hour difference in the equator crossing

11 time, requiring shorter time lags for collocations severely limits the data coverage at lower latitudes.

Global ERS-2 winds spanning the entire mission are only available based on the

CMODIFR2 Geophysical Model Function (GMF). This wind speed is underestimated in comparison with QuikSCAT by 0.6 m/s, but wind directions are consistent. This global wind speed bias is a consequence of applying the CMODIFR2 to the ERS-2. This GMF was derived for ERS-1 but applied to the next ERS-2 without any adjustments. The wind speed bias is reduced to -0.2m/s after partial reprocessing of the CMODIFR2-based wind speed using the new CMOD5.n GMF, but keeping the wind direction unchanged. An additional component of the difference in wind speed results from biases in GMFs used in processing the two data sets and is parameterized as a function of ERS-2 wind speed and direction relative to the mid-beam azimuth. After applying the GMF-related correction to ERS-2 winds, the global wind speed difference between ERS-2 and QuikSCAT decreases to -0.1m/s. But, QuikSCAT wind speed remains systematically lower (by 0.5 ms-1) than ERS-2 over regions of cold

SST<10oC. This difference results from temperature-dependence in the viscous damping of surface waves which has a greater impact on shorter waves and thus preferentially impacts

QuikSCAT. After applying the SST-related correction to the QuikSCAT wind speed, the global mean wind speed difference between ERS-2 and QuikSCAT becomes negligible.

Analysis of the QuikSCAT minus ERS-2 wind speed difference reveals significant asymmetry versus the wind direction relative to the ERS-2 mid-beam azimuth. This azimuth dependence can’t be explained by errors in GMF used for ERS-2 processing because any

GMF is symmetric in the relative wind direction. Closer examination of the backscatter coefficients for ERS-2 beams reveals inconsistency between the for-beam and aft-beam, which could be responsible for the asymmetry. This finding along with an apparent wind

12 speed bias in CMODIFR2-based product suggests the need for complete reevaluation and reprocessing of ERS-2 scatterometer data.

Acknowledgements. This research was supported by the NASA International Ocean

Vector Wind Science Team (NASA NNXIOAD99G).

13 References

Atlas, R., R. N. Hoffman, J. Ardizzone, S. M. Leidner, J. C. Jusem, D. K. Smith, and D.

Gombos (2011), A Cross-calibrated, Multiplatform Ocean Surface Wind Velocity Product

for Meteorological and Oceanographic Applications. Bull. Amer. Meteor. Soc., 92, 157–

174. doi: 10.1175/2010BAMS2946.1

Bentamy, A., P. Queffeulou,, Y. Quilfen, K. Katsaros,1999: Ocean surface wind fields

estimated from satellite active and passive microwave instruments, IEEE T. Geoscience

and Remote Sensing, 37 (5) , 2469-2486

Bentamy A., K B. Katsaros, M. Alberto, W. M. Drennan, E. B. Forde, 2002: Daily surface

wind fields produced by merged satellite data. American Geophys. Union, Geophysical

Monograph Series Vol. 127, 343-349.

Bentamy, A., H.-L. Ayina, P. Queffeulou, and D. Croize-Fillon (2007), Improved Near Real

Time Surface Wind Resolution over The Mediterranean Sea, Ocean Sci., 3, 259-271.

Bentamy, A., S. A. Grodsky, J. A. Carton, D. Croizé-Fillon, and B. Chapron, 2012: Matching

ASCAT and QuikSCAT winds, J. Geoph. Res., 117, C02011,

doi:10.1029/2011JC007479.

Bourassa, M. A., D. M. Legler, J. J. O’Brien, and S. R. Smith, 2003: SeaWinds validation

with research vessels, J. Geophys Res., 108, C2, 3019.

Bourassa M., A., and Coauthors, 2009: Proceedings of OceanObs’09: Sustained Ocean

Observations and Information for Society. 21-25 Venice Italy.

(http://www.oceanobs09.net/proceedings/cwp/Bourassa-OceanObs09.cwp.08.pdf )

Bourles, B., et al. (2008), The Pirata Program: History, accomplishments, and future

directions, Bull. Am. Meteorol. Soc., 89, 1111– 1125.

14 Center for Satellite Exploitation and Research (CERSAT), 1994: WNF products – User

manual, IFREMER/CERSAT, C2-MUT-W-01-IF, pp. 86, available at (

http://cersat.ifremer.fr/content/download/2414/16735/file/mutw01.pdf ).

Crapolicchio, R., G. De Chiara A. Paciucci, and P. Lecomte, 2007: The ERS-2 scatterometer:

instrument and data performances assessment since the beginning of the mission. Proc.

‘Envisat Symposium 2007’, Montreux, Switzerland 23–27 April 2007 (ESA SP-636,

July 2007).

De Chiara, G., and H. Hersbach, 2009: ERS-2 scatterometer cycle report. ESA publication.

http://earth.esrin.esa.it/pcs/ers/scatt/reports/pcs_cyclic/wscatt_rep_144.pdf. pp. 63.

Ebuchi, N., H. C. Graber, and M. J. Caruso, 2002: Evaluation of wind vectors observed by

QuikSCAT/SeaWinds using ocean buoy data. J. Atmos. Oceanic Technol., 19, 2049-

Fairall, C.W., E.F. Bradley, J.E. Hare, A.A. Grachev, and J.B. Edson (2003), Bulk

Parameterization of Air–Sea Fluxes: Updates and Verification for the COARE Algorithm,

J. Clim., 16, 571–591.

Freilich, M.H. (1997), Validation of vector magnitude data sets: Effects of random component

errors. J. Atmos. Ocean. Tech., 14, 695-703.

Grodsky, S. A., V. N. Kudryavtsev, A. Bentamy, J. A. Carton, and B. Chapron (2012), Does

direct impact of SST on short wind waves matter for scatterometry?, Geophys. Res.

Lett., 39, L12602, DOI: 10.1029/2012GL052091.

Jet Propulsion Laboratory (JPL), 2006: QuikSCAT Science Data Product User's Manual

Overview & Geophysical Data Products, ftp://podaac-

ftp.jpl.nasa.gov/allData/quikscat/L2B/docs/QSUG_v3.pdf

McPhaden, M. J., et al. (1998), The Tropical Ocean-Global Atmosphere observing system: A

decade of progress, J. Geophys. Res., 103, 14,169– 14,240.

15 Meindl, E.A., and G.D. Hamilton, 1992: Programs of the National Data Buoy Center, Bull.

Amer. Meteor. Soc., 4, 984-993.

Quilfen, Y., 1995: ERS-1 off-line wind scatterometer products. IFREMER Tech. Rep., 75 pp.

Quilfen Y., Chapron, B., Vandemark, D., 2001: "The ERS Scatterometer Wind Measurement

Accuracy: Evidence of Seasonal and Regional Biases", J. Atmos. and Oceanic Techn.,

18, no. 10, pp. 1684-1697.

Sobieski, P. W., C. Craeye, and L. F. Bliven, 1999: Scatterometric signatures of multivariate

drop impacts on fresh and salt water surfaces, Int, J. Remote Sens., 20, 2149-2166.

Weissman, David E., Mark A. Bourassa, Jeffrey Tongue, 2002: Effects of Rain Rate and

Wind Magnitude on SeaWinds Scatterometer Wind Speed Errors. J. Atmos. Oceanic

Technol., 19, 738–746. doi: 10.1175/1520-0426(2002)019<0738:EORRAW>2.0.CO;2

16 Table 1: Difference between NDBC hourly winds and collocated scatterometer data. Values determined for the common period ERS-2 and QuikSCAT are in parenthesis. Wind Speed Wind Direction Length 9985(3659)

Bias 0.66(0.80) -5(-4) ERS-2 (CMOD2IFR) Std 1.21(1.19) 19(19)

Cor 0.94(0.94) 1.80(1.79)

Length 57714(7720) Bias 0.01(0.03) -3(-5) QuikSCAT Std 1.03(1.02) 16(16) Cor 0.95(0.95) 1.87(1.86) Length 9985(3659)

Bias 0.05(-0.07) -5(-4) ERS/N (CMOD5.n) Std 1.4(1.35) 19(19)

Cor 0.9(0.91) 1.80(1.79)

17 Figure 1. (a) 10m neutral buoy wind speed (NDBC and TAO) binned in 1m/s of ERS-2 wind speed. Histogram of WERS is shown against the right-hand axis. (c) Difference in buoy and ERS-2 wind directions binned in ERS-2 relative wind direction (versus the mid-beam azimuth), horizontal lines indicate the 10o corridor. Histogram of ERS-2 relative wind direction is shown against the right-hand axis. All directions are in azimuth angle from the north (degN). Zero relative wind direction corresponds to antenna looking along the wind vector. Gray shading is STD in each bin.

18 Figure 2. Gross comparison of collocated QuikSCAT (QS) and ERS-2 winds. (a) QS wind speed (WQS ) binned in 1m/s of ERS-2 wind speed (WERS ). Gray shading is the STD of WQS in each bin. (b) Histogram of WERS . (c) Difference in QS and ERS wind directions binned in ERS-2 relative wind direction (versus to the mid-beam azimuth), horizontal lines indicate the 10o corridor. (d) Histogram of ERS-2 relative wind direction. All directions are in azimuth angle from the north (degN). Zero relative wind direction in c) and d) corresponds to antenna looking along the wind vector.

19 Figure 3. (a) Time mean difference of collocated wind speed from QS and ERS-2 (

WQS WERS ), (b) STD difference, and (c) temporal correlation. QuikSCAT rain flag and MRP<0.5 are both applied.

20 Figure 4. Time mean difference between collocated QuikSCAT and ERS2 wind speed: (a) ERS wind based on IFREMER CMOD; (b) ERS wind speed from CMOD5N (ERS/N); (c)

ERS/N wind corrected for GMF dependence, WERS / N  W1(WERS / N , ) ; (d) ERS/N winds corrected by W1 and QS winds corrected for SST dependence, WQS  W 2(WQS , SST ) .

21 Figure 5. (a) Zonally average wind speed difference, (b) histogram of wind speed difference for (1) WQS WERS , (2) WQS WERS / N , (3) WQS  (WERS / N  W1) , (4)

(WQS  W 2)  (WERS / N  W1) . Numbers in the right column of the legend in panel (b) are median wind speed differences in m/s. See also Fig. 4 for notations.

22 Figure 6. (a) QuikSCAT minus ERS/N wind speed difference W1 binned in ERS/N wind speed and wind direction relative to the mid-beam azimuth. (b) SST-related correction for QuikSCAT wind speed ( W 2 , from Bentamy et al., 2012).

23 Figure 7. Time mean (a) GMF-related wind speed correction and (b) SST-related wind speed correction applied to the entire ERS-2/QuikSCAT collocations. Units are m/s.

24 Figure 8. Observed radar backscatter ( 0 ) minus backscatter simulated with the corrected 0 QuikSCAT wind speed and direction ( CMOD5.n ) versus incidence angle for each ERS-2 beam.