We developed a ResNet methodology based on Convolutional Neural Network (Zanchetta and Zecchetto, 2021), able to estimate the wind direction at a spatial resolution of 500 m by 500 m without external information. The ResNet model can derive the wind field even in the absence of wind streaks, in presence of convective turbulence structures, atmospheric lee waves, and ships. It is indicated to extract wind information over small areas, as the example of Venice lagoon. In this work the wind fields have been producet using the directions from ResNet and the scatterometer-based Geophysical Model Function CMOD7 (Stoffelen et al., 2017).
The possibility offered by ResNet led us to investigate the characteristics of the strongest winds blowing on the northern Adriatic Sea and Venice Lagoon, Italy. The area of interest is subjected to high spatial and temporal variability of wind, a peculiarity of many coastal areas, making it a very demanding site.
The structure of the wind systems inside and outside the lagoon has been studied in terms of spatial variability of speed, direction and vertical velocity wek in the Ekman layer derived by ResNet.
The layout of wek exhibits contiguous cells of upward and downward motion elongated orthogonally to the wind direction with periodicity of 5.4 km. This spatial variability seems to be a signature of the atmospheric Ekman pumping, produced by local variations of direction and speed.
An example of results from ResNet and OCN is reported in Fig. 1, which shows the ResNet (left panel) and the OCN SAR wind fields over the Venice lagoon: differently from the ESA OCN winds, the unprecedented resolution obtained with the ResNet allows an exhaustive coverage of the Venice lagoon, making possible to investigate the spatial structure of wind fields. For instance, under northeastern storms (Bora), the wind speed increases from northern to southern lagoon by 30% in average, in agreement with a case study carried out on experimental data (Zecchetto et al., 1997).
SAR winds derived by ResNet have been compared with the in-situ and ECMWF model data, showing on average, a 9% of underestimation and 7% of overestimation respectively, in the range from 4 ms-1 to 25 ms-1. The overestimation of SAR derived winds with respect to ECMWF confirms the results obtained in the Adriatic basin from comparisons between scatterometer and ECMWF winds (Zecchetto et al., 2015), while the underestimation with respect to the in-situ data conveys the difference between ECMWF and in-situ winds of ~10%.
The importance of a correct determination of the wind direction has been tested by comparing the SAR wind fields produced using ResNet and ECMWF wind directions, which may differ locally up to ±30º: these discrepancies may produce local differences of wind speed as large as ±2 ms-1.
Detailed analysis of selected cases raised the issue of the lack of data with true spatial resolution of O(1) km and within half hour from the satellite pass time necessary for exhaustive comparisons.


References

Stoffelen, A., Verspeek, A., Vogelzang, J., Verhoef, A., 2017. The CMOD7 Geophysical Model Function for ASCAT and ERS Wind Retrievals. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 10, 2123–2134, doi:10.1109/JSTARS.2017.2681806 .

Zecchetto, S. , G. Umgiesser and M. Brocchini, Hindcast of a Storm Surge Induced by Local Real Wind Fields in the Venice Lagoon, Continental Shelf Research, Vol.17 No.12,1513-1538, 1997

Zecchetto, S., della Valle, A., De Biasio, F., 2015. Mitigation of ECMWF–scatterometer wind biases in view of storm surge applications in the Adriatic Sea. Adv. Space Research 55, 1291–1299. doi:10.1016/j.asr.2014.12.011 .

Zanchetta, A. and S. Zecchetto, Wind direction retrieval from Sentinel-1 SAR images using ResNet, Remote Sensing of Environment, 253, 2021 (https://doi.org/10.1016/j.rse.2020.112178)

The Chinese French Ocean Satellite (CFOSAT) is an innovative space mission dedicated to the global observation and monitoring of the ocean sea state and the sea surface vector winds. CFOSAT operates two Ku-band rotating radars: the nadir/near-nadir Ku-band wave scatterometer (SWIM) and the dual-polarization, moderate incidence angle, Ku-band wind scatterometer (SCAT). This unique instrumental configuration provides regular collocated measurements of radar backscatter to retrieve sea surface state parameters, including significant wave height, directional wave spectrum, and wind vector. Two sensors also give the opportunity to improve the quality of the retrieved parameters by combining both data sources. In particular, this approach can be applied for the improvement of SCAT wind retrievals using SWIM observations.

The effective backscattering properties of SWIM and SCAT do not perfectly match the commonly used Ku-band Geophysical Model Function (GMF) due to various reasons like radar antenna design, swath patterns and noise signal distortions. On the other hand, observations for different incidence angles have different sensibility to sea surface parameters: short and long waves, surface currents, surface temperature, etc. The joint use of multi-instrument measurements within a common processing framework, i.e. scatterometer wind vector inversion Maximum Likelihood Estimator procedure, brings a potential risk of a significant error multiplication due to models and observation data mismatch.

The relation between collocated backscatter (σ0) measurements and various environmental parameters could be justified with the new common GMF which describes geophysical and CFOSAT-specific instrumental properties for all onboard sensors in a unified form. Such alternative Ku-band GMF was developed using a neural network (NN) approach. The traditional set of GMF variables (wind vector, incidence angle, polarization, ….) was extended with various additional geophysical parameters which can impact the signal properties: significant wave height, sea surface current vector, sea surface temperature, ice concentration, precipitation rate. The NN learning data set is based on CFOSAT measurements and collocated model data as provided by IFREMER Wave and Wind Operational Center (IWWOC) with SWISCA S Level 2 product. To avoid model biasing, special attention was addressed to the normalization and uniformization of input values during the learning process. As well, the numerical learning strategy was adapted to reduce the negative impact of using numerical weather prediction models (NWP) in the backscatter measurements regression task. The derived NN GMF reproduces the main features of NSCAT-4 GMF for moderate incidence angles and TRMM/GPM GMF for near-nadir observations. However, instrument-specific features are clearly present as well.

The resulting NN GMF could be considered as the approximation of Ku-band radar cross-section as a function of a multi-parameter environment. This function allows separating the impact of different geophysical variables on the backscattering coefficient value. The flexible nature of the proposed approach naturally enables the inclusion of any additional sea state variables to GMF. Additionally, it provides a robust platform for rapid signal calibration and re-adjustment during mission exploitation. We anticipate the implementation of the demonstrated model to extend the existing SCAT data processing with collocated SWIM nadir/near-nadir observations and additional NWP variables. As well, this approach can be suggested for the implementation in other scatterometry processing chains associated with different instruments and sensing microwave bands.

Radiance measurements from spaceborne microwave instruments are the most impactful observations used in Numerical Weather Prediction (e.g. Eyre, English and Forsythe 2020). Sophisticated data assimilation methods such as 4D-Var have been critical to this success, enabling direct assimilation of raw radiances. However, until recently, different Earth System components such as ocean, land and atmosphere were always handled separately, meaning those radiances which are sensitive to more than one component are still assimilated sub-optimally. The development of coupled data assimilation methodologies enables us to take another big step in the use of radiances, simultaneously and consistently fitting the state in multiple sub-systems to the same observations. This requires improved surface radiative transfer models.

For the ocean, although in general physically based models are used in data assimilation, at least for modelling passive microwave observations, the uncertainty is not well known and often different models are used for different spectral bands, and for active and passive sensing instruments. Furthermore, for active-instrument, empirical Geophysical Model Functions are used, which are very accurate (~0.1 dB), but physically-based methods have lower accuracy (Fois, 2015). In attempting error budget closure, lack of knowledge of uncertainty in surface emission was a limiting factor (GAIA-CLIM: www.gaia-clim.eu/). An International Space Science Institute team was created (English et al. 2020) to address this gap, taking the best available model components, integrating, testing across all spectral bands and characterizing as far as possible the uncertainty. The resulting reference model will then be provided as community software on GitHub.

In this short presentation, the choices made assembling this model will be explained, building on the starting point of the LOCEAN model of Dinnat et al. (2003). Samples of characterization undertaken will also be summarized. This includes comparison to SMAP, AMSR2 and GMI (e.g. Kilic et al. 2019) and early work to evaluate in the infrared and for active sensors. Finally, the plans for making code available will be briefly presented. This model will also be used to generate training data for fast models, e.g. Fastem (English and Hewison 1998), as used in operational data assimilation and climate re-analysis.

References

Dinnat, E. P., Boutin, J., Caudal, G., and Etcheto, J., 2003 : Issues concerning the sea emissivity modeling at L band for retrieving surface salinity, Radio Sci., 38, 8060, https://doi.org/10.1029/2002RS002637

English, S., Prigent, C., et al., 2020: Reference-quality emission and backscatter modeling for the Ocean, B. American Meteorol. Soc., 101(10), 1593-1601. https://doi.org/10.1175/BAMS-D-20-0085.1

English S.J. and Hewison T.J., 1998: Fast generic millimeter-wave emissivity model, Proc. SPIE 3503, Microwave Rem. Sens. Atmos. Env., https://doi.org/10.1117/12.319490

Eyre, J.R., English, S.J., Forsythe, M., 2020: Assimilation of satellite data in numerical weather prediction. Part I: The early years. Q J R Meteorol Soc. 2020; 146: 49– 68. https://doi.org/10.1002/qj.3654

Fois, F., 2015, Enhanced ocean scatterometry, PhD Delft University of Technology, Delft, the Netherlands, doi = 10.4233/uuid:06d7f7ad-36a9-49fa-b7ae-ab9dfc072f9c .

Kilic, L., Prigent, C., Boutin, J., Meissner, T., English, S., & Yueh, S., 2019: Comparisons of ocean radiative transfer models with SMAP and AMSR2 observations., J. Geophys. Res.: Oceans, 124, 7683– 7699. https://doi.org/10.1029/2019JC015493

As more than 70% of the earth surface is covered by water, exchanges of heat, gases and momentum at the air-sea interface are a key part of the dynamical earth system and its evolution. The ocean surface wind plays an essential role in the exchange at the atmosphere-ocean interface. It is therefore crucial to accurately represent the wind forcing in physical ocean model simulations. Scatterometers provide high-resolution ocean surface wind observations, but have limited spatial and temporal coverage. On the other hand, numerical weather prediction (NWP) model wind fields have better coverage in time and space, but do not resolve the small-scale variability in the air-sea fluxes. In addition, Belmonte and Stoffelen (2019) documented substantial systematic errors in global NWP fields on both small and large scales, using scatterometer observations as a reference.

Trindade et al. (2020) combined the strong points of scatterometer observations and atmospheric model wind fields into ERA*, a new ocean wind forcing product. ERA* uses temporally-averaged differences between geolocated scatterometer wind data and European Centre for Medium-range Weather Forecasts (ECMWF) reanalysis fields (ERA-Interim) to correct for persistent local NWP wind vector biases. Verified against independent observations, ERA* reduced the variance of differences by 20% with respect to the uncorrected NWP fields.

We present a new hourly ocean wind forcing product that will be included in the Copernicus Marine Service (CMEMS) catalogue. To best serve the ocean modelling community, this Level 4 product will include global bias-corrected 10-m stress-equivalent wind (De Kloe et al., 2017) and surface wind stress fields at 0.125 degree horizontal spatial resolution. The near real-time (NRT) version of the product is based on the ECMWF operational model (OPS*) and the reprocessed (REP) version on the ERA5 re-analysis (ERA5*). Like any CMEMS product, the new wind product will be freely and openly available for all operational, commercial and research applications.

References:
Belmonte Rivas, M. and A. Stoffelen (2019): Characterizing ERA-Interim and ERA5 surface wind biases using ASCAT, Ocean Sci., 15, 831–852, doi: 10.5194/os-15-831-2019.

Kloe, J. de, A. Stoffelen and A. Verhoef (2017), Improved use of scatterometer measurements by using stress-equivalent reference winds, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 10 (5), doi: 10.1109/JSTARS.2017.2685242.

Trindade, A., M. Portabella, A. Stoffelen, W. Lin and A. Verhoef (2020), ERAstar: A High-Resolution Ocean Forcing Product, IEEE Trans. Geosci. Remote Sens., 1-11, doi: 10.1109/TGRS.2019.2946019.

Local variability of sea surface wind has a significant impact on the mesoscale air-sea interactions and the wind-induced oceanic response, such as temperature variability and circulation patterns. Recent advances in the wind quality control of Advanced Scatterometer (ASCAT) show that wind variability within a wind vector cell can be characterized using certain quality indicators derived from ASCAT data, such as the inversion residual (namely the maximum likelihood estimator, MLE) and the singularity exponent (SE) derived from singularity analysis.

This study is aimed at quantifying the ASCAT subcell wind variability over the global ocean surface. It is assumed that the spatial variability is proportional to the variance associated with time-series of collocated moored buoys winds. As such, 10-min sampled buoy winds are used to examine the subcell wind variability following Taylor’s hypothesis, which allows for a temporal dimension to be converted into a spatial dimension, and vice versa. The time window (centered on the buoy measurement collocated with ASCAT acquisition) used for calculating the mean buoy winds and the subcell spatial variability is set equal to 25 km. Then the sensitivity of ASCAT quality indicators to the subcell wind variability is evaluated. The results indicate that SE is more sensitive than MLE in characterizing the wind variability, but they are rather complementary in flagging the most variable winds. Consequently, an empirical model is derived to relate the subcell wind variability to the ASCAT MLE and/or SE values.

Although the overall procedure is based on the one-dimensional temporal analysis and the empirical model cannot fully represent the two-dimensional spatial variability as depicted by the scatterometer, it is probably the first attempt to assign a subcell wind variability value for each wind vector cell within the ASCAT swath. The empirical method presented here is effective, straightforward, and could be applied to other scatterometer systems. The next step is therefore generate global wind variability maps which can be used in a wide variety of scientific and operational applications.

The presence of horizontal spatial structures in the sea surface temperature (SST) field is known to influence the atmospheric response at various time scales. The ESA CCI (Climate Change Initiative) GLAUCO (Global and Local Atmospheric response to the Underlying Coupled Ocean) project aims to characterize the wind, cloud and rainfall response to the SST structures at daily and sub-daily time scales, with a focus on the physical mechanisms responsible for that. In the literature, two main mechanisms have been identified: the Downward Momentum Mixing mechanism (DMM) mechanism and the Pressure Adjustment (PA) one. According to DMM, a positive change in SST along the wind decreases the stability of the lower atmosphere, which enhances the vertical mixing of horizontal momentum. This results in a net acceleration of the low level flow, producing wind divergence over SST fronts (from relatively cold water to relatively warm water). According to PA, the presence of a warm SST patch induces a local pressure low, responsible for pressure gradients that generate secondary circulations. Surface wind convergence (divergence) is then produced over local SST maxima (minima).
The long-term, consistent and unbiased climate data records produced within the ESA CCI project (and its extensions) are used. These data sets enable to robustly estimate the long-term statistics of the atmospheric response at fast (daily and sub-daily) scales. A “globally local” approach is pursued, where regional differences can be assessed and compared using observational products that are consistent at the global level. As different levels of processing are available, often associated with a different grid spacing, the dependence of the results on the size of the resolved SST structures can also be assessed. This can shed some light on the ability of general circulation models in representing these small-scale fast air-sea interactions and their impact on the atmospheric dynamics.
Typically, to determine the importance of the thermal mechanisms introduced above, one calculates correlation coefficients or the slope of the binned distributions (named coupling coefficients) of: downwind SST gradient and wind divergence for DMM, and SST Laplacian and wind divergence for PA. However, the advection has been observed to break the correlation between SST Laplacian and wind divergence, so that the PA mechanism has been often overlooked in the literature. As the pressure response is produced in all directions, we propose to measure the correspondence of the across-wind SST second spatial derivative and the across-wind divergence to identify the action of the PA mechanism. It is found that this new metrics detects a signal only when small-scale SST forcing is present.
By applying this new across-wind metrics to high resolution satellite data, namely the ESA CCI SST data at 0.05° and the L2 Metop-A ASCAT wind field swaths at 12.5 km, new interesting features of the wind response to the SST forcing appear. First of all, the signature of the PA mechanism appears in regions where this mechanism was thought not to be present. Moreover, both DMM and PA mechanisms show strong seasonal behaviors and regional differences. Non-linearities and asymmetries in the response according to the sign of the forcing field emerge, highlighting that the assumption of linear atmospheric response that has often been done does not hold at fine spatio-temporal scales. Thus, for a proper characterization of the air-sea fluxes, high-resolution simultaneous observations of SST, surface currents and MABL properties are needed, as pursued by the Earth Explorer 10 (EE10) candidate mission Harmony and the EE11 candidate mission Seastar.