Ocean Surface Wind Research Articles

Ocean sound levels in the northeast Pacific recorded from an autonomous underwater glider.

Ocean gliders are a quiet and efficient mobile autonomous platform for passive acoustic monitoring and oceanographic measurements in remote marine environments. During July 20—August 6 2012, we used a Teledyne Webb Research Slocum G2 glider equipped with a hydrophone logging system to record ocean sound along a 458 km north to south traverse of the outer continental shelf break along the U.S. Pacific Northwest coast. Glider derived recordings yielded a unique perspective on the variation of ambient sound with depth, where natural wind generated surface processes were identified as a dominant acoustic contributor to spectral levels in the region. Near and far-field vessel radiated noise were also found to add significant energy to ambient conditions. Spatially distributed measurements of ambient sound levels recorded from the glider were consistent with long-term spectral estimates from fixed station, deep ocean hydrophone array measurements during the 1990–2000’s in the region. Ocean sound level measurements captured by a mobile glider are shown to be an effective and valuable asset for describing ocean surface wind conditions and characterizing spatial and temporal changes in the underwater acoustic environment over a broad regional scale.

Comparison of QuikSCAT, WRF and buoy ocean surface wind data off Valparaiso Bay, Chile

Abstract The winds that affect the surface of the ocean are also important for a vast array of activities, either operational or scientific, hence the importance of being able to adequately predict this quality. Because of the preceding fact, a Weather Research and Forecasting model was used to perform simulations at the surface of the Ocean, for winds derived from different boundary conditions (NCEP-CFSR, ERA-Interim and NCEP-FNL) and configured with different spatial resolution (25, 5 and 1 km), with the objective of evaluating which of these data sets delivers the more precise wind simulation at the surface of the ocean. A comparative analysis was performed between the different outputs of the WRF model, QuikSCAT satellite and in situ observations of a buoy installed off the central coast of Chile. The results showed that the WRF model, overestimates the wind magnitudes, across all boundary conditions or spatial resolution. Additionally, depending on the in situ wind magnitude (>6 ms−1), the model predicts adequately wind magnitude and direction. Spatial comparisons were performed between QuikSCAT and WRF outputs at the Chilean coast to evaluate any possible differences. The modeled winds showed a tendency to be stronger than those measured by Satellites and the bigger differences appeared closer to the shore. The wavelet coherence and phase analysis, confirmed that the model delivers precise wind information for frequencies greater than the daily cycle. Finally, the results of the simulation produced by the ERA-Interim analysis showed lower errors in terms of temporal and spatial variability of surface winds.

An Assessment of NASA’s GMAO MERRA-2 Reanalysis Surface Winds

Abstract The quality of MERRA-2 surface wind fields was assessed by comparing them with 10 years of measurements from a wide range of surface wind observing platforms. This assessment includes a comparison of MERRA-2 global surface wind fields with the ones from its predecessor, MERRA, to assess if GMAO’s latest reanalyses improved the representation of the global surface winds. At the same time, surface wind fields from other modern reanalyses—NCEP-CFSR, ERA-Interim, and JRA-55—were also included in the comparisons to evaluate MERRA-2 global surface wind fields in the context of its contemporary reanalyses. Results show that MERRA-2, CFSR, ERA-Interim, and JRA-55 show similar error metrics while MERRA consistently shows the highest errors. Thus, when compared with wind observations, the accuracy of MERRA-2 surface wind fields represents a clear improvement over its predecessor MERRA and is in line with the other contemporary reanalyses in terms of the representation of global near-surface wind fields. All reanalyses showed a tendency to underestimate ocean surface winds (particularly in the tropics) and, oppositely, to overestimate inland surface winds (except JRA-55, which showed a global tendency to underestimate the wind speeds); to represent the wind direction rotated clockwise in the Northern Hemisphere (positive bias) and anticlockwise in the Southern Hemisphere (negative bias), with the exception of JRA-55; and to show higher errors near the poles and in the ITCZ, particularly in the equatorial western coasts of Central America and Africa. However, MERRA-2 showed substantially lower wind errors in the poles when compared with the other reanalyses.

Monitoring of tropical cyclone structures in ten years of RADARSAT-2 SAR images

The cross-polarized synthetic aperture radar (SAR) onboard the RADARSAT-2 satellite has the capability to simultaneously observe the tropical cyclone (TC) structure and ocean surface winds with high spatial resolution of ~1 km and large swath images of ~450 × 450 km. We propose a methodology to extract TC centers, radius of maximum winds, intensities and the azimuthal wave-number one asymmetric surface wind structures from 75 RADARSAT-2 cross-polarized SAR images over the period from 2008 to 2017. We present a quantitative investigation on both the symmetric mean flow and the asymmetries of azimuthal wave-number one in the surface wind fields retrieved from these SAR images. The results demonstrate that the radius of maximum wind decreases as maximum wind increases, which is consistent with the TC dynamic models for axisymmetric convective heating, and that TCs are more asymmetric when they are weak, and close to symmetric when they are strong.

In-Orbit Performance of the Constellation of CYGNSS Hurricane Satellites

AbstractThe NASA Cyclone Global Navigation Satellite System (CYGNSS) constellation of eight satellites was successfully launched into low Earth orbit on 15 December 2016. Each satellite carries a radar receiver that measures GPS signals scattered from the surface. Wind speed over the ocean is determined from distortions in the signal caused by wind-driven surface roughness. GPS operates at a sufficiently low frequency to allow for propagation through all precipitation, including the extreme rain rates present in the eyewall of tropical cyclones. The spacing and orbit of the satellites were chosen to optimize frequent sampling of tropical cyclones. In this study, we characterize the CYGNSS ocean surface wind speed measurements by their uncertainty, dynamic range, sensitivity to precipitation, spatial resolution, spatial and temporal sampling, and data latency. The current status of each of these properties is examined and potential future improvements are discussed. In addition, examples are given of current science investigations that make use of the data.

A Geophysical Model Function for Wind Speed Retrieval From C-Band HH-Polarized Synthetic Aperture Radar

Synthetic aperture radar (SAR) imagery is routinely acquired at HH-polarization in high-latitude areas for measuring surface wind over the ocean. However, in the contrary of VV-polarization, there is no HH-polarization geophysical model function (GMF) exists to directly retrieve wind speed from SAR images. In general, HH-polarized normalized radar cross section (NRCS) is thus converted into VV-polarization and then conventional CMOD functions are used with auxiliary wind direction information for wind speed retrieval. In this letter, we propose a new GMF for SAR ocean surface wind speed retrieval, called CMODH, which relates the C-band NRCS acquired at HH-polarization over the ocean, to the 10-m height wind speed, incident angle, and relative wind direction. We first use more than 220 000 ENVISAT ASAR radar backscatter measurements collocated with ASCAT winds to derive the CMODH coefficients. Subsequently, 1459 RADARSAT-2 (RS-2) and 428 Sentinel-1A/B (Sl-1A/B) HH-polarized SAR acquisitions under different wind speeds are matched to in situ buoy observations to validate CMODH. The statistical comparisons between SAR-observed and simulated NRCS show a bias of −0.07 dB and a root-mean-square error of 1.62 dB for RS-2, and −0.01 dB and 2.48 dB for S1-1A/B. These results suggest that the proposed CMODH has the potential to directly retrieve ocean surface wind speeds using C-band SAR images acquired at HH-polarization, with no need for NRCS transformation by using various empirical and theoretical polarization ratio models.

On the seasonal and inter-annual variability of the equatorial Indian Ocean surface winds

Variability of the Equatorial Indian Ocean (EIO) winds play crucial roles in driving the upper EIO dynamics and modifying the ocean–atmosphere interactions in the Indian Ocean (IO). This study, using a satellite-sensed high-resolution (0.25° × 0.25°) monthly winds, has revealed many salient features of the seasonal and inter-annual variability of winds at various dynamically significant regions of the EIO. Though annual mean wind pattern of the EIO shows westerlies east of 60°E, the different local areas in this ocean exhibit significant seasonal wind variations. Migrations and fluctuations of southerlies and westerlies primarily determine the EIO wind variability. In the western (eastern) EIO, SW and NE monsoon winds are stronger (weaker) and of longer (shorter) duration. Near to the equator (1°S–1°N), in the central EIO, weaker winds occur during SW and NE monsoons, whereas stronger winds during transition periods. Near the equator, meridional winds show a significant annual period, especially in the western EIO. But zonal winds exhibit semi-annual period east of 55°E with a peak in the central ocean associated with the active westerlies during spring and fall, and annual period west of 55°E in the western EIO due to monsoon reversals. Westerlies are stronger during the fall compared to spring. Zonal wind variability in the central EIO is the essential deciding factor for the zonal wind variability occurring in the whole EIO. Both Zonal Sea Level Pressure Gradient (ZSLPG) and Momentum Advection (MA) fields determine the dynamics of zonal winds in the EIO. Zonal winds are weak in the western EIO because of the semi-annual harmonics of both ZSLPG and MA, though with substantial amplitude, exhibiting opposite accelerations there. But stronger zonal winds occur in the central EIO (60°E–80°E) because of the same accelerations set by the relatively weaker ZSLPG and MA fields. The zonal winds show substantial inter-annual variability compared to the meridional winds. The inter-annual variability of zonal winds shows the predominance of two modes, with EOF1 50% and EOF2 20% of the total variance. Zonal wind pattern of EOF1 resembles the monsoon transition seasons, while that of EOF2 describes the SW monsoon. In the EIO, Indian Ocean Dipole (IOD) influences the patterns of winds and SST anomalies better than that of El Nino–Southern Oscillation (ENSO). Spatiotemporal variability of the EIO winds during ENSO and IOD indicates the prevalence of stronger zonal wind anomalies in the eastern and central EIO and weaker anomalies in the western EIO. Although spatial patterns of wind anomalies are similar during El Nino and Positive IOD (PIOD) periods, the monthly variations, spatial extensions, and intensities of anomalies are much different while comparing. Prospectus for understanding the EIO winds variations can help to understand the changes occurring in the wind-driven surface and subsurface zonal currents.

Marine Environment Distinctions and Change Law Based on eCongnition Remote Sensing Technology

Zhang, X.-L.; Li, Z.; Li, D., and He, Y.-K., 2019. Marine environment distinctions and change law based on eCongnition remote sensing technology. In: Gong, D.; Zhu, H., and Liu, R. (eds.), Selected Topics in Coastal Research: Engineering, Industry, Economy, and Sustainable Development. Journal of Coastal Research, Special Issue No. 94, pp. 107–111. Coconut Creek (Florida), ISSN 0749-0208.The extraction of marine environment distinctions is of utmost importance to the development of marine economy and environmental conservation. Traditional extraction method has some defects such as less observation data, large time span and vulnerability to vile weather on the ocean. The eCongnition remote sensing technology features fast classification, high precision and rises superior to the bad weather of the ocean. This paper attempts to apply the eCongnition remote sensing technology to the extraction of marine environment features in order to seek out its changing laws. The findings show that the eCongnition remote sensing technology can analyze the spectral, spatial and textural identities of the ocean. By extracting the characteristics of offshore fishery resources and ocean surface wind speed information, it is found that the marine fishery has a direct bearing on the marine environment factors, i.e. the surface temperature and chlorophyll-a concentration. The optimal ocean surface temperature is 29.5°C, and the optimal chlorophyll-a concentration is 0.4 mg/m3; the greater the distance from the sea level, the higher the wind speed on the ocean surface. Experimentally, this study provides the clues to the application of eCongnition remote sensing technology in marine feature extraction.

Retrieval of sea surface salinity from SMAP L‐band radiometer: A novel approach for wind speed correction

AbstractThe present article introduces a new sea surface salinity (SSS) retrieval technique from the Soil Moisture Active Passive (SMAP) L–band (1.4 GHz) microwave radiometer. The ocean surface wind speed (WS) correction, which is very critical in any of the SSS retrieval techniques, is accounted for with a different approach. A model function between ΔSSS (the difference between SSS of flat and rough ocean surfaces) and ocean surface WS is developed and used to correct the impact of ocean surface roughness due to the winds on SSS. In addition to WS correction, the retrieval technique appropriately accounts for the effects of atmospheric attenuation on L‐band frequency due to water vapour and oxygen, and ocean surface temperature on SSS. The performance of the estimations is statistically evaluated by comparing the estimations with the in situ Argo, buoy SSS, and SMAP version 3, 70 km resolution SSS observations in global oceans during 2016. The retrieved salinity well captured the global distribution patterns of SSS, including low and high values. Comparison of retrieved salinity with Argo and SMAP products show a root‐mean‐square error (RMSE) of 0.32 and 0.39 psu, a Pearson correlation coefficient of 0.94 and 0.93, and a scatter index of 0.0093 and 0.011 respectively on a monthly time‐scale. In addition, retrieved SSS promisingly captured day‐to‐day variability in SSS observed by tropical moored buoy SSS with RMSE of 0.36 psu and a Pearson correlation coefficient of 0.94. The validation confirms the potential ability of the new retrieval technique in capturing SSS changes on daily and monthly time‐scales.

Tracking fishing ground parameters in cloudy region using ocean colour and satellite-derived surface flow estimates: A study in the Bay of Bengal

ABSTRACT A new conceptual framework, based on ocean bio-physical observations from different satellites, has been proposed to track fishing ground parameters to identify Potential Fishing Zones (PFZ) in the Bay of Bengal (BoB). The proposed technique also attempts to provide a short-term forecast based on feature propagation, even under cloudy conditions. The current study has been carried out to understand the link between possible fish catch availability and satellite-derived parameters such as chlorophyll concentration, sea surface temperature, sea-level anomaly, ocean surface currents and wind vectors. Net Ekman transport obtained from the upfront and downfront components of the wind relative to the frontal direction provides valuable information on the forecast of PFZ regions and its possible shift. Comparison of the identified PFZs with limited in situ fish catch data proves that the suggested approach holds the promise in improved monitoring of possible fish catch locations. The study provides a novel approach for the monitoring and short-term forecasting of PFZ even for cloud-contaminated regions by using all possible ocean information from space-based platforms.

C-Band Right-Circular Polarization Ocean Wind Retrieval

We report an investigation of ocean-surface wind speed retrieval from C-band RADARSAT Constellation Mission (RCM) Synthetic Aperture Radar (SAR) images using a new channel of coright-circular polarization (RR-pol) in compact polarimetry (CP) option. The analysis of simulated RCM quad-polarized CP SAR data and collocated in situ buoy measurements suggests that the RR-pol is much less sensitive to wind directions than the other three polarizations in the CP option. A method is proposed for the RR-pol radar signal as a function of wind speed and incidence angle. We demonstrate that C-band RR-pol has the potential for ocean high wind retrieval, especially for cyclone studies.

A Geophysical Model Function for S‐Band Reflectometry of Ocean Surface Winds in Tropical Cyclones

AbstractOcean surface wind remote sensing using reflectometry measurements from S‐band (2.3 GHz) digital communication satellite transmissions is demonstrated in tropical cyclones. Twelve days of airborne data were collected during the 2014 Atlantic hurricane season. An empirical model function, relating the sensed ocean surface mean square slope to wind speed, was first developed by fitting mean square slope estimates to comparison wind speed data from the Stepped Frequency Microwave Radiometer on the same aircraft and wind direction data from the Hurricane Weather Research and Forecasting model. This model function was then applied for wind speed retrievals using an independent set of reflectometry measurements and validated against four data sources: Stepped Frequency Microwave Radiometer, the Hurricane Weather Research and Forecasting model, dropsondes, and flight‐level winds. Good agreement between these retrievals and comparison data was found, with a root‐mean square error between 4.9 and 6.6 m/s and a bias between −1.4 and −0.2 m/s for wind speeds up to 50 m/s.

SEASTAR: A Mission to Study Ocean Submesoscale Dynamics and Small-Scale Atmosphere-Ocean Processes in Coastal, Shelf and Polar Seas

High-resolution satellite images of ocean color and sea surface temperature reveal an abundance of ocean fronts, vortices and filaments at scales below 10 km but measurements of ocean surface dynamics at these scales are rare. There is increasing recognition of the role played by small scale ocean processes in ocean-atmosphere coupling, upper-ocean mixing and ocean vertical transports, with advanced numerical models and in situ observations highlighting fundamental changes in dynamics when scales reach 1 km. Numerous scientific publications highlight the global impact of small oceanic scales on marine ecosystems, operational forecasts and long-term climate projections through strong ageostrophic circulations, large vertical ocean velocities and mixed layer re-stratification. Small-scale processes particularly dominate in coastal, shelf and polar seas where they mediate important exchanges between land, ocean, atmosphere and the cryosphere e.g. freshwater, pollutants. As numerical models continue to evolve towards finer spatial resolution and increasingly complex coupled atmosphere-wave-ice-ocean systems, modern observing capability lags behind, unable to deliver the high-resolution synoptic measurements of total currents, wind vectors and waves needed to advance understanding, develop better parameterizations and improve model validations, forecasts and projections. SEASTAR is a satellite mission concept that proposes to directly address this critical observational gap with synoptic two-dimensional imaging of total ocean surface current vectors and wind vectors at 1 km resolution and coincident directional wave spectra. Based on major recent advances in squinted along-track Synthetic Aperture Radar interferometry, SEASTAR is an innovative, mature concept with unique demonstrated capabilities, seeking to proceed towards spaceborne implementation within Europe and beyond.

Ocean Surface Wind Speed Retrieval Using Simulated RADARSAT Constellation Mission Compact Polarimetry SAR Data

We investigated the use of C-band RADARSAT Constellation Mission (RCM) synthetic aperture radar (SAR) for retrieval of ocean surface wind speeds by using four new channels (right circular transmit, vertical receive (RV); right circular transmit, horizontal receive (RH); right circular transmit, left circular transmit (RL); and right circular transmit, right circular receive (RR)) in compact polarimetry (CP) mode. Using 256 buoy measurements collocated with RADARSAT-2 fine beam quad-polarized scenes, RCM CP data was simulated using a “CP simulator”. Provided that the relative wind direction is known, our results demonstrate that wind speed can be retrieved from RV, RH and RL polarization channels using existing C-band model (CMOD) geophysical model function (GMF) and polarization ratio (PR) models. Simulated RR-polarized radar returns have a strong linear relationship with speed and are less sensitive to relative wind direction and incidence angle. Therefore, a model is proposed for the RR-polarized synthetic aperture radar (SAR) data. Our results show that the proposed model can provide an efficient methodology for wind speed retrieval.

Air-Sea Fluxes With a Focus on Heat and Momentum

Turbulent and radiative exchanges of heat between the ocean and atmosphere (hereafter heat fluxes), ocean surface wind stress, and state variables used to estimate them, are Essential Ocean Variables (EOVs) and Essential Climate Variables (ECVs) influencing weather and climate. This paper describes an observational strategy for producing 3-hourly, 25-km (and an aspirational goal of hourly at 10-km) heat flux and wind stress fields over the global, ice-free ocean with breakthrough 1-day random uncertainty of 15 W m-2 and a bias of less than 5 W m-2. At present this accuracy target is met only at OceanSITES reference station moorings and research vessels (RVs) that follow best practices. To meet these targets globally, in the next decade, satellite-based observations must be optimized for boundary layer measurements of air temperature, humidity, sea surface temperature, and ocean wind stress. In order to tune and validate these satellite measurements, a complementary global in situ flux array, built around an expanded OceanSITES network of time series reference station moorings, is also needed. The array would include 500 - 1000 measurement platforms, including autonomous surface vehicles, moored and drifting buoys, RVs, the existing OceanSITES network of 22 flux sites, and new OceanSITES expanded in 19 key regions. This array would be globally distributed, with 1 - 3 measurement platforms in each nominal 10° by 10° boxes. These improved moisture and temperature profiles and surface data, if assimilated into Numerical Weather Prediction (NWP) models, would lead to better representation of cloud formation processes, improving state variables and surface radiative and turbulent fluxes from these models. The in situ flux array provides globally distributed measurements and metrics for satellite algorithm development, product validation, and for improving satellite-based, NWP and blended flux products. In addition, some of these flux platforms will also measure direct turbulent fluxes, which can be used to improve algorithms for computation of air-sea exchange of heat and momentum in flux products and models. With these improved air-sea fluxes, the ocean’s influence on the atmosphere will be better quantified and lead to improved long-term weather forecasts, seasonal-interannual-decadal climate predictions, and regional climate projections.

Integrated Observations of Global Surface Winds, Currents, and Waves: Requirements and Challenges for the Next Decade

Ocean surface winds, currents, and waves play a crucial role in exchanges of momentum, energy, heat, freshwater, gases, and other tracers between the ocean, atmosphere, and ice. Despite surface waves being strongly coupled to the upper ocean circulation and the overlying atmosphere, efforts to improve ocean, atmospheric, and wave observations and models have evolved somewhat independently. From an observational point of view, community efforts to bridge this gap have led to proposals for satellite Doppler oceanography mission concepts, which could provide unprecedented measurements of absolute surface velocity and directional wave spectrum at global scales. This paper reviews the present state of observations of surface winds, currents, and waves, and it outlines observational gaps that limit our current understanding of coupled processes that happen at the air-sea-ice interface. A significant challenge for the coming decade of wind, current, and wave observations will come in combining and interpreting measurements from (a) wave-buoys and high-frequency radars in coastal regions, (b) surface drifters and wave-enabled drifters in the open-ocean, marginal ice zones, and wave-current interaction ``hot-spots'', and (c) simultaneous measurements of absolute surface currents, ocean surface wind vector, and directional wave spectrum from Doppler satellite sensors.

Space of solutions to ocean surface wind measurement using scatterometer constellations

Satellite wind vector data are integral to atmospheric models and forecasts, but current measurement limitations make some synoptic, mesoscale activities difficult to observe. Using miniaturized electronics and advanced deployable mechanisms, new satellite wind scatterometers may be possible that increase spatial, temporal, and wind resolution and coverage. We propose a simple parametric model of the space of satellite wind scatterometer designs, their performances, and the transformation between design and performance, together called a solution space model. We explore two applications of this model: understanding how advances have expanded the design space and searching for alternative approaches to satellite wind scatterometry. Recent advances enable a greater capability-to-volume ratio, which enables constellations of small, low-cost scatterometers that co-operate in a variety of modes. We present two example concepts for constellations of co-operative satellite wind scatterometers. We estimate that a constellation of CubeSat scatterometers may affordably measure global ocean vector winds every three hours.

Effect of Faraday Rotation on L-Band Ocean Normalized Radar Cross Section and Wind Speed Detection

Observational evidence that L -band backscatter measurements are distorted by Faraday rotation (FR) is presented using match-ups consisting of the phased-array type L -band synthetic aperture radar-2 (PALSAR-2) data, scatterometer winds, and FR estimated from total electron contents (TEC) and magnetic fields. Investigating these data reveals that the PALSAR-2-derived ocean surface HH normalized radar cross section (NRCS) tends to decrease slightly as the FR angle increases and the HV one tends to increase the most. At an incidence angle of 27.5°, the measured HV NRCS increases by about 15 dB or more, whereas the FR angle increases from 0° to 15°. The variation of the PALSAR-2 observations with FR and its incident angle dependence are roughly explained by a model based on the scattering matrix of linearly polarized backscatter signatures. The error of wind speeds estimated by the L- band HH geophysical model function of ocean surface winds from the match-ups shows a tendency to underestimate wind speeds as FR increases. This is explained by the fact that the HH NRCS decreases as FR increases. Meanwhile, it is suggested that because ocean cross-polarized (HV) data with large co- and cross-polarization ratio has strong sensitivity to FR due to the inclusion of the co-polarization component, correction for the influence of the FR angle is essential for accurate ocean wind estimation.

Machine Learning Approaches for Detecting Tropical Cyclone Formation Using Satellite Data

This study compared detection skill for tropical cyclone (TC) formation using models based on three different machine learning (ML) algorithms-decision trees (DT), random forest (RF), and support vector machines (SVM)-and a model based on Linear Discriminant Analysis (LDA). Eight predictors were derived from WindSat satellite measurements of ocean surface wind and precipitation over the western North Pacific for 2005–2009. All of the ML approaches performed better with significantly higher hit rates ranging from 94 to 96% compared with LDA performance (~77%), although false alarm rate by MLs is slightly higher (21–28%) than that by LDA (~13%). Besides, MLs could detect TC formation at the time as early as 26–30 h before the first time diagnosed as tropical depression by the JTWC best track, which was also 5 to 9 h earlier than that by LDA. The skill differences across MLs were relatively smaller than difference between MLs and LDA. Large yearly variation in forecast lead time was common in all models due to the limitation in sampling from orbiting satellite. This study highlights that ML approaches provide an improved skill for detecting TC formation compared with conventional linear approaches.

Subsurface temperature estimation from remote sensing data using a clustering-neural network method

The use of remote sensing observation to estimate subsurface oceanic variables, including subsurface temperature anomaly (STA), is essential for the study of ocean dynamics and climate change. Here we report a new method that combines a pre-clustering process and a neural network (NN) approach to determine the STA using ocean surface temperature, surface height, and surface wind observation data at the global scale. Gridded monthly Argo data were used in the training and validation procedures of the method. Results show that the pre-clustered NN method was better than the same method without clustering, while also outperforming a clustered linear regressor and the random forest method recently reported. The new method was tested over a wide range of time (all months from 2004 to 2010) and depth (down to 1900 m). Overall, our best estimation resulted in an overall root-mean-squared error of 0.41 °C and a determination coefficient (R2) of 0.91 at the 50 m level for all months. The R2 decreased to 0.51 at 300 m but was still better than the calculation without pre-clustering. This method can be expanded to estimate other key oceanic variables and provide new insights in understanding the climate system.