GLOBWAVE: A GLOBAL WAVE DATA PORTAL

Farquhar, C.R.1, Deighton, H.1, Busswell, G.1, Snaith, H.M. 2, Ash, E.3, Collard, F.4, Piolle, J-F.5, Poulter, D. J. S. 2, & Pinnock, S.6

1 Logica, Keats House, Leatherhead, Surrey, UK 2 National Oceanographic Centre, Southampton, UK 3 Satellite Oceanographic Consultants, UK 4 Collecte Localisation Satellites, France 5 Ifremer, France 6 ESA, ESRIN, Frascati, Rome, Italy

ABSTRACT

The GlobWave Project is an initiative funded by the European Space Agency (ESA) and subsidised by CNES through the Data User Element (DUE), a programmatic element of the 3rd period of the Earth Observation Envelope Programme (EOEP-3), an optional ESA programme. GlobWave is led by Logica, with key expertise provided from Satellite Oceanographic Consultants Ltd (SatOC), Collecte Localisation Satellites (CLS), Ifremer and National Centre (NOC) and aims to improve the uptake of satellite-derived -wave and data by the scientific, operational and commercial user communities. The project, running from January 2009 – December 2012, covers the development of an integrated set of information services based on satellite wave data, and the operation and maintenance of these services for a demonstration period.

This paper describes the datasets provided and the activities carried out through the project which are intended to make it easier for the global user community to use satellite wave data, to facilitate routine comparison with wave models and to stimulate the development of satellite wave data assimilation.

1. INTRODUCTION

The ESA GlobWave project is a three year initiative, funded by ESA and CNES, serving the needs of satellite wave product users across the globe. GlobWave provides free access to consolidated satellite wave data and products from both SAR and altimeter missions in a common netCDF format. Both historic and near real-time data are provided, with the latter being available within 1 hour of receiving the data from the relevant agency. In addition, GlobWave provides comparisons with in situ measurements, interactive data analysis tools and a pilot spatial and spectral wave forecast verification scheme for operational forecast production centres.

2. BACKGROUND

Significant efforts have been made in the past to provide integrated and harmonized satellite data sets in other fields of oceanography, such as the Medspiration project for surface temperature and the AVISO portal for sea surface height [1]. Despite the important benefits of the application of wave observations from space, and the relatively good availability of satellite wave data, uptake by the potential user community had been found to be less than optimal. For example, only very few meteorological centres routinely assimilate altimeter wave data and fewer still assimilate the information available from SAR. GlobWave aims to address these needs and the needs of the user community gathered from a regular series of user consultations held since 2007 by providing a "one stop shop" for satellite wave data and to encourage widespread added value initiatives, like the assimilation of altimeter and SAR data into models.

Information on wind-driven sea surface waves is of high importance to shipping, offshore industries, coastal engineering, weather forecasting, coastal zone management, and even tourism. Safe transport of goods and people by ships, and of fishing fleets, is dependent on a timely knowledge of the . Although ships and offshore installations are designed to withstand extreme weather conditions, the risk of accidents is higher under severe and unusual sea states, such as and cross . Marine engineering operations, such as those performed by the oil industry, aquaculture and offshore wind power operators, are also very sensitive to sea state.

In addition, sea state is an important factor governing the air-sea fluxes of momentum, heat, water vapour and gas transfer, and needs to be accounted for in modelling the interaction of the sea with the atmosphere for accurate weather forecasting and climate research. It is also a required input parameter of the -atmosphere coupling schemes of climate models, and is fundamental to the corrections required to derive climate quality sea surface topography. Waves affect sediment transport along coastlines, contributing to erosion and changing coastal morphology, and waves can combine with storm surges to increase the risk of coastal flooding.

3. GLOBWAVE DATA

Since it began in January 2009, GlobWave has successfully increased the uptake of satellite-derived wind-wave and swell data by the scientific, operational and commercial user communities. This has been achieved by adhering to the core objective of providing a uniform, harmonized, quality controlled, multi-sensor set of satellite wave data and ancillary information in a common format, with a consistent characterization of errors and biases [2].

3.1 GlobWave Satellite Database

GlobWave satellite wave data is collected from both altimeters (ERS-1, ERS-2, ENVISAT, Topex/POSEIDON, Jason-1, Jason-2, CryoSAT, GEOSAT and GEOSAT Follow On) and from ESA Synthetic Aperture Radar (SAR) missions, namely ERS-1, ERS-2 and ENVISAT. In the future, such measurements will be continued by several missions, including ESA's upcoming Sentinel series. GlobWave satellite data is available both in near real-time and delayed mode. The delayed mode data consists of 12 data streams from 9 satellites from as far back as 1985. The near real-time data consists of 4 data streams from 3 satellites and is available from GlobWave within about 3 hours of the actual observation from space.

To aid widespread usage, all GlobWave data streams come in the same format, which is netCDF-3 with CF (v1.4) conventions. Each product file will contain relevant wave related parameters from the native product plus additional quality information such as various error characteristics and ancillary data such as and , wind speed and direction.

In addition, GlobWave provides comparisons with in situ data for the purposes of calibration and validation of the satellite data. Such in situ data can be from buoys networks around the globe including POSIEDON, UK Met Office, Météo-France, NODC and CDIP.

Another core objective is to demonstrate new types of satellite wave data products, such those based on new retrieval techniques, new types of satellite data, merged data from different sensors, or combinations of model and satellite data . Finally, GlobWave is focused on developing and trailing a pilot facility, following the JCOMM Expert Team on Wind Waves and Storm Surges recommendations, to permit operational agencies to routinely spatially compare their wave models with satellite wave data.

3.2 Online Tool for Satellite verses In Situ Matchup Database

In February 2012, GlobWave updated their online query tool, which contains a powerful search engine to find buoy / satellite overlaps based on a range of criteria including sensor / measurement, date, depth and distance to shore. Recent updates enable the display of ancillary data such as wind fields, storm paths and currents. The tool also contains extensive visualisation tools to help understand the relationships between data sets. Such visualisations include time series, histograms, directional histograms, scatterplots and time / frequency plots. New display features include display along a section, virtual buoys, intercomparisons and areal statistics. To access the tool and for further details on its capabilities please visit the GlobWave portal as shown in Figure 1.

Figure 1: GlobWave Online Tool

3.3 Error Characterisation Analysis: Accuracy of Satellite Data

Satellite data quality is tested by comparison with both in situ data and other satellite streams. Quality is recorded in two types of reports:

I. An Annual Quality Control Report which uses delayed mode GlobWave data to perform crossover analysis and comparisons with in situ buoys. It also performs interesting comparisons between delayed mode and near real-time significant measurements for Envisat, Jason-1 and Jason-2. II. Four Quarterly Quality Control Reports are also produced each year. Whereas the Annual Report focuses on delayed mode data, these analyse near real-time data and are made available on the portal shortly after each quarter. They contain a summary by month of the quality levels of the GlobWave near real-time data sets. 3.4 GlobWave Data - Statistics Areas

The physical quantities measured namely altimeter () and SAR (Swell wave height, dominant swell direction and dominant swell wavelength) from different satellites over different regions have been compared. The Global Wave Statistics (GWS) have been calculated over selected areas representing all the major ocean basins, and being of a size suitable for meaningful statistical analysis (typically 10 degrees latitude by 20 degrees longitude). The selected areas include regions of high wave activity such as the North Atlantic and Southern Ocean, as demonstrated in Figure 2.

Figure 2: Areas selected for GWS analysis

3.5 Altimetry Statistics Comparisons

This section gives the Global Wave Statistics (GWS) comparisons for altimeter Significant Wave Height (SWH) measurements in selected areas for appropriate time periods and satellite combinations. The aim of this study was not to produce useful products, as the areas are too large with too much spatial variability, but rather to produce statistics from individual altimeters and to investigate differences between them. To achieve this, the GWS have been calculated for the following time periods and satellite combinations as shown in Table 1:

Time period Satellites 01/01/1996 – 31/12/2002 ERS-2, TOPEX/Poseidon 01/01/2003 – 31/12/2008 Envisat, Jason-1, GFO

Table 1: Time Period and Satellite Combinations

The time periods were chosen as the maximum number of whole years when the satellite combinations were all operating at the same time, and are considered long enough to provide meaningful comparisons. The more recent Jason-2 time series (which could also be compared with Jason-1 and Envisat) is not yet long enough for more than a one or two-year analysis, and as such the results would be affected by the sampling variability. For each area the analysis has been carried out for satellite pairs over the whole time period. For region 1 a monthly analysis has also been performed to look at seasonal variations.

Quantile-Quantile plots: The main means of analysis used for the altimeter statistics comparisons is a Quantile-Quantile (QQ) plot, which compares the percentile values of SWH from each altimeter based on all data for an area and time period. Selected key percentiles are marked as red dots on the plots.

An example comparison of data together with the QQ plot is given in Figure 3, and the same comparison with the histogram density plotted on a log scale is given in Figure 4. The normal histogram shows very similar distributions estimated from the two altimeters. The log scale histogram allows more detailed comparison at higher wave heights and shows that GFO measured slightly more high waves than Envisat. This is visible in the QQ plot above the 0.95 quantile. The “wandering” QQ- plot at the highest SWH values (at around 12m) is no doubt due to sampling variability - note that although there are about 600k records from each altimeter the values from each transect of the area (at 6 – 7 km intervals) will have considerable correlation.

The QQ plots show the same information as the histogram comparisons but seem easier to interpret; hence these are favoured for presenting the comparison results. Examples of histograms for each area are also given in the overall results in order to compare the different ocean regions.

Figure 3: Histogram and QQ plot for Significant Wave Height (SWH) from Envisat and GFO in area 2.

Figure 4: Histogram with log scale and QQ plot for Significant Wave Height (SWH) from Envisat and GFO in area 2.

3.6 SAR Wave Statistics

Statistical occurrence of cross seas is an important parameter that can be extracted from SAR wave directional spectra which could help minimize hazards encountered by maritime users. The GWS from Envisat SAR have been calculated for the years 2003 to 2010 inclusive, with statistics produced both globally and for each GWS region. Further analysis was also carried out to investigate seasonal variations. Note that Envisat ASAR wave mode is one of a number of self-exclusive acquisitions modes and Wave mode acquisition is a background mission and only collected after other priorities, such as commercial data requests or operational oil spill monitoring, are taken into consideration.

SAR Wave statistics can be used to locate cross seas, where two sets of waves travelling from different directions cross at an angle of 45° or more. This creates very steep short-crested waves that can be dangerous for shipping [3]. A cross sea is estimated by examining the occurrence of several simultaneous wave partitions within each individual SAR wave spectra and only identifying those that have significant dominant direction difference (over 45 degrees) and sufficient energy (over either 0.5 or 1m swell SWH). The global occurrences of these two classes of cross seas are displayed in Figure 5 and Figure 6.

Figure 5: Percentage of Cross Sea Occurrence 2003 - 2010. White line indicates the maximum extent of Antarctic sea ice

Figure 6: Percentage of Cross Sea Occurrence 2003 - 2010. White line indicates the maximum extent of Antarctic sea ice

As the figures show, mild cross seas are quite frequent in general and occur mostly in the eastern equatorial Pacific, where distant swells from southern and northern mid latitude storms overlap. Moderate to high cross seas seems to happen in more specific regions such as the southern Pacific but also in the North Sea, Agulhas current region and the coast of south west Australia. Swell crossing behind large islands can occur very regionally such as east of Shetlands or the Kerguelen Islands. 3.7 Pilot Extension to the JCOMM Wave Forecast Verification Scheme (WFVS)

The Pilot Extension to the WFVS has now been operational for over a year. Reports showing intercomparisons between wave models and satellite observations are automatically generated daily and monthly for meteorological centres, namely the UK Met Office, SHOM / PREVIMER, FNMOC, NOAA-NCEP, ECMWF, Puertos del Estado and BOM. Work is currently being undertaken to incorporate other participating centres. These reports highlight discrepancies between the different models and satellite observations, as well as showing any systematic issues with forecasts for the model domains. The following plots (Figure 7) show a specific region in the north Atlantic, with the normalised mean bias and the scatter index of the SWH from the UKMO model verses the satellite altimeter (Ku band) SWH values during June.

Figure 7: The normalised mean bias and the scatter index plot of a specific region in the north Atlantic

The reports are configured for each participant, with participants able to elect which sections of the report they wish to have generated. Specific regional plots are being included for some participants. All the generated reports contain links to where the figures can be viewed online and each figure, together with its underlying data, can be downloaded in a range of formats. Users are also able to generate specific figures online to investigate particular time periods or diagnostics, even if they are not included in the regular reports.

4. USERS OF GLOBWAVE DATA

The GlobWave project has been widely utilised by the wave community. The GlobWave portal, which contains information on both the project and the wider wave community is very popular, with an average of 4500 hits per month. The easy and free access to the data has been very popular, with a record 2,200,000 files being downloaded in February 2012, prior to the Envisat failure. There is currently in excess of 170 registered GlobWave users, who download and use the data in a wide range of research and operational projects. For example, the Royal Australian Navy and the New Zealand Defence Force has used GlobWave data to examine the climatic and oceanographic conditions that accompanied attacks by pirates in the Horn of Africa region between 2010 and 2011. Due to the security situation in this region, no buoy data was available in the region off Somalia where piracy is most common. Satellite observations where used to provide the oceanographic information required. By examining SWH and wind speed measurements within the GlobWave data, the researchers found a strong relationship between successful pirate activity and wave height and wind speed [4].

A study by OceanWeather Inc used GlobWave global data to identify extreme sea states. This study found that GlobWave’s extensive data quality information, the homogenized dataset and quality control to be particularly useful and identified a significantly larger number of extreme waves (on 12 m or higher) than a previous study using individual data streams [5]. Finally, as GlobWave collated provides relatively new data types, it can be used to develop new types of information, such as using SAR to locate cross seas, as described earlier.

5. RESULTS AND CONCLUSION

The GlobWave project has shown that the oceanographic community has a need for wave data and by providing such data in a standard format and collating all data together, they will use ocean wave data more regularly. Given below are testimonials received from the users of GlobWave data in various projects.

Laure Grignon from HR Wallingford said: “GlobWave has provided additional data to validate offshore conditions estimated from global and regional wind and wave models, representative of an average in space and time rather than an instantaneous measure of the wind or wave height. Due to the success of these results, we will continue to use GlobWave data to calibrate and validate global and regional wind and wave models.”

Christian Appendini, from National Autonomous University of Mexico said: “We have done a 30-year wave hindcast for the Gulf of Mexico and Caribbean Sea, which we have validated with buoys, but this data is sparse in time and space, so we wanted to use the GlobWave data for such purposes. As this is our first time using satellite data it has taken us some time to get what we need. On the other hand, if GlobWave data is unavailable we wouldn't have the opportunity to use satellite data.”

Every effort has been made by the GlobWave study to make our satellite wave data: – Easy to locate – Easy to retrieve – Easy to use – Of known quality

It is quite evident from this study that satellite data has potential applications in various fields both research and commercial sectors and having these types of data in one place (“One stop shop”) with “data fit for purpose” quality will encourage the uptake of satellite data by end users. With initiative like GlobWave, the goal to increase the number of users of satellite wave data is not too far to achieve.

6. REFERENCES

[1] I. S. Robinson, J.-F. Piollé, P. Leborgne, C. Donlon, and O. Arino, “MEDSPIRATION: A European Contribution to the Global Ocean Data Assimilation Experiment High Resolution Sea Surface Temperature Pilot Project,” Proceedings of the MERIS (A)ATSR Workshop 2005 (ESA SP-597). 26 - 30 September 2005 ESRIN, Frascati, Italy, ESRIN, Rome, pp 16.1, 2005.

[2] G. Busswell, E. Ash, J.F. Piolle, D. Poulter, H. Snaith, F. Collard, H. Sheera, and S. Pinnock, “Project GlobWave,” Proceedings of the Third International Workshop SeaSAR 2010, held at ESA ESRIN (Frascati, Italy), 25 – 29 January 2010. ESRIN, Rome, 2010.

[3] ESA, The world’s now in one place, ESA, http://www.esa.int/esaEO/SEMIGIRHPOG_index_0.html, 2011.

[4] D.E. Cook, S. Garrett and M. Rutherford, “Climatic controls on piracy in the Horn of Africa region, 2010 – 2011,” Proceedings of the 19th International Congress of Biometeorology, 4 – 8 December 2011, Auckland, New Zealand, Paper 00380, ISBN 978-0-86869-132, 1 – 6 pp, 2011.

[5] V. Cardone, “Critical issues for the specification of unbiased and homogeneous marine surface wind reanalyses”, Proceedings of the Third International Workshop on Advances In the use of Historic Marine Climate Data (MARCDAT-III, Frascati, Italy, 2 – 6 May 2011), JCOMM Technical Report No. 59, TSR Workshop 2005 (ESA SP-597). 26 - 30 September 2005 ESRIN, Frascati, Italy, ESRIN, Rome, 2011.