Fiducial Reference Measurements for Altimetry

This work defines the concept of Fiducial Reference Measurements (FRM) for altimetry calibration. It has emerged out of the requirement for reliable, consistent, standardized and undisputable Earth observation records. FRM observations are to become comparable worldwide, insensitive to instrument, satellite, setting, location and measurement conditions among others. At first, the paper singles out the various contributing error sources to the calibration error budget in altimetry. Secondly, it evaluates measurement uncertainty as it originates from each constituent in the error propagation chain. Third, it sets the foundation for connecting measurement uncertainty so that it could be traceable to international metrology standards (speed of light, atomic time, for example). Finally, this paper presents procedures, protocols and best practices for arriving at FRM standards in sea-surface but also in transponder calibration. This analysis is based on experience gained over the last 15 years of operation of the permanent facility for altimetry calibration in west Crete, Greece.

The international timekeeping community prepares several official timescales, and there exist a plethora of alternatives used by others, often with no official recognition. The advantages and disadvantage of several official timescales are presented; however, they are all interconvertible using information readily available on the internet. A discussion of traceability and the statistical errors is included; all of these scales are ultimately traceable to UTC. Many of the considerations are not relevant if the error tolerance is above a few tens of nanoseconds.

Synchronisation of a remote clock to a time reference can be challenging. Within the timing community these challenges have been addressed, and robust time transfer and calibration techniques have been developed offering differing levels of synchronization accuracy to the international time reference UTC, Coordinated Universal Time. These techniques can be applied to timing equipment at ground-based Cal/Val (Calibration/Validation) sites in Western Crete and elsewhere to achieve FRM (fiducial reference measurements) for altimetry, satisfying their requirement for SI (International System of units) traceability. Continuous monitoring of the remote sites is required to maintain traceability to the reference time, and a holdover clock may also be needed. This paper discusses how UTC or TAI (International Atomic Time) could be used as a time reference for timestamped measurements taken at Cal/Val sites, improving measurement uncertainty and linking fiducial reference measurements for satellite altimetry back to the SI unit of time: the second.

By convention the absolute bias in sea surface height (SSH) is the difference between the altimeter and the in-situ reference SSH heights above the Earth ellipsoid. Both the absolute and the relative bias of the CryoSat-2 and Sentinel-3A missions are derived in this study at four stations along the German coasts.

Firstly, the coastal data processed in Delay Doppler altimeter (DDA) mode, also called SAR mode (SARM), are shown to be less noisy than data in pseudo-low resolution mode (PLRM), which is comparable to the conventional low-resolution mode (LRM). The best agreement with in-situ data is reached by the SARM data retracked with the SAMOSA+ coastal retracker (hereafter SAR/SAMOSA+) from the ESA GPOD SARvatore service.

Secondly the absolute bias and its standard deviation are computed for each mission and product type. The mean mission absolute bias depends on location and altimeter product. Both the absolute and relative biases are small and the standard deviation is smaller than 4 cm and larger than the bias. Departures between absolute biases evaluated at different stations are possibly related to geoid inaccuracy in the coastal zone. Finally, the smaller standard deviation of the bias time series confirms that SAR altimetry is more accurate than PLRM, the minimum standard deviation is 2 cm.

Performance of the Sentinel-3A SRAL altimeter data is evaluated in the coastal waters of the Gulf of Finland, Baltic Sea. Sea Surface Heights (SSH) were computed along three ascending passes nearby three tide gauge (TG) sites in Estonia. These SSH were compared to a high-resolution marine geoid model in conjunction with a TG corrected hydrodynamic model (HDM). The SAMOSA2 and SAMOSA+ retracker retrieved SSH values were inter-compared as well. The quality assessment yielded the root mean square error of 115 and 99 mm for the SAMOSA2 and SAMOSA+ retrieved Sentinel-3A SSH data, respectively. The near-zero mean of discrepancies shows that there is no significant systematic bias between the geodetic infrastructure and Sentinel-3A SSH data.

This work presents the latest absolute and relative calibrations for the altimeters of the Sentinel-3A and Jason-3 at the Permanent Facility for Altimeter Calibration in west Crete, Greece. Results have been determined at first with the transponder at the CDN1 Cal/Val site on the mountains of west Crete using the ascending Sentinel-3A Pass No. 14 and the descending Jason-3 Pass No.18. Then, sea-surface calibration has been carried out with the descending Sentinel-3A Pass No. 335 and the ascending Jason-3 Pass No.109 based on the Cal/Val facility on Gavdos island. For Sentinel-3A results have been established for cycles 3–27 using Level 2 (Level 0 for transponder) and Non-Time Critical data. For Jason-3, cycles 5–80 and the S-GDR-D data have been worked with the transponder calibration, while for the results with sea-surface calibration, cycles 1–80 with the I-GDR-D data have been implemented.

Sentinel-3A produces biases of the order of a few mm either with the transponder (+2.7 mm) or with the sea-surface calibration (7.3 mm and −4.4 mm). The altimeter of Jason-3 presents a range bias at +22.7 mm (No.18) with the transponder and −36.7 mm for the bias in sea-surface height, respectively. Finally, comparison of sea-surface heights observed by Sentinel-3A relative to Jason-3 demonstrates a difference of +4 cm within a period of ±3 days about 20 km south of Gavdos.

The inter-mission cross-calibration is a basic prerequisite for long-term sea level change studies on all spatial scales. Especially, for climate studies the consistent combination of successive missions is essential. This study uses a global multi-mission crossover analysis in order to investigate the performance of the Copernicus Sentinel-3A altimetry mission and its consistency to consecutive missions such as Jason-3. The first 1.5 years of data show an inter-mission bias of 3.6 cm with respect to Jason-3 with a linear trend of 4.0 mm/year. When using Pseudo Low Resolution Mode (PLRM) instead of Delay-Doppler data the bias increases whereas the trend decreases. Given the short time period under investigation these numbers should not be overrated, however, a careful future monitoring is necessary.

The quality of existing bathymetry models for the Arctic Ocean is evaluated through visual comparison and the response of modelled tides. The high resolution ArcTide 2017 hydrodynamic model was used to evaluate the bathymetry in selected shallow water regions where tides are significant. The Southern Barents Sea was identified as a problematic region where inconsistencies were identified, resulting from methods used to patch in regional models and incorrect definitions of coastlines and depths. More generally, the investigation shows that careful verifications are needed to ensure seamless transitions between bathymetry datasets.

More accurate bathymetry in the Arctic Ocean is needed and we outline the development of a new Arctic bathymetry using bathymetry inversion which uses a combination of the existing Arctic bathymetry and topography inverted from a band-pass filtered version of the most recent DTU17 gravity field. We also illustrate the regions where we find adequate spatial correlation to perform such inversion.

A preliminary analysis of sea level (SL) changes around Venice from three tide gauges (TGs) (one off-shore: AAPTF, one at the coast: DSL, and one inside the lagoon: PS) to characterize the variability during 1993–2015 and relative SL trends, is provided. As no global positioning system (GPS) data covering the same period was available to the authors for the three tide gauges, the analysis is restricted to changes relative to the land. Monthly SL means from the European Space Agency (ESA) Sea Level Climate Change Initiative (CCI) altimeter-derived product are also used. A comparison between the monthly mean time series of CCI and AAPTF has been performed using the nearest CCI grid point to the location of AAPTF: the centered Root Mean Square Difference (cRMSD: the root mean square difference of the two series with the respective means removed) resulted 6.33 cm, while the Pearson’s linear correlation reached 0.75. Much higher agreement was found, as expected, between the monthly mean records of AAPTF and PS TGs: the cRMSD was 1.03 cm, and the linear correlation 0.99. We obtained a trend of 6.65 mm year⁻¹ at AAPTF over the Satellite Altimetry (SA) era (1993–2015). A smaller trend has been found here from altimetry (4.25 mm year⁻¹). The differences might be explained in terms of Vertical Land Motion (VLM) which was not accounted for in the TGs time series, to the different processing of TGs and altimeter data (in the altimeter signal the Dynamic Atmospheric Correction is removed), and/or uncertainties in this area due to the current CCI product that is based on open ocean altimetry. In general, the altimetry trends derived from CCI are spatially higher in the Adriatic Sea than Global Mean Sea Level (GMSL) in most of the region, with greater values than average in Venice. Reprocessing of along-track altimeter data sets with consistent coastal processing for all missions is expected to enhance SL accuracy and with a better refining of raw trends. The SA era is too short to delineate or discuss an affordable climatology, as decadal SL variations cannot be accounted for by such a short time series, nonetheless the rates calculated in this study are fruitfully compared each other and with those derived by SA.

The Copernicus POD (Precise Orbit Determination) Service is part of the Copernicus PDGS Ground Segment of the Sentinel-1, -2, and -3 missions. It is responsible of generating precise orbital products and auxiliary data files for their use as part of the respective PDGS processing chains.

For Sentinel-3, the CPOD (Copernicus Precise Orbit Determination) Service generates three types of products: the official Near Real Time (NRT) product, the Short Time Critical (STC) and the Non-Time Critical (NTC). The STC and NTC orbit products are generated as backup of the primary orbit products from CNES (Centre National d’Etudes Spatiales). The accuracy requirements (in the radial direction) are 10, 4 and 3 cm, respectively, but with the goal of achieving 8, 3 and 2 cm, respectively. Actual accuracies for Sentinel-3A are significantly better than the requirements and goals, as it will be shown in this paper.

Considering the importance of the Sentinel-3 orbit products for the radar altimetry processing the orbit validation is crucial. The validation of the different Sentinel-3 orbital products from the Copernicus POD Service, therefore, consists of several independent steps including orbit overlap analysis, direct orbit comparison, but also cross-validation with SLR and DORIS measurements. The different orbit validation steps are described and results are shown for the entire mission time until March 2018 of Sentinel-3A.

The most recent released global marine gravity field from DTU Space takes into account the new SARAL/AltiKa geodetic mission initiated in 2016 along with new improved Arctic processing of the Cryosat-2 mission. With its 369 days repeat cycle, Cryosat-2 provides one repeat of geodetic mission data with 8 km global resolution each year since its launch in 2010. Together with the Jason-1 end-of-life geodetic mission in 2012 and 2013, we now have more than five times as many geodetic missions sea surface height observations compared with the old ERS-1 and Geosat geodetic missions.

The DTU17 has been derived focusing on improving the coastal and Arctic gravity field, enhancing the shorter wavelength of the gravity field (10–15 km). For DTU17, we find a substantial improvement in marine gravity mapping as shown through comparison with high quality airborne data flown north of Greenland in 2009.

The new Synthetic Aperture Radar (SAR) data from the Sentinel-3 satellites will provide the community with valuable new information in coastal areas and in the Arctic, due to the higher along-track resolution obtained through the Delay-Doppler processing. The SAR data also allows for a more detailed study of the ocean surface, since these make small-scale variations visible. Combined, data from the Sentinel-3 satellites creates a tremendous possibility for improving tidal models and mean sea surfaces near the coast, where these models are currently using extrapolation to provide information. However, some challenges are also becoming more apparent in areas where satellite altimetry have not previously been available. Such as the discrepancies between tidal models near the coast, which are amplified because the Sentinel-3 satellites fly in a sun-synchronous orbit. Acquiring satellite altimetry data in coastal and sea ice covered areas also highlights some issues with the current wet tropospheric correction, calculated from measurements by the on-board microwave radiometer, leading us to the conclusion that it is safest to use a WTC from a model – at least in coastal and sea ice prone areas.

Through both atmospheric and oceanic circulation, heat is transferred between the equator and the poles. Possible ways in which the Arctic ecological systems can be affected by warmer temperatures include: changes in amount and duration of snow and ice cover; frequency and extent of spring floods; changes in the ratio of precipitation minus evapotranspiration; amounts of water transport from lakes and rivers from snow and permafrost melting; and a decrease in frozen precipitation. A key component in transferring heat is through freshwater exchange in and out of the Arctic. Hence, accurate mapping freshwater fluxes and potentially its changes with time is vital to describe the heat transfer and its possible temporal changes.

Our results demonstrate how ESA’s Earth Observation data together with in-situ measurements can be used to improve the mapping of the major Arctic Ocean freshwater fluxes. In this paper, we outline how four of the five major freshwater fluxes can be determined using present day Earth Observation data exclusively. These are: discharge from rivers; inflow through ice and melt run off; outflow of freshwater in sea ice; and in/outflow of freshwater through ocean currents. We subsequently present key finding and estimates of these four freshwater fluxes and compare our results with estimates based on in-situ data provided through previous studies.

This work provides the essential elements for a scientific and operational roadmap with guidelines and practical directions to calibrate satellite altimeters under a new established standard of Fiducial Reference Measurements for Altimetry. According to this new principle, ground facilities, instrumentation and procedures, set up for calibration and validation (Cal/Val) of observations and products in altimetry, shall follow well-documented procedures and protocols, transparent to all involved. It shall also deliver uncertainty budgets of the Cal/Val results which should be built upon metrological standards and capable of being traced to Système International units. At first, this paper describes some guidelines for establishing new facilities for satellite altimetry Cal/Val with respect to their geographical location. Secondly, it gives requirements for maintaining an efficient performance and functionality for the Cal/Val facility. Third, it presents the optimal design and setup for the facility’s instrumentation in relation to the Cal/Val technique employed. Finally, this work recommends in-situ data set formats.