Towards a swath-to-swath sea-ice drift product for the Copernicus Imaging Microwave Radiometer mission

Abstract. Across spatial and temporal scales, sea-ice motion has implications on ship navigation, the sea-ice thickness distribution, sea ice export to lower latitudes and re-circulation in the polar seas, among others. Satellite remote sensing is an effective way to monitor sea-ice drift globally and daily, especially using the wide swaths of passive microwave missions. Since the late 1990s, many algorithms and products have been developed for this task. Here, we investigate how processing sea-ice drift vectors from the intersection of individual swaths of the Advanced Microwave Scanning Radiometer 2 (AMSR2) mission compares to today’s status-quo (processing from daily averaged maps of brightness temperature). We document that the “swath-to-swath” (S2S) approach results in many more (two orders of magnitude) sea-ice drift vectors than the “daily-maps” (DM) approach. These S2S vectors also validate better when compared to trajectories of on-ice drifters. For example, the RMSE of the 24 hour Arctic sea-ice drift is 0.9 km for S2S vectors, and 1.3 km for DM vectors from the 36.5 GHz imagery of AMSR2. Through a series of experiments with actual AMSR2 data and simulated Copernicus Imaging Microwave Radiometer (CIMR) data, we study the impact that geo-location uncertainty and imaging resolution have on the accuracy of the sea-ice drift vectors. We conclude by recommending that a “swath-to-swath” approach is adopted for the future operational Level-2 sea-ice drift product of the CIMR mission. We outline some potential next steps towards further improving the algorithms, and making the user community ready to fully take advantage of such a product.

radiometric correction and better geolocation does not significantly improve the comparison with buoys in the Arctic [with respect to using daily maps, our addition] and identify as future work to document whether orbital data, rather than the 24hour averaged TBs used by each group, offers significant improvements.
Here, we report on such a study that was conducted in the context of the design phase for a future satellite mission: the Copernicus Imaging Microwave Radiometer (CIMR). CIMR is a conically-scanning microwave radiometer mission under study at the European Space Agency for the Expansion phase (2026)(2027)(2028)(2029)(2030) of the European Union's Copernicus Space Component. At time of writing, the reference document for the CIMR mission is the Mission Requirement Document (Donlon et al., 2020).
The aim of our research is thus two-fold: 1) to document the pros and cons of processing sea-ice drift vectors from individual orbits vs from daily averaged maps, and 2) to discuss the implications for future sea-ice motion capabilities of the CIMR mission. This paper is structured as follows: satellite and in-situ data are introduced in Sect. 2, while the methodologies for sea-ice motion tracking and product validation are covered in Sect. 3. Sect. 4 documents our results, Sect. 5 covers a discussion in the context of the CIMR mission, and we conclude in Sect. 6.

Orbit-based brightness temperature data
We accessed Level-1b brightness temperature (TB) data (Version 2 calibration) of the Global Change Observation Mission 1st-Water (GCOM-W1) Advanced Microwave Scanning Radiometer 2 (AMSR2) directly from the Japan Aerospace Exploration Agency (JAXA). For this study, we used the brightness temperatures at both Vertical and Horizontal polarizations of the Ka (36.5 GHz) and W (89 GHz) imagery channels.   freezing conditions in both regions, and to maximize the number of available on-ice drifters available for validation, as covered in the next section.

GPS trajectories of on-ice drifters
To validate satellite-based sea-ice drift vectors, we access GPS trajectories for on-ice drifters in the Arctic and Antarctic ( Fig. 1). In the Arctic, buoys are in the Beaufort Gyre, and in the Transpolar Drift. In the Antarctic, all buoys are in the central Weddell Sea. The colors in Fig. 1 represent the time of the observation records within the two 3-months periods. As expected, sea-ice motion is faster in the Weddell Sea than in the Arctic, and the trajectories are more elongated.
A variety of buoy types are included in our validation data, but we are only concerned with three pieces of information per trajectory record: timestamp, latitude, and longitude. Most buoys report positions on an hourly basis. The Ice-Tethered Profiler data were collected and made available by the Ice-Tethered Profiler Program (Toole et al., 2010;Krishfield et al., 2008) based at the Woods Hole Oceanographic Institution (http://www.whoi.edu/itp, last accessed 1 st June 2020). A variety of other buoys were accessed from the data portal meereisportal.de (Grosfeld et al., 2016, last accessed 1 st June 2020), including all Antarctic buoys, and the buoys deployed at and around the Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) site (Krumpen et al. 2020).

Sea Ice Concentration data from EUMETSAT OSI SAF
The Sea Ice Concentration (SIC) data from European Organization for the Exploitation of Meteorological Satellites (EUMETSAT) Ocean and Sea Ice Satellite Application Facility (OSI SAF) was accessed and transformed into an ice/no-ice mask using a threshold of 40% SIC: grid cells with a SIC above 40% are considered as ice-covered, and can be used for the computation of sea-ice drift vectors. The Interim Climate Data Record "v2" based on Special Sensor Microwave Imager Sounder (SSMIS) data is used . Algorithms and processing chains are described in Lavergne et al. (2019).

Sea-ice motion tracking
The sea-ice motion tracking methodology implemented here, including the quality control steps, is very similar to that described by Lavergne et al. (2010) and implemented in the operational chains of the EUMETSAT OSI SAF (Lavergne et al., 2016). We recall below two unique features of this processing chain.
First, it implements the Continuous Maximum Cross-Correlation (CMCC) motion tracking algorithm. The CMCC stems from the well-known MCC (Ninnis et al. 1986) but implements a continuous optimization of the crosscorrelation function (rather than a discrete optimization in MCC). The continuity is enabled by on-the-fly linear interpolation of image pixels. This continuous optimization strongly reduces the "quantization noise" present in many MCC-based sea-ice drift products. Lavergne et al. (2010) documented how the CMCC-based ice motion vectors were more accurate than those based on MCC (see also Hwang 2013, Sumata et al. 2014).
Second, it implements an iterative quality control step to detect and correct a few questionable (aka "rogue") vectors. Those vectors are at the maximum of the cross-correlation function, but point in a direction completely different from the neighbouring vectors. All block-based motion tracking algorithms need such a quality control step (e.g. Girard-Ardhuin and Haarpaintner, 2006;Tschudi et al. 2020), but most authors remove the rogue vectors and the vector field has missing data cells. The quality control step of Lavergne et al. (2010) both detects the questionable vectors and -most of the time-corrects them, reducing the occurrence of data gaps.
As in Lavergne et al. (2010) and in the EUMETSAT OSI SAF sea-ice drift product (Lavergne et al. 2016), we process seaice drift vectors with a grid spacing of 62.5 km on two polar stereographic grids (Arctic and Antarctic).

"swath-to-swath" and "daily maps" sea-ice drift products
The sea-ice drift product run at the EUMETSAT OSI SAF based on the algorithms of Lavergne et al. (2010) is processed from daily gridded maps of satellite signal: daily averaged maps of TB from the JAXA AMSR2 and United States Department of Defence (DoD) SSMIS, and daily averaged backscatter coefficients (corrected to 40° incidence angle) of the EUMETSAT Advanced SCATterometer (ASCAT). These daily averaged maps aggregate satellite signal from 00 UTC to 23:59 UTC for a given day, and their valid time is 12 UTC. We name this type of sea-ice drift product a "daily maps" product (noted DM).
It is noteworthy that, in addition to averaged maps of satellite signal from which DM sea-ice drift vectors are computed, we prepare the corresponding maps of daily averaged satellite sensing time: the average of the observation time from all the swaths during the 24 hours period of the daily averaging ( Fig. 2 in Lavergne et al. 2016). This mean sensing time is an important information attached to each DM sea-ice drift vector: even though they are from daily maps with valid time 12 UTC, the DM vectors will be used and validated taking into account these space varying mean start and stop times.
For this study, we prepared both DM products and "swath-to-swath" products (noted S2S Hemisphere (see Sect. 2.1). In both 3-month periods, we process sea-ice drift vectors with durations (aka time-span) ranging from ~100 minutes (the separation time between two consecutive orbits) to 48 hours. The duration of an ice drift vector is the difference between the timestamps of the start and stop image from which motion is estimated (these images are either individual swaths for S2S products or daily maps of DM products).

Collocation of satellite and in-situ drift vectors
Comparison between any two data sets first requires deciding on a collocation strategy, in space and time. To collocate seaice motion vectors is somewhat different from collocating other geophysical variables, in that the reference data source (a timeseries of GPS records from the ice drifting buoy or ship) is not directly comparable to what the satellite product measures (a net Lagrangian displacement vector over a time period). For this validation exercise, we follow the approach of Lavergne et al. (2010), Hwang (2013) and others to compute equivalent net Lagrangian displacement vectors from the in-situ trajectories as part of the collocation step. In short, for each satellite vector to be validated, the in-situ record closest to the start position of the satellite vector is selected and -if within a suitable geographical radius-the in-situ net Lagrangian displacement vector is computed from its GPS records, using the start and end time of the satellite vector. We rely on a nearest neighbor approach in both space and time domains (nearest position, nearest time) without interpolation.
For a collocation pair to enter in the collocated matchup database, the start time of the in-situ displacement has to be within plus or minus 3 hours of the start time of the nearest satellite drift vector, the duration of the in-situ vector has to be within plus or minus 1 hour of that of the satellite vector, and the distance between the start positions of the in-situ and satellite vectors has to be less than 30 km. To avoid over-representation in the case of buoy clustering (e.g. buoy arrays or the MOSAiC site) only the closest buoy to a satellite vector is kept. In addition, the collocated matchup database is filtered so that no directly neighbouring satellite vectors co-exist in it, in order to reduce the effects of the correlation lengths stemming from the satellite retrieval.
For the S2S product, the start and end time of the drift vectors are the time stamp of the individual satellite swaths. For the DM products, they are the space-varying mean observation (overflight) times at the start and end images, meaning that the satellite product is not used as if starting and stopping at 12 UTC everywhere in the product grid.

Simulation of CIMR orbits and swaths
The swath of CIMR (>1900 km) will be larger than that of AMSR2 (1450 km). To study the impact of a wider swath on the characteristics of a future S2S sea-ice drift product from CIMR, we simulate some of its orbits and swaths. CIMR is to fly along a sun-synchronous dawn-dusk orbit with a 98.7° inclination. Additional orbit and instrument parameters relevant to the simulations are in Donlon et al. (2020). propagation is performed using the mean keplerian orbit propagation mode available in the orbit library. The spacecraft attitude is modelled applying local normal pointing and yaw steering law, and the instrument swath edges are defined in terms of direction look angles. Finally the zone visibility functions were used to compute the coverage mask of the instrument swath over a Northern and a Southern Hemisphere grid. The visibility library includes functions to calculate the intersection of the swath points with the Earth ellipsoid, which are invoked internally by the zone visibility routine.
The coverage mask (binary map) of individual swath is then used to mask a daily-averaged map of AMSR2 TB to create a simulated CIMR TB swath, that enters the sea-ice motion software. Our simulated CIMR swaths are thus only for studying the impact of the wider swaths, not the better spatial resolution.

Comparative space/time coverage of the DM and S2S vectors
Given the full daily coverage of GCOM-W1 AMSR2 (except the observation gap at the pole in the Arctic), the number of potential DM sea-ice drift vectors is a function of the sea-ice extent only. Conversely, the number of potential S2S sea-ice drift vectors varies with sea-ice extent and with the area of overlap between the individual swaths. This area of overlap depends on the relative orientation of two swaths, thus on the time difference between the orbits. Consecutive orbits will allow a larger overlap than, e.g. orbits that are 4-8 hours apart. Due to the orbit configuration of the GCOM-W1 mission, two orbits that are roughly 24 hours apart will have large overlaps.  In NH, the majority of S2S vectors are computed in a band between 75°N and 80°N (up to 90 S2S drift vectors per grid cell), and slowly decays towards 85°N with spatial patterns that are typical of satellite swath geometry. The sharp decay poleward of 85°N is due to the polar observation hole of the AMSR2 instrument, with a width of ~ 0.5° (see Fig. 2, left panel). Two factors amplify this decay. First, there is less overlap between swaths at very high latitude (edges of the swaths). Second, the motion tracking algorithm works with sub-windows (e.g. 11x11 image pixels) and this limits the retrieval of drift vectors in the immediate vicinity of areas with image missing data (e.g. the polar observation hole, sea-ice edge, coastal region, etc…).
In fact not a single S2S vector (nor DM vector) are retrieved north for 88.5°N from the GCOM-W1 AMSR2 data. The transition from ~90 S2S drift vectors per grid cell to 40 and less equatorward is visible on both NH and SH maps. In SH, most of the sea-ice cover is south of 75°S, which is the case all year round. Still, even at the outskirts of the SH sea-ice cover, we have 5-15 S2S vectors -compared to a single DM vector -per grid cell. The numbers above pertain to the configuration where drift vectors are processed at a grid spacing of 62.5 km: reducing the grid spacing would directly increase the number of DM and S2S vectors.
A first advantage of adopting a S2S ice drift processing for microwave radiometer satellite data is thus to take full advantage of the individual swaths, and get access to many more sea-ice drift vectors than in the DM configuration.
These S2S Fig. 4 should thus be seen as an illustration of a general pattern (repartition over the whole period, with peaks). The S2S duration histograms (Fig. 4, right panels) are also controlled by the characteristics of the orbit, swath width and the sea-ice cover. They document that a wide spectrum of drift durations are recorded by S2S drift products, with peaks near 0, 24, and 48 hours in the NH and more spread in the SH.
These peaks generally correspond to when the swaths overlap most.
All in all, the short analysis conducted here documents that S2S and DM ice-drift products have distinct characteristics. S2S ice-drift products offer many more vectors, and this number varies with latitude. The two types of products cover the 11 245 250 temporal domain differently. The S2S products have a broad but not homogeneous coverage in start times, stop times, and duration. Our next step is to investigate if the accuracy of the S2S ice-drift product is better, similar, or worse than that of DM ice-drift products. We assess this accuracy against collocated in-situ drifter GPS trajectories. Results are reported in the next section.

Validation of 24 hour drift vectors against buoy data
We collocate all S2S and DM sea-ice drift vectors with on-ice drifters trajectories (Sect. 3.3) for both 3-month periods in the Northern and Southern Hemispheres (Sect. 3.2). Here we analyse the statistics from this validation.   The validation statistics are somewhat worse in the SH than in the NH. This is probably due to the low number of validation data, and the fact that some are at the outskirts of the sea-ice cover, in dynamical regimes that are challenging for motion tracking from coarse resolution radiometry imagery (Fig. 1).
These numbers mean that a typical 24 h drift displacement vector can be measured with an uncertainty of typically 1 km in each component by an S2S product, even though the original imagery is at a rather coarse resolution (the -3dB footprint of the AMSR2 36.5 GHz channels is 7x12 km, see Table 1). As expected, the uncertainty of DM sea-ice drift vectors is larger (~30% increase). In the next section, we repeat the validation experiment but this time studying the full range of durations, from ~100 mn to 52 h.  Table 2 correspond to the conditions around the 24 h tick-mark.

Impact of geo-location accuracy of the imagery
Geo-location accuracy of the satellite images from which drift vectors are computed is a key component of the uncertainty budget. Even very high resolution satellite images (such as those from SAR) will give poor sea-ice drift vectors if the images are not correctly geo-located.
To simulate the impact of geo-location accuracy on the S2S drift vectors, we purposely mis-register the GCOM-W1 AMSR2   Geo-location accuracy is of importance for all satellite-based products, but will have an exacerbated impact on sea-ice drift products. Indeed, if both satellite images from which the drift vectors are computed have a geo-location error, and the two geo-location errors are in opposite directions, the drift vector will be strongly affected. Fig. 7 gives a schematic view of the impact of a constant geo-location error in the flight direction (which is very close to what we simulated in Fig. 6).

Figure 7: Impact of a constant geo-location error in the flight direction on S2S retrievals. Each panel corresponds to a different relative angle θr between the two orbits from which sea-ice drift vectors are estimated. The true drift vector is in black, the geolocation offsets in the first and second orbits are of the same length, and create an erroneous drift vector (gray). The errors in the dX and dY components are εdX and εdY.
In Fig. 7, we illustrate how the relative angle between two swaths sustaining the computation of an S2S vector impacts the magnitude of the retrieval error in the presence of a constant geo-location offset along the flight direction. The four panels correspond to four different relative angles. When the two swaths are close to parallel (θr ≈ 0), the geo-location error does not result in significant drift errors. This corresponds to time separation between the AMSR2 swaths of ~100 mn, ~24 h, Before closing this section on geo-location errors, we discuss the fact that the DM drift vectors seem much less affected by the artificial geo-location error introduced for Fig. 6 than the S2S vectors are (compare the error-bar symbols of Fig. 5 and Fig 6, and especially that the bias is zero in both). When building daily composite maps of averaged TB, the impact of the geo-location is smeared, but not canceled. Indeed, and especially at lower latitudes, locations on the sea-ice cover are often observed by sequences of subsequent orbits, which means a constant geo-location error from each orbit will still result in a non-zero mean geo-location error in the daily averaged map. Nevertheless, the bias in drift components of DM vectors is close to zero because of the repeat cycle of the observations of (in our case AMSR2) orbits: a location on the sea-ice cover will be observed on average roughly at the same time of the day one or two days apart. This repeat cycle of the satellite orbit ensures that the non-zero mean geo-location error is similar at all locations in the daily maps, thus that the impact on the DM drift vectors is small (θr ≈ 0).
All in all, we conclude from this section that S2S drift vectors are more sensitive to geo-location errors than DM drift vectors are, and that the retrieval of accurate sea-ice drift vectors from individual swaths puts stringent requirements on the geolocation accuracy of passive microwave missions if all swath overlap pairs should be processed. Even seemingly limited geo-location errors such as those simulated in this section will result in inaccurate (biased) estimates when orbits overlap in opposite flight directions.

Impact of the resolution and microwave frequency of the imagery
The key principle of motion tracking algorithms (including those for sea-ice drift) is to identify local intensity patterns on one image, and track it on another image. The accuracy of the motion vectors will thus be better if the satellite images offer intensity patterns that are clearly defined and stable with time. The sharpness and stability of satellite microwave images of the sea-ice surface depend both on the spatial resolution achieved by the FoV, and the microwave frequency and polarization of the imagery channel being processed. Indeed, channels (or instruments) with coarser spatial resolution will provide blurred images of the microwave emissions, with less or weaker intensity patterns to track. Channels with higher sensitivity to stable characteristics of the surface (e.g. sea-ice type, snow-depth, etc…) and/or lower sensitivity to the varying atmosphere above the sea ice (e.g. cloud liquid water path, air temperature) will offer sharper and more stable intensity patterns to track.
We investigate the impact of resolution and microwave frequency of the imagery by running an experiment where GCOM-W1 AMSR2 Ka (36.5 GHz) and W (89.0 GHz) imagery channels are first resampled on a pan-Arctic grid with spacing of 3.125 km. Such a fine spacing is an oversampling of the true resolution of both microwave channels (see Table 1). The   is observed when processing the full resolution 3.125 km images. In fact a slight worsening can be observed for the 36.5 GHz channels.
Thanks to their higher microwave frequency, the GCOM-W1 AMSR2 89 GHz imagery channels have a distinctly better resolution than the 36.5 GHz channels (Table 1). However, this finer resolution does not seem to translate into a much better accuracy of the S2S drift vectors, as the RMSEs and biases reported on Fig. 8 are very similar for both microwave frequencies.
Here we can hypothesize that the finer spatial resolution of the 89 GHz channels are balanced by the relative lack of intensity (and/or temporal stability) of intensity patterns to track over a ~24 hours period. Notably, the 36.5 GHz imagery is emitted from deeper in the ice/snow media, is more sensitive to sea-ice type and snow-depth (rather stable in time), and less sensitive to the atmosphere above the ice. The 89 GHz is emitted closer to the top of the ice/snow media and is thus more sensitive to temperature changes. It is also much more sensitive to liquid water path in the atmosphere, that can easily blur or hide intensity patterns at the surface. We note that the lack of improvement of the RMSE metric from 6.25 km to 3.125 km could also be due to our rather simple approach to resampling the 89 GHz channels to the 3.125 km grid.
However, it is not obvious that advanced gridding techniques (e.g. Backus-Gilbert) can do much better as they could be challenged by the lack of sufficient overlap between neighboring 89 GHz Field of Views. The slight increase of RMSE seen for the 36.5 GHz channels from 6.125 km to 3.125 km also indicates that overly oversampling the native spatial resolution of the imagery channels is not beneficial, and that it seems more efficient for sea-ice drift retrievals to aim at an optimum resampling resolution for each channel, rather than the finest possible spacing.
Still, the spatial resolution of the images does not only drive the accuracy of the drift vectors, but also the overall resolution of the vector field. Indeed, a finer image will allow motion tracking algorithms to be run with smaller image sub-windows, and thus allow the creation of a denser vector field. For example, the finer resolution of the AMSR2 89 GHz channels allows for tracking smaller areas down to 70 km x 70 km, and processing denser vector fields (e.g. 31.25 km, Ezraty et al. 2007).
Increasing the grid spacing to 31.25 km was attempted in our study but did not result in significantly different results for either the 36.5 GHz or 89 GHz imagery (not shown). In fact, the result was a slightly larger RMSE for both frequency channels, and at all coarsening kernels. This might be because tracking smaller image sub-windows leads to tracking fewer intensity patterns per vector. As is true for the retrieval of many other geophysical variables, increasing the spatial resolution often leads to increasing the noise level. The increased RMSEs might also reflect that most of our validation data are from the central Arctic Ocean and mostly sample drift events that are coherent over large spatial scales. The higher resolution drift vectors might be more accurate in cases where sharp spatial gradients of the drift field are observed, either in the event of sea-ice deformation, or in places with stable velocity gradients such as in the East Greenland Sea. However these conditions are not well enough represented in our validation data set to have a positive impact on the overall statistics.

Simulating the spatial coverage of S2S drift vectors from CIMR
The vectors should be observed towards lower latitudes (needed to better sample the core of the sea-ice cover in the Southern Hemisphere) and higher latitudes (needed to monitor the sea-ice cover close to the North Pole).
We simulate 48 hours of CIMR orbit and swath coverage (Sect. 3.4) and enter these swaths in our S2S sea-ice drift processing chain. We repeat the analysis of spatial distribution presented in Sect. 4.1 with AMSR2 data.  The total number of S2S vectors for the Arctic coverage is 143,101 for CIMR (compared to 110,400 for AMSR2 and 1,729 for the DM product). Over Antarctic sea-ice, the number is 108,601 for CIMR (compared to 74,737 for AMSR2 and 3,617 for the DM product). The wider swath of CIMR thus results in 30-40% more S2S vectors compared to AMSR2.
When comparing Fig. 9 (CIMR) to Fig. 3 (AMSR2) in the Northern Hemisphere, a striking improvement is seen towards the North Pole. In Sect. 4.1 we explained how the polar observation hole of the AMSR2 imagery, despite being of only 0.5°l atitude, has a wider impact on the coverage of S2S vectors near the pole. The CIMR orbit and swath width are specifically optimized to allow full sub-daily coverage of the polar regions including the poles, thus "no hole at the pole" (Donlon et al., 2020). Our analysis shows that the number of S2S vectors from CIMR reduces towards the pole, but that there are vectors all the way to the pole on a daily basis.
We underline that the actual number of S2S vectors from CIMR will be larger that discussed here, simply because the increased spatial resolution of the Ku-and Ka-band imagery (Table 1) will allow a denser vector field than we processed (62.5 km grid-spacing).

Discussion
In this study, we processed Arctic and Antarctic sea-ice drift vectors from the AMSR2 imagery, and compare them to GPS trajectories of on-ice buoys. We use a state-of-the-art algorithm (Lavergne et al., 2010), as implemented in the processing chain of the EUMETSAT OSI SAF. Our focus is to compare two approaches: a) processing sea-ice drift vectors from the intersection of individual swaths (S2S-type products) and b) the current status-quo which is to process sea-ice drift vectors from daily average maps of satellite signal (DM-type products). We show that not only many more vectors can be processed with an S2S-type product on a daily basis (Sect. 4.1), but that these vectors also validate better against GPS trajectories of on-ice buoys (Sect. 4.2 and 4.3). We also document the impact that geo-location uncertainty (Sect. 4.4) and spatial resolution of the imagery (Sect. 4.5) have on the accuracy of the sea-ice drift field.
We link the better accuracy of the S2S vectors to two aspects of the motion tracking algorithm. First, the imagery of individually-gridded swaths presents sharper intensity features over sea ice, which results in tracking more accurate sea-ice motion vectors. Once they are averaged over a daily period, the satellite imagery is blurred by the very motion we want to measure. Second, the start and end times assigned to S2S vectors are more accurate than those assigned to DM vectors.
Better defined start and end times lead to better collocation with buoy trajectories, which improves the validation statistics.
Beyond validation, having more accurate start and end times will help matching S2S vectors with (e.g. hourly) fields from ocean/ice forecast models. This will have most impact in cases where the motion field has strong temporal gradients, e.g.
when a low-pressure system travels over sea ice.
To the best of our knowledge, the better accuracy of S2S vectors from passive microwave satellite missions has not been systematically documented. Maslanik et al. (1998) note that investigations on S2S vectors had been conducted but that this does not significantly improve the comparison with buoys. It must be underlined that we now have acccess to much better data than in the late 1990s, both in terms of satellite imagery (AMSR2 versus the SSM/I) and in terms of buoy data. Indeed, most Arctic buoy data in the late 1990s were 3-hourly trajectories with geo-location via the Argos positioning system, which could have degraded the validation statistics.
Existing routinely generated sea-ice drift products such as those from the EUMETSAT OSI SAF, IFREMER CERSAT, or JAXA's AMSR2 ground processor could readily move from DM sea-ice drift products to an S2S configuration with limited investment, since the core of motion tracking algorithm is the same. The positive impact will include better accuracy, better timeliness, and more sea-ice drift information for their users. Both EUMETSAT OSI SAF and IFREMER CERSAT products closer to 24 hours. Our results (e.g. Fig. 5) show that S2S products from AMSR2 can be moved towards 24 hours and shorter which can be easier for Data Assimilation applications.
When it comes to new satellite missions and services, and specifically CIMR, we recommend that S2S sea-ice drift products are adopted from the start in the ground segment, so that the processing chains are dimensioned accordingly and users and downstream services can prepare for it.
The CIMR mission has several characteristics that can be exploited to prepare sea-ice drift information with a higher accuracy and resolution than currently available from any passive microwave satellite mission. We discuss some of these here. First, CIMR will provide a suite of microwave frequency channels at a much higher spatial resolution than AMSR2 (see Table 1), e.g. "better than 5 km" at 36.5 GHz (Ka-band), and 5 km at 18.7 GHz (Ku-band). In Sect. 4.4, we illustrated that imaging resolution is a key element for the final accuracy of sea-ice drift vectors (Fig. 8) and for the spatial resolution of sea-ice drift products (the spacing between neighboring vectors). Particularly, we showed that the high-frequency channels of the AMSR2 instrument (89 GHz) did not bring extra accuracy, probably because the radiation was emitted at a shallower depth in the snow/ice layer, and the resulting intensity patterns were less stable with time and thus harder to cross-correlate.
As for the retrieval of other ocean and sea-ice parameters, the true advantage of CIMR will not be to access to highresolution radiometric images in absolute terms, but rather to give access to high-resolution radiometric images at microwave frequencies that until now were only accessible at medium-to-coarse resolution (> 10 km). While the 36.5 GHz channels will make the main contribution to the sea-ice drift accuracy during the cold winter season, the 18.7 GHz channels at 5 km will have a strong impact in the melting and summer season (Kwok, 2008). In any case, in order to fully succeed as a sea-ice drift mission, CIMR will also have to deliver a high geo-location accuracy. In Sect. 4.3, we documented how geolocation errors, especially if they are systematic with respect to the imaging dimensions, rapidly grow into prohibitive retrieval errors for S2S-type of products. In this respect, S2S products are more affected than DM products, because the latter average the geo-location error throughout the day and -due to the orbital configuration of most polar orbiting missions-the same areas are revisited roughly at the same times a few days apart. Since CIMR will use a large rotating deployable mesh antenna reflector, the geo-location accuracy translates into a stringent requirement on the pointing accuracy of the antenna.
In addition, it is expected that small remaining systematic geo-location errors will be asessed and corrected against coastlines (Wiebe et al., 2008). From the results presented here, we argue that the accuracy (especially the bias) of a future S2S sea-ice drift product from CIMR, assessed against on-ice GPS buoy trajectories, would constitute an independent check of the geolocation accuracy of the mission, e.g. as part of the Calibration Validation (Cal/Val) phase.
The high spatial resolution, and high geo-location accuracy CIMR Level-1B (geo-located brightness temperatures in swath projections) products will not directly be input to the Level-2 S2S drift processing. Indeed, the motion tracking algorithms require remapped brightness temperatures as input. Following other missions (e.g. NASA SMAP), CIMR will also process a CIMR's Level-1C product will be the input of its Level-2 S2S drift product, which has several advantages including the collocation of all CIMR's channels onto a set of directly overlapping EASE2 grids (Brodzick). In Fig. 8, we found that the accuracy of the drift vectors stalled (89 GHz) and even slightly degraded (36.5 GHz) when grid spacing approached or went beyond the true resolution of the imagery, possibly because our gridding methodology was basic and might have introduced artefacts at small grid spacing. The gridding algorithms implemented in the CIMR Level-1C product should be carefully designed to not introduce such artefacts and retain the true resolution of the Level-1B information, so as not to reduce the final accuracy of the Level-2 S2S drift product.
A number of features of the CIMR mission prompt specific research and development before they can be fully exploited in the operational sea-ice drift processing. We mention some of these here, as a way forward for the development of CIMRspecific algorithms. CIMR will offer high-resolution brightness temperature imagery at the microwave frequencies listed in Table 1. Being the first radiometer to offer such a high spatial resolution at, e.g. 6.9 GHz and 10.7 GHz, the potential contribution of these frequency channels to a sea-ice drift product will have to be investigated. Since they are emitted from deeper in the snow and sea-ice layer, they could potentially contribute to the summer melt period, in addition to 18.7 GHz.
CIMR will offer both a forward and a backward scan, separated with four minutes at the center of the swath (Donlon et al., 2020). Considering the typical sea-ice drift speed and the resolution of the CIMR channels, we do not expect to be able to detect motion taking place during such a short time. However, the swaths corresponding to the forward and backward scans are independent, overlapping, and mostly simulatenous images that can both be input to a sea-ice motion tracking algorithm.
This might be used to reduce the noise of retrieved sea-ice drift vectors, either by averaging the two independent drift vectors, or as an additional input to the quality control step and the detection of rogue-vectors (see Sect. 3.1).
Independently of the R&D elements outlined above, the design and dimensioning of a CIMR Level-2 sea-ice drift Payload Data Ground Segment (PDGS) will need careful consideration for the specificities of a S2S sea-ice drift product. Typically, a Level-2 processing chain produces a single Level-2 product per input Level-1 file. This will not be the case for the CIMR Level-2 drift product. Instead, each incoming Level-1C file can potentially be paired with several past Level-1C files, and each pair results in a Level-2 sea-ice drift product. When dimensioning the data throughput of the sea-ice drift processor, we must make sure that all relevant pairs are processed before the next Level-1C file is available. One can think of several schemes for selecting those interesting pairs, including the drift duration (time separation between the two Level-1C files) and the angle between the swaths, which may relate to geo-location noise (see Sect. 4.3). The setup we adopted in this study (pairing a Level-1 file with all preceding Level-1 files within 48 hours) resulted in about 40 Level-2 sea-ice drift files per Level-1 file (>500 Level-2 files per day) which can be overwhelming, but does fully sample the temporal variability of the sea-ice motion field. In preparing for the CIMR mission, one also has to consider that today's users, especially from the modelling community, are not used to these S2S products. Adopting this type of product might require some dedicated efforts, that can luckily be conducted in advance, for example using AMSR2 data as we did here. As with the other parameters to be observed by CIMR, a Level-3 sea-ice drift product should be prepared that combines e.g. a day's worth of S2S products into a complete map of sea-ice motion vectors. Yet, such merging algorithms do not exist for sea-ice drift vectors. Finally, we note that sea-ice drift vectors are today processed at high resolution from Sentinel-1 SAR images, e.g. in the Copernicus Marine Environment Monitoring Service (CMEMS), and that this type of SAR-based product will continue with other missions such as the Radarsat Constellation Mission (RCM), the Expansion Sentinel (formerly known as High Priority Candidate Mission) Radar Observing System for Europe -L-Band (ROSE-L), and later the Sentinel-1 "next generation" platforms. Despite the increase in SAR missions, the coverage of the Arctic and Antarctic sea-ice at a sub-daily frequency will remain prohibitive in the foreseeable future. A Level-4 Ice Drift Analysis (IDA) product merging S2S drift products from both CIMR and several SAR missions would fill a data gap by providing a high-level sea-ice motion product at high resolution and accuracy, with a daily complete coverage. Such a Level-4 product does not exist today. It could be developed from AMSR2 and Sentinel-1 data in preparation for the Copernicus Expansion Sentinels polar missions.

Conclusions
We investigate the feasibility and impact of adopting a "swath-to-swath" (S2S) versus "daily-map" (DM) framework for the processing of sea-ice motion from modern passive microwave satellite missions such as JAXA's AMSR2, in preparation for the future CIMR mission. We find that S2S sea-ice drift vectors obtained from AMSR2 imagery are more accurate than the corresponding DM vectors when compared to GPS trajectories from on-ice buoys, both in the Arctic and Antarctic. An S2S configuration also results in many more drift vectors on a daily basis: the number varies with latitude and depends on the orbital and swath characteristics of the satellite mission. Since S2S drift vectors can be prepared for each new incoming swath, this configuration yields much better timeliness, which is beneficial for several operational applications such as support to navigation and short term sea-ice forecasting. One potential limitation to the S2S configuration is that it is more sensitive to inaccurate geo-location, especially if the geolocation errors are systematic (e.g. a shift in the flight direction).
As far the the CIMR mission is concerned, we recommend the adoption of a S2S configuration for the Level-2 sea-ice drift product in the operational ground segment. Considering the microwave frequency channels, target spatial resolution, swath width, and geolocation accuracy specified for the CIMR imagery, its Level-2 sea-ice drift product will allow unprecedented spatial resolution, coverage and accuracy for a microwave radiometer mission. Several other new characteristics of the CIMR mission (e.g. the relatively high spatial resolution at 6.9 and 10.8 GHz, the backward and forward scans) will also contribute to an enhanced sea-ice drift product, but this requires further research.
We finally note that such Level-2 S2S sea-ice drift products will be new to a large fraction of the user community, and their downstream uptake must be prepared. This includes the preparation of Level-3 daily products from the CIMR mission only, as well as Level-4 daily products merging several sources such as CIMR, SARs (e.g. Sentinel-1 and ROSE-L) and potentially on-ice buoy trajectories. Such Level-4 Ice Drift Analysis (IDA) will require the development of dedicated algorithms and processing chains.

Data availability
A subset of the S2S and DM data used in this study are made available for inspection during the review process. The data are hosted at the Norwegian Meteorological Institute. They are formatted as NetCDF4 (classic) files, and follow the Climate and Forecast (CF) and Attribute Convention for Data Discovery (ACDD) conventions. All sea ice-drift products made available here are from the GCOM-W1 AMSR2 36.5 GHz imagery. The following data are prepared: Northern Hemisphere covering 15 to 30 November 2019: • S2S drift vectors (DOI); • DM drift vectors (DOI); Southern Hemisphere (Weddell Sea) covering 15 to 31 July 2019: • S2S drift vectors (DOI); • DM drift vectors (DOI); The four DOIs are being minted at the "Norwegian Meteorological Institute, Free Meteorological Data" data center (https:// www.re3data.org/repository/r3d100011014) and will be added to the Discussion manuscript as soon as they are available.
We selected those two periods because they exhibited dynamic events in the sea-ice drift fields, including sharp spatial gradients and rotation patterns caused by low pressure systems. This should help demonstrate the different characteristics of the DM and S2S approach.

Competing Interests
The authors declare that they have no conflict of interest.

Author contribution
TL designed and carried out the study. MPS performed the CIMR orbit simulations. ED optimized some of the sea-ice drift software. TL prepared the manuscript with contributions from all co-authors.