From "how does it work" to "what's the difference with existing satellites", all the way through "what is the surface current?" ... here is a selection of frequently asked questions about SKIM.
Q: What is the depth at which SKIM measures the current?
A: SKIM gives a horizontal current velocity averaged over roughly the top two meter of the ocean. We know this from the theory of near-nadir radar back-scatter (Nouguier et al. 2018), in which the current only influences the back-scatter through the "Doppler shift" k.U term of the wave dispersion relation, where k is the wavenumber of the considered ocean waves, and U is the effective current velocity that advects the wave
phase. Generalizing Nouguier et al. (2018) to a vertically sheared current, we have a U(k) ~ integral of U(z) exp (2kz) dz as given by Stewart and Joy (1974). As a result, each wave train with wavenumber k contributes to the mean slope velocity (msv) with a velocity that is C(k) + U(k). Whereas the wave contribution through C(k) gives the mean slope velocity, the current gives a mss-weighted average of U(k).
So ... how deep is it in the end? Using the HF radar analogy with a good proxy of the depth given by lambda/(4 pi) for a linear shear (Stewart and Joy 1974), where lambda is the Bragg scattering wavelength, we can compute the weighted sum of the mss to find that the "depth of measurement" typically ranges beween 0.5 and 0.7 m, hence it is an average over the top 1 m. Certainly this will be worth checking in properly designed experiments (e.g. with shallow freshwater plumes as in Ivonin et al. JGR 2005) or in the laboratory.
Q: Don’t we already have current measurements ?
A: Currents are measured by current-meters on moorings, surface drifters and coast-based HF radars. Part of surface currents are estimated from a combination of Mean Dynamic Topography (MDT) given from satellite missions such as GOCE, and sea level anomaly (SLA) measured by satellite altimeters such as ERS-1 and now Jason 3 and Sentinel 3. These nadir altimeter-based estimates follows a pressure balance assumption, cannot resolve scales smaller than 200 km. At these large scales they work fine in many places but they have very large errors :
- near the equator: scales shorter than 40 days and 400 km are not resolved
- on shelf seas (where tides and wind-driven dynamics prevail)
- near the ice edge
- in strong western boundary currents (Gulf Stream, Agulhas … )
Also the spatial resolution of both MDT and SLA is limited to 100 km (see Rio et aL. 2018) Temporal resolution of MDT and SLA is 10 years and 20 days)
NB: The same thing is true for the winds: Aeolus will measure wind speed directly, whereas now it is estimated from various methods …
Q: How can you measure a horizontal current by looking almost straight down?
A: The principle of the SKIM measurement is that we measure the vertical velocity of the water or ice surface at known horizontal positions. If the surface is frozen, the vertical velocity is just the surface slope times the horizontal velocity. If the surface is moving (with ocean waves), the surface velocity contains also the wave phase speed. Hence the radar measures the apparent velocity of the surface. The radar can separate echoes as a function of their distance (red and pink bands in the radar beam) and so we get measurements of surface velocities for each "range bin".
Ice drift measurements
Q: Writing about "frozen surface", what velocity will SKIM measure over sea ice?
A: SKIM measures either a power-weighted mean velocity, which, in the presence of ice should be ice drift. This has not been demonstrated yet, except for Doppler centroid measurements with Sentinel 1 that provide one component of the velocity vector. Although the backscatter of ice at 0 to 12° incidence is less than that of oceans, we should get a higher instrument noise but almost no wave bias, giving similar performance in ice and in water.
SKIM and rivers
Q: Can we measure river velocities with SKIM?
A: Yes of course... but... watch out for observation conditions (low wind) and sparse swath filling. Because SKIM does not measure everywhere across the 300 km swath, some rivers will be missed, even if they are wider than the footprint (6 km)... and right now there is no plan to acquire data inland (coasts and estuaries are covered in the acquisition mask). Big estuaries > 30 km will be covered. Now it is possible that we could still get good doppler data from coastal rivers narrower than the footprint ... this is something to investigate (signal to noise ratio ..). Be careful also that without wind to roughen the surface there will be no echo to be measured, hence no velocity data. At 12° incidence this problem of "black water" is more severe than with SWOT at 4°. Before SKIM, you can already look into Doppler centroid data from Sentinel 1 IW mode: the processing should be radically improved shortly and with 1-component only and the orientation of the river, this may well be enough.