CoAuthors

Angelica Tarpanelli (Research Institute for Geo-Hydrological Protection, Italy); Elena Zakharova (Institute of Water Problem, RAS, Moscow, Russia); Jérôme Benveniste (European Space Agency, Italy)

The river level from altimetry is mainly portrayed at virtual stations.
Hence, at locations where repeat missions such as Envisat, AltiKa (before June 2016),
the Jason missions and the recently launched missions Sentinel-3A and 3B cross the river.
These missions can provide us with river levels at a temporal resolution equivalent
with the repeat time a the given missions. For the Jason missions, Envisat/Altika,
and the Sentinel-3, this is 10, 35 and 27 days, respectively.

For missions such as CryoSat-2 and AltiKa (after June 2016) the virtual station approach is
not an option due to the drifting track pattern. Hence, construction time series of water
level is challenging due to the topographic contribution present at each and that the
amplitude of the water level might be different also.

The water level variations in rivers, hence the phases of annual signals and the amplitude,
can be quite different. The seasonal pattern in Arctic rivers, for instance,
is dominated by a flood wave in the spring when the ice and snow start to melt.
In the Lena river, the water rises with up to 20-25 meter within a few weeks. In general
narrow rivers are more sensitive to sudden events, such as extreme rain, compared to large rivers.
Hence, we might not capture the entire signal at the virtual stations.

Here we use data from CryoSat-2, SARAL/AltiKa, and Sentinel-3A/3B for a river segment to make
a joint solution of the river level. We experiment with different approaches to reconstruct
the river level time series. One of the approaches is a state-space model composed
of a process part consisting of an AR1 process
and an observational part, where the error follows a mixture between a Gaussian and a
Cauchy distributions. The spline functions account for the change in topography
and a potential variation in the water level amplitude along the river.

The model approach has the advantage that the river level time series can be reconstructed
at any location along the considered river segment.

Here We show the results for rivers (Lena, Ob, Po, Mississippi, and the Amazon)
of different sized and in different environments. When validating with in situ data
we find that we are able to map the river levels, especially for the smaller rivers, in greater detail
compared to what can be obtained at the virtual stations.

The presented work is funded by ESA through the projects Ridesat and ArcFlux.