Abstract's details

Reconstructing the spatial and temporal elevation signals on large lakes from ICESat-2

Karina Nielsen (DTU Space, Denmark)


Heidi Ranndal (DTU Space, Denmark); Ole B. Andersen (DTU Space, Denmark)

Event: 2022 Ocean Surface Topography Science Team Meeting

Session: Science IV: Altimetry for Cryosphere and Hydrology

Presentation type: Type Oral

Satellite altimetry is now a well-established method to measure water level changes in lakes and rivers to understand climate change, for water resource management, and security.

The water level in lakes can generally be estimated with a relative RMSE of 10-20 cm when compared to in situ data. Contrary to what we expect, the RMSE is sometimes higher for larger lakes, not because of a poorer performance of the altimeters but due to a spatial signal in the elevation relative to the vertical reference potentially from un-modeled geoid signals, wind, or other static fields. Hence, when reconstructing the water level time series over large lakes, tracks crossing at different locations, are affected by a different contribution in elevation. This is especially problematic for geodetic missions like CryoSat-2 and SARAL/AltiKa, but also if more than one track is applied for the repeat missions.

To reduce the error due to a spatially distributed elevation signal on large lakes we suggest setting up a model where the spatial and temporal signal is estimated simultaneously. This is done in a state-space model where the spatial part is modeled via a Gaussian Markov Random Field and the temporal part is modeled via a random walk. To separate the spatial and temporal signals require a good coverage in space and time which is achieved with ICESat-2. A further advantage of ICESat-2 is the high precision that enables us to detect smaller variations.

Hence, we can use the highly precise ICESat-2 data to construct correction grids to account for the spatially distributed elevation signal. Such grids can be used to correct the other missions like CryoSat-2 and Sentinel-2 to improve the reconstructed water level time series.

We show examples from the lakes Tanganyika, Lake Malawi, Titicaca, Baikal, Issykul, and others. We also demonstrate how the reconstructed water level time series based on other missions like CryoSat-2 and Sentinel-3 is improved when using the ICESat-2 based correction grids.

Oral presentation show times:

Room Start Date End Date
Sala Grande Thu, Nov 03 2022,17:15 Thu, Nov 03 2022,17:25
Karina Nielsen
DTU Space