Abstract's details

Exploitation of the ENA Ground-Based Water Vapour Radiometers in Satellite Altimetry

Bernard Vasconcellos (DGAOT, Faculdade de Ciências, Universidade do Porto, Portugal)


Clara Lázaro (DGAOT, Faculdade de Ciências, Universidade do Porto, Portugal); Telmo Vieira (DGAOT, Faculdade de Ciências, Universidade do Porto, Portugal); M. Joana Fernandes (DGAOT, Faculdade de Ciências, Universidade do Porto, Portugal)

Event: 2022 Ocean Surface Topography Science Team Meeting

Session: Instrument Processing: Propagation, Wind Speed and Sea State Bias

Presentation type: Type Poster

Contribution: PDF file


Satellite altimetry has become a standard tool for many Earth Observation studies applied to sea, glaciers, rivers, inland waters and lakes, in order to measure the height of a water body above a reference ellipsoid. The signal emitted by the altimeter sensor is highly dependent on the wet tropospheric correction (WTC), due to its interaction with water vapour suspended in this atmospheric layer. Since water vapour does not behave as a well-mixed gas, its prediction becomes a delicate task. Therefore, microwave radiometers are on board the altimeter satellites in order to determine the WTC very accurately over open-ocean. Despite the ability of this passive sensor to produce collocated measurements with the altimeter, other sources of WTC are needed in locations where observations are invalidated due to the presence of non-ocean surfaces in its footprint, such as in coastal areas and high latitudes.
This study aims to investigate ground-based radiometers (MWRGB) as a reliable source of water vapour measurements for deducing the WTC of altimetric observations. For this purpose, the WTC from the MWRGB is assessed by comparison with other four external WTC sources: (1) microwave radiometers on board (MWROB) altimetry missions such as Sentinel-3 A and B, SARAL/AltiKa, and Jason-3; (2) Global Navigation Satellite Systems (GNSS); (3) radiosonde (RS); and (4) ECMWF (European Centre for Medium-Range Weather Forecasts) - ERA5 atmospheric model. The first three comparisons can be collocated or not while the latter is collocated through spatial and temporal interpolation. Among all the comparisons, the only one that is not independent is the comparison with radiosonde, since the information provided by this source is also introduced in the MWRGB retrieval algorithms.
MWRGB from one observatory of the Atmospheric Radiation Measurements (ARM) user facility have been used in this study, which is the ENA (Eastern North Atlantic). In addition, retrievals from two ARM algorithms were used in order to evaluate which one best suits the needs of Satellite Altimetry – NN or MWRRETV2.
For the ENA observatory, collocated comparisons, or up to 40 km, show RMS of WTC differences in the range 1.02 cm - 1.41 cm. The collocated comparison with ERA5 show RMS of 1.09 cm - 1.19 while the comparison with GNSS, which is non-collocated at only 51 m, shows a higher value of 1.41 cm. Furthermore, the comparisons with MWROB up to 40 km present an RMS in the range of 1.02 cm - 1.30 cm. For the comparison with RS, which is non-collocated at 89 km, an RMS of nearly 2.37 cm is found, which may be explained by the large distance between the two datasets, close to the limit of the WTC spatial correlation scale.
The intra-algorithm assessment showed that in general the NN and MWRRETV2 algorithms have great similarity in their results, with a variation of RMS of WTC differences in the range of 0 – 2.8 mm. Therefore, for the needs of Satellite Altimetry up to 40 km, the NN algorithm proves to be a reliable source for deducing WTC, due to the near real-time latency of its retrieved data.
At last, two neural network algorithms were tuned to estimate the WTC directly from MWRGB brightness temperatures (TB) observations, which are: (WTCGB_2TB) using 2 inputs - TB from both the 23.8 and 30 GHz bands; and (WTCGB_3TB), using 3 inputs – the former two TB with further inclusion of the TB from the 90 GHz band. Thus, the training dataset consisted of 100,000 samples which refer to 1 year of observations, 3 or 2 TB as inputs, and model-interpolated WTC for the same instants as output. An independent assessment for the WTC values retrieved from both algorithms was carried out against GNSS data. WTCGB_2TB presented an RMS of 1.42 cm while the WTCGB_3TB obtained a better performance of 1.34 cm. Furthermore, the comparison with the NN algorithm, for the same period, showed an RMS of 1.41 cm, which was higher than the result found for the WTCGB_3TB.

Poster show times:

Room Start Date End Date
Mezzanine Tue, Nov 01 2022,17:15 Tue, Nov 01 2022,18:15
Mezzanine Thu, Nov 03 2022,14:00 Thu, Nov 03 2022,15:45
Bernard Vasconcellos
DGAOT, Faculdade de Ciências, Universidade do Porto