Lázaro Clara (Universidade do Porto, Faculdade de Ciências & Centro Interdisciplinar de Investigação Marinha e Ambiental (CIIMAR/CIMAR), Portugal); Telmo Vieira (Universidade do Porto, Faculdade de Ciências, Portugal)

Previous studies by Fernandes et al. [2014] show that the main errors in the level 2 (L2) corrections present in altimeter products over inland water regions are on the tropospheric (dry and wet) corrections. These errors are mainly related to the fact that these corrections are often provided either at sea level or at the level of the orography associated to the adopted atmospheric model, often ECMWF. The dry tropospheric correction (DTC) has a relatively well-known mean variation with height of 2.5 cm per each 100 m. Unlike the DTC, the height dependence of the wet tropospheric correction (WTC) is more difficult to model and is function not only of surface height but also of the WTC itself. Height errors of, e.g., 100 and 500 m, induce errors of 5% and 28%, respectively (1 cm and 5.6 cm for a correction of 20 cm).
In the scope of the Sentinel-3 Hydrologic Altimetry PrototypE (SHAPE) project, the University of Porto has been analysing the errors present in the DTC and WTC of the CryoSat-2 (C2) products and developing new L2 corrections, aiming at getting improved products for C2 and Sentinel-3 (S3). This study is being conducted on selected regions of interest (ROI), covering rivers and lakes where C2 is operating in SAR and SAR-In modes, such as the Amazon and Danube rivers, the Titicaca and Baikal lakes.
The analysis of the DTC and WTC errors requires a proper inspection of the height variations within each ROI. For this purpose, the ACE2 digital elevation model (DEM) has been used. In addition, for each river, mean river profiles derived in the scope of this project by AlongTrack (ATK) from Jason-2 data and mean lake levels derived from Envisat data have been used.
The DTC and WTC present in the C2 L1B products have been compared against corrections computed from the ECMWF operational model at various levels: i) the level of ECMWF model orography; ii) the level of the ACE2 DEM; iii) the level of mean lake or river profile.
Concerning the WTC, the correction present in C2 products has also been compared with GNSS-derived WTC, whenever GNSS data are available within the ROI, as e.g. the Danube River. Since S3 microwave radiometer (MWR) is a 2-band radiometer similar to that of Envisat, model-derived WTC are also compared to Envisat derived WTC in the central parts of the lakes, where radiometer data are valid, while S3 data are not available.
Results show that the model-derived corrections present in C2 products seem to be referred to the model orography that can depart from the mean river profile or mean lake height by hundreds of metres. ACE2 DEM is a better altimetric surface than ECMWF orography. However, height errors up to e.g.120 m in the Amazon and 210 m in Lake Titicaca exist in ACE2.
DTC errors up to 3-5 cm exist in C2 products, in most ROI. Although in statistical terms these errors are small, they are systematic and if not accounted for, induce significant errors e.g. in the determination of mean river profiles or lake level time series. Comparison with surface pressure observations shows that DTC absolute error due to ECMWF pressure uncertainty is negligible.
For the analysed ROI, the WTC errors related with the height dependence of the correction are smaller than those for the DTC, up to a few cm.
Overall, results indicate that DTC and WTC errors are of the order of 1 cm for most ROI, providing that the corrections are computed at the mean river profile (using the height of the closest point in the profile) or mean lake level, previously determined from satellite altimetry. Full assessment of model-derived WTC errors using GNSS and MWR derived path delays is ongoing. For completeness, since all analysed ROI are located in regions of small variability in the WTC, this assessment must be extended to regions of higher WTC variability.