Abstract's details
A one-dimensional variational approach for wet tropospheric correction retrieval in the perspective of high resolution altimetry mission: defining the the background error covariance matrix
CoAuthors
Event: 2015 Ocean Surface Topography Science Team Meeting
Session: Instrument Processing: Corrections
Presentation type: Type Oral
Contribution: PDF file
Abstract:
The future altimetry missions planned for the coming months (Jason-3, Sentinel-3) or for the coming years (SWOT) aim to deliver a measurement of the topography at a finer spatial resolution, a higher temporal rate and over heterogeneous surfaces, open ocean but also coastal regions, hydrological targets and over ice and sea-ice.
In this perspective, the role played by the wet tropospheric correction (WTC) is critical, due to its large temporal and spatial variability and its crucial weight in the final budget error.
Current algorithms are based on empirical approaches parameterized using measured (radiosonde) or modeled (numerical weather prediction analysis) atmospheric profiles; in both cases, a radiative transfer model relates the integrated content of WTC to the top of the atmosphere brightness temperatures (TB) and a relation is fitted between the two datasets.
This method is valid over open ocean only, where a model of the emissivity is available. The performances are then degraded wherever the instrumental measurements are contaminated by other surfaces (land, ice, sea-ice); solutions exist to correct for this contamination but will not be able to satisfy the future constraints on the retrieval errors over coastal regions or hydrological surfaces.
In this context, a one-dimensional variational approach (1D-VAR) for wet tropospheric correction retrieval is a good candidate to provide a unique method well adapted to all surfaces. Where current algorithms directly provide an integrated value of WTC, this latter aims to estimate the atmospheric profiles that best explain the TOA TB measurements. The WTC is then computed from integration of the retrieved profiles. Depending on the surface, the emissivity is provided by a model (open ocean) or emissivity atlas (other surfaces) estimated from TOA measured brightness temperatures. Previous work (Desportes et al. 2010) has already shown the potential of 1D-VAR over coastal regions.
In this presentation, we will focus on the background error covariance matrix. The method applied for its computation is presented and a set of matrices is defined according to specific atmospheric conditions. A sensitivity analysis of the impact of this set of matrices on the retrieval performances is performed using AMR Jason-2 TB and AMR WTC as a reference.
In this perspective, the role played by the wet tropospheric correction (WTC) is critical, due to its large temporal and spatial variability and its crucial weight in the final budget error.
Current algorithms are based on empirical approaches parameterized using measured (radiosonde) or modeled (numerical weather prediction analysis) atmospheric profiles; in both cases, a radiative transfer model relates the integrated content of WTC to the top of the atmosphere brightness temperatures (TB) and a relation is fitted between the two datasets.
This method is valid over open ocean only, where a model of the emissivity is available. The performances are then degraded wherever the instrumental measurements are contaminated by other surfaces (land, ice, sea-ice); solutions exist to correct for this contamination but will not be able to satisfy the future constraints on the retrieval errors over coastal regions or hydrological surfaces.
In this context, a one-dimensional variational approach (1D-VAR) for wet tropospheric correction retrieval is a good candidate to provide a unique method well adapted to all surfaces. Where current algorithms directly provide an integrated value of WTC, this latter aims to estimate the atmospheric profiles that best explain the TOA TB measurements. The WTC is then computed from integration of the retrieved profiles. Depending on the surface, the emissivity is provided by a model (open ocean) or emissivity atlas (other surfaces) estimated from TOA measured brightness temperatures. Previous work (Desportes et al. 2010) has already shown the potential of 1D-VAR over coastal regions.
In this presentation, we will focus on the background error covariance matrix. The method applied for its computation is presented and a set of matrices is defined according to specific atmospheric conditions. A sensitivity analysis of the impact of this set of matrices on the retrieval performances is performed using AMR Jason-2 TB and AMR WTC as a reference.