Abstract's details
From ERA-Interim to ERA5: impact of the latest ECMWF reanalysis in the computation of radar altimeter Wet Path Delays
CoAuthors
Event: 2019 Ocean Surface Topography Science Team Meeting
Session: Instrument Processing: Propagation, Wind Speed and Sea State Bias
Presentation type: Type Poster
Contribution: not provided
Abstract:
The accurate determination of sea surface heights (SSH) using satellite altimetry requires Wet Path Delays (WPD) estimations with the same accuracy. The most accurate way to measure the effect of the wet troposphere in the altimetric signals is from on-board Microwave Radiometers (MWR) measurements. However, the WPD retrievals from MWR become invalid and cannot be used over coastal and inland waters and, in addition, some satellites do not possess an on-board MWR (e.g. CryoSat-2). An alternative WPD source can be atmospheric parameters provided by Numerical Weather Models (NWM), e.g. those from the European Centre for Medium-Range Weather Forecasts (ECMWF).
ERA-Interim, the fourth reanalysis from ECMWF, has been used in the context of tropospheric corrections for satellite altimetry. It provides parameters at 6-hours intervals and 0.75°x0.75° spatial sampling. This reanalysis has recently been replaced by the latest ERA5 reanalysis, which has a much higher spatial (0.25°x0.25°) and temporal (1-hour) resolutions and an improved troposphere modelling, with particular interest for the WPD computation. ERA5 is the first ECMWF reanalysis available at 1-hour intervals and provides many improvements, in comparison with its predecessor ERA-Interim. The relevance of model reanalyses is well recognised, for example in climate applications or over periods for which the ECMWF operational model was less accurate (before 2004).
Motivated by these improved resolutions, the focus of this study is the global assessment of the impact of using ERA5 in the computation of WPD. Firstly, this assessment is performed in order to quantify the impact of using different resolutions of ERA5, mainly the temporal resolution and, secondly, to evaluate the global improvement of ERA5 over ERA Interim. Aiming to identify the best compromise between spatial/temporal resolutions, accuracy and computational time, WPD estimated from ERA5 using various combinations of these resolutions were inter-compared. Additionally, the same WPD computed from ERA5, as well as computed from ERA-Interim are compared with independent retrievals from MWR.
Independent comparisons with WPD derived from MWR (on board EnviSat and Jason-2) show that the effect of using ERA5 at 1-h instead of 6-h intervals is small, significant only for latitudes 30°-60°S and 30°-60°N. Results indicate that hourly intervals do not have significant impact on the WPD from ERA5, being a temporal resolution of 3-h high enough to ensure the same level of accuracy. Concerning the spatial resolution, this study shows that the effective useful spatial resolution of ERA5 is a bit worse than its native resolution, since ERA5 spatial samplings better than 1°x1° do not generate significant improved WPD estimations.
Results also show that the global RMS of the differences MWR-ERA5 is 1.2 cm, varying with latitude between 0.6 and 1.7 cm. Considering the differences between retrievals from MWR and WPD computed from ERA-Interim, the global RMS is 1.4 cm, varying with latitude between 0.8 and 1.9 cm. These results show a global improvement of 0.2 cm from ERA-Interim to ERA5, however for some latitude bands the improvement can reach 0.4 cm. These results illustrate the considerable impact of using the ERA5 in the computation of radar altimeter wet path delays.
As recognized for the previous atmospheric models, results obtained from the independent comparisons show that ERA5 cannot map the WPD small space and time scales, evidencing the limitations of the latest ECMWF reanalysis, being the measurements from MWR, whenever valid, the most accurate way to account for the effect of the wet troposphere in satellite altimetry. Results of this study provide relevant information to ensure that when NWM-derived WPD are used in satellite altimetry, the best compromise is achieved between accuracy and computational time.
ERA-Interim, the fourth reanalysis from ECMWF, has been used in the context of tropospheric corrections for satellite altimetry. It provides parameters at 6-hours intervals and 0.75°x0.75° spatial sampling. This reanalysis has recently been replaced by the latest ERA5 reanalysis, which has a much higher spatial (0.25°x0.25°) and temporal (1-hour) resolutions and an improved troposphere modelling, with particular interest for the WPD computation. ERA5 is the first ECMWF reanalysis available at 1-hour intervals and provides many improvements, in comparison with its predecessor ERA-Interim. The relevance of model reanalyses is well recognised, for example in climate applications or over periods for which the ECMWF operational model was less accurate (before 2004).
Motivated by these improved resolutions, the focus of this study is the global assessment of the impact of using ERA5 in the computation of WPD. Firstly, this assessment is performed in order to quantify the impact of using different resolutions of ERA5, mainly the temporal resolution and, secondly, to evaluate the global improvement of ERA5 over ERA Interim. Aiming to identify the best compromise between spatial/temporal resolutions, accuracy and computational time, WPD estimated from ERA5 using various combinations of these resolutions were inter-compared. Additionally, the same WPD computed from ERA5, as well as computed from ERA-Interim are compared with independent retrievals from MWR.
Independent comparisons with WPD derived from MWR (on board EnviSat and Jason-2) show that the effect of using ERA5 at 1-h instead of 6-h intervals is small, significant only for latitudes 30°-60°S and 30°-60°N. Results indicate that hourly intervals do not have significant impact on the WPD from ERA5, being a temporal resolution of 3-h high enough to ensure the same level of accuracy. Concerning the spatial resolution, this study shows that the effective useful spatial resolution of ERA5 is a bit worse than its native resolution, since ERA5 spatial samplings better than 1°x1° do not generate significant improved WPD estimations.
Results also show that the global RMS of the differences MWR-ERA5 is 1.2 cm, varying with latitude between 0.6 and 1.7 cm. Considering the differences between retrievals from MWR and WPD computed from ERA-Interim, the global RMS is 1.4 cm, varying with latitude between 0.8 and 1.9 cm. These results show a global improvement of 0.2 cm from ERA-Interim to ERA5, however for some latitude bands the improvement can reach 0.4 cm. These results illustrate the considerable impact of using the ERA5 in the computation of radar altimeter wet path delays.
As recognized for the previous atmospheric models, results obtained from the independent comparisons show that ERA5 cannot map the WPD small space and time scales, evidencing the limitations of the latest ECMWF reanalysis, being the measurements from MWR, whenever valid, the most accurate way to account for the effect of the wet troposphere in satellite altimetry. Results of this study provide relevant information to ensure that when NWM-derived WPD are used in satellite altimetry, the best compromise is achieved between accuracy and computational time.