# Abstract's details

# Advantages and drawbacks of the filtered solution for dual-frequency ionospheric correction from altimeter

**CoAuthors**

**Event: **2022 Ocean Surface Topography Science Team Meeting

**Session: **Regional and Global CAL/VAL for Assembling a Climate Data Record

**Presentation type: **Type Poster

The ionosphere is the part of Earth’s upper atmosphere ionized by solar radiation. The resulting electron activity plays an important role in delaying the propagation of the altimeter pulses, accounting for errors of more than 10 cm in the evaluation of the distance between the satellite and the ocean surface. Such induced delay must therefore be corrected to obtain reliable altimetry observations.

To measure this ionosphere-induced delay, satellite altimeters use observations from two signals emitted at the same time at two different frequencies (e.g. these correspond to the wavebands–Ku and –C in the case of the Jason and Sentinel-3 missions). Because of the frequency difference, the two signals are differently delayed as they cross the ionosphere. The ionospheric correction can thus be estimated as a function of the difference in delay between the two signals (hence the term “dual-frequency ionospheric correction”) after the Sea State Bias (SSB) corrections are applied. Since it includes measurements errors from two different frequency bands, as well as errors associated with the choice of the SSB correction algorithm applied, the ionospheric correction is inherently noisy. Furthermore, since it is directly proportional to the ionospheric total electron content (TEC) which is primarily controlled by solar radiation, the correction is characterized by large spatial and temporal variability. Such low frequency variability is primarily controlled by three main factors:

• Solar activity

• Local time

• Latitudinal location

While applying the raw ionospheric correction improves SLA observations when the correction values are large, it becomes ineffective (if not even detrimental) when the values are small due to the small signal to noise ratio. Since the instrumental noise primarily controls the high-frequency variability, it is required to low-pass filter the ionospheric correction to improve the overall quality of the final SLA products.

Here, we describe the iterative filtering approach used for the production of Sentinel-3 and Jason-3 GDR-F altimetry products. The filter was designed to achieve two goals:

• Base the correction on as many dual-band ionospheric observations as possible

• Improve the correction where altimetric observations are discontinuous or isolated

To achieve the first goal, the selection of the ionospheric observations used for the correction is independent from the quality of sea level observations. However, while this maximizes the number of observations selected, at the same time increases the number of potential outliers. The iterative filtering applies a median and a Lanczos filter in sequence, in order to progressively reduce the number of outliers in the ionospheric observations used to compute the final filtered correction. To achieve the second goal, since the filtered correction has long spatial correlation scales, a spline interpolation is applied to fill gaps in the interpolated correction up to few hundreds km.

The advantages of such approach were assessed by comparing the resulting sea level anomaly fields with those obtained using the unfiltered ionospheric correction. Our results show that the SLA variance reduction from the iterative filtered ionospheric correction is always larger than that from the raw correction, regardless of the values of the correction. Thus, the iterative filtered correction outperforms the raw correction over all ocean regions and for any level of solar activity:

• Within the tropics and/or during years of strong solar activity (i.e. high correction value conditions),when the raw correction already reduces the variance of SLA observations, the iterative correction further improves them.

• At low latitudes and/or during years of weak solar activity (i.e. low correction value conditions), when the raw correction introduces additional variance to SLA observations, the iterative filtered correction can still improve (or at least not degrade) them.

Due to the use of the Lanczos filtering and Spline interpolation, the iterative filtered correction increases the number of valid correction values in the open ocean regions impacted by intense rain events. On the other hand, the number of edited values is substantially increased along the continental coast-lines and the Antarctic ice margins.

To measure this ionosphere-induced delay, satellite altimeters use observations from two signals emitted at the same time at two different frequencies (e.g. these correspond to the wavebands–Ku and –C in the case of the Jason and Sentinel-3 missions). Because of the frequency difference, the two signals are differently delayed as they cross the ionosphere. The ionospheric correction can thus be estimated as a function of the difference in delay between the two signals (hence the term “dual-frequency ionospheric correction”) after the Sea State Bias (SSB) corrections are applied. Since it includes measurements errors from two different frequency bands, as well as errors associated with the choice of the SSB correction algorithm applied, the ionospheric correction is inherently noisy. Furthermore, since it is directly proportional to the ionospheric total electron content (TEC) which is primarily controlled by solar radiation, the correction is characterized by large spatial and temporal variability. Such low frequency variability is primarily controlled by three main factors:

• Solar activity

• Local time

• Latitudinal location

While applying the raw ionospheric correction improves SLA observations when the correction values are large, it becomes ineffective (if not even detrimental) when the values are small due to the small signal to noise ratio. Since the instrumental noise primarily controls the high-frequency variability, it is required to low-pass filter the ionospheric correction to improve the overall quality of the final SLA products.

Here, we describe the iterative filtering approach used for the production of Sentinel-3 and Jason-3 GDR-F altimetry products. The filter was designed to achieve two goals:

• Base the correction on as many dual-band ionospheric observations as possible

• Improve the correction where altimetric observations are discontinuous or isolated

To achieve the first goal, the selection of the ionospheric observations used for the correction is independent from the quality of sea level observations. However, while this maximizes the number of observations selected, at the same time increases the number of potential outliers. The iterative filtering applies a median and a Lanczos filter in sequence, in order to progressively reduce the number of outliers in the ionospheric observations used to compute the final filtered correction. To achieve the second goal, since the filtered correction has long spatial correlation scales, a spline interpolation is applied to fill gaps in the interpolated correction up to few hundreds km.

The advantages of such approach were assessed by comparing the resulting sea level anomaly fields with those obtained using the unfiltered ionospheric correction. Our results show that the SLA variance reduction from the iterative filtered ionospheric correction is always larger than that from the raw correction, regardless of the values of the correction. Thus, the iterative filtered correction outperforms the raw correction over all ocean regions and for any level of solar activity:

• Within the tropics and/or during years of strong solar activity (i.e. high correction value conditions),when the raw correction already reduces the variance of SLA observations, the iterative correction further improves them.

• At low latitudes and/or during years of weak solar activity (i.e. low correction value conditions), when the raw correction introduces additional variance to SLA observations, the iterative filtered correction can still improve (or at least not degrade) them.

Due to the use of the Lanczos filtering and Spline interpolation, the iterative filtered correction increases the number of valid correction values in the open ocean regions impacted by intense rain events. On the other hand, the number of edited values is substantially increased along the continental coast-lines and the Antarctic ice margins.