Meetings: Documents

Aquarius Level 2 Algorithm: Recent Developments
[16-Apr-13] Meissner, T., Hilburn, K., Wentz, F., and Scott, J.
Presented at the 2013 SMOS-Aquarius Science Workshop
The Aquarius L-band radiometer/scatterometer system is designed to provide monthly salinity maps at 150 km spatial scale to an accuracy of 0.2 psu. The sensor was launched on June 10, 2011, aboard the Argentine CONAE SAC-D spacecraft. The L-band radiometers and the scatterometer have been taking science data observations since August 25, 2011.
Our presentation describes recent developments in the Aquarius Level 2 salinity retrieval algorithm, which is run in the Aquarius Data Processing System ADPS at Goddard Space Flight Center. We discuss the most important advances in the algorithm that have occurred since the currently released ADPS version 2.0.
Specific topics are:
  1. Correction for effects of the wind roughened ocean surface.
  2. Correction for intruding celestial radiation, foremost from the galaxy.
  3. Use of the SAC-D K/Ka-band microwave radiometer MWR for correcting atmospheric liquid water absorption and rain flagging.
The most important improvement in the performance of the salinity retrieval is due to an update in the surface roughness correction (item 1). This effect turns out to be a major driver in the salinity retrieval error budget. The surface roughness correction that is used in the current version 2.0 ADPS data product is based on auxiliary input of wind speed and direction fields from NCEP and VV-pol backscatter measurements form the Aquarius L-band scatterometer.
The accuracy of the surface roughness correction improves significantly if one uses an Aquarius retrieved wind speed instead of the NCEP fields. This Aquarius wind speed is derived from a combination of HH-pol scatterometer and H-pol radiometer observations and assimilates the NCEP wind speed as background field. We have validated this Aquarius wind speed and shown that its accuracy is about 0.7 m/s, which matches the accuracy of other satellite derived wind speed products, e.g. from SSM/I or WindSat. It is a major improvement compared to the NCEP wind speed, whose accuracy is only about 1.2 m/s. This increased accuracy of the surface wind speed, which is the crucial input to the surface roughness correction algorithm, leads in turn to more accurate salinity values. Further smaller improvements in the performance of the surface roughness correction can be achieved by including significant wave height data from the NOAA Wave Watch 3 Model as auxiliary input. The RMS between the Aquarius salinity and the HYCOM reference salinity field drops from about 0.55 psu in the current version 2.0 to about 0.41 psu in the new version.
The Aquarius salinity data exhibit biases between the ascending and descending swaths during certain times and in certain locations. The main cause for these biases is the intrusion of reflected galactic radiation (item 2). The algorithm that is employed in the current version 2.0 data release leads to an overcorrection at low wind speeds and an undercorrection at high wind speeds. The situation improves significantly in the new algorithm with the use of an Aquarius derived wind speed instead of the NCEP wind speed in the correction for the reflected galactic radiation. Nevertheless, residual errors in the correction for the reflected galactic radiation and accordingly some of the ascending - descending biases remain also in the new L2 algorithm. We will analyze the reasons and discuss possible mitigations.
The MWR, on-board SAC-D, provides spatially and timely collocated observations with Aquarius at 23.8 GHz and 36.5 GHz after appropriate resampling (item 3). These observations can be used to calculate the atmospheric absorption due to cloud liquid water at L-band, which needs to be removed in the Level 2 retrieval algorithm. They also provide a method to flag the Aquarius observations for rain. The availability and use of the real-time MWR data lead to a small additional improvement of the Aquarius salinities over the current version 2.0 data, which utilize NCEP cloud water profiles as auxiliary input for this purpose. Our talk gives an overview over the MWR cloud liquid water retrieval algorithm and its application in the Aquarius L2 processing.
The last part of our presentation contains a validation study for the new salinity product, which consists in a 3-way intercomparison between Aquarius, SMOS and in-situ salinity measurements from moored RAMA, TRITON, TAO, and PIRATA buoys. The Aquarius - moored buoy comparison shows that that the new Aquarius salinity meets the aforementioned mission requirement of 0.2 psu. By comparing SMOS and Aquarius salinity fields we assess the importance of the roughness correction and the presence of the L-band scatterometer for reaching this level of accuracy, which is a major difference between the SMOS and Aquarius missions.

View Document (PDF, 11.1 MB)