Au_MultiSensor1.1 notes(CSIRO Marine-Remote Sensing-Ocean Currents )

Notes on Multisensor 1.1 (AMSR-E, ATSR-2, AVHRR, Reynolds) SST

Notable images

26/10/93 cold around PNG (drifters agree)
4/11/97 El Nino time again
31/3/94 cold Tasman
22/8/94 upwelling off Java
8/3/95 Upwelling off Robe and Lakes Entrance
19/4/95 cold Adelaide-Sydney
6/6/95 still cold, NB diversion of East Australian Current
18/6/95 ..as documented by a drifter
21/3/97impact of Tropical Cyclone Justin off QLD
26/6/98 The biggest data error I've spotted so far (off Qld)
5/1/00 big warm area SW of WA
18/3/01 cold E Indian (NB some Argo floats), warm elsewhere
10/7/01 warm off Sydney and SW of Tas
1/12/01 warm again in the south
25/12/02 another warm summer, strong EAC. Many Argo floats now
13/4/03 an Argo float samples TC Inigo
7/5/03 and the first Argo float near Tasmania samples an extemely warm eddy
26/3/04 TC Faye off NWS. NB No Leeuwin

Production

This is a very preliminary version of a merged AMSR-E, ATSR-2, AVHRR, surface_obs product that we are currently working on as part of our contribution to the GODAE High-Resolution SST Pilot Project (GHRSST-PP). The goal of the project is to fully exploit the relative strengths of different sources of estimates of sea surface temperature to produce an optimal combined product.

The AVHRR observations offer the greatest frequency of sampling and wide coverage but suffer from gaps and errors due to atmospheric effects. The ATSR-2 observations offer the greatest accuracy (by viewing the same patch of ocean from multiple angles through the atmosphere) but the sampling frequency and coverage are far from complete. Finally, the AMSR-E microwave radiometer has the coarsest resolution of all, but has the great advantage of being able to see through cloud. In-situ observations are free of atmospheric effects and can be very frequent but are only, of course, available from a limited number of locations.

This preliminary product is to satisfy an immediate need for SST estimates suitable for assimilation into our hydrodynamic model. For this application, being free of large errors is a higher priority than having the greatest accuracy at high resolution. Accordingly, we have constructed a 'patchwork quilt' using the following, listed in preference order:

a six-day average of the AMSR-E data (supplied by Remote Sensing Systems)
a 12-day average of the ATSR-2 observations (provided by ESA via the Rutherford Appleton Laboratory (UK) (with 0.2degrees added to make it more compatible with the way AVHRR data has traditionally been calibrated)
our 15-day histogram-composite of Australian LAC AVHRR data, but only where and when this is formed from a sufficiently large number of cloud-cleared estimates
Reynold's weekly-averaged SST product.

The 'seasonal anomaly' is the difference from the CARS2000 monthly climatology, linearly interpolated to the day of the SST estimate.

Revision history:

Version 1.0: May 2004. 1993-2001
Version 1.1: July 2004 1993-2004, added AMSR-E.

Enquiries or comments (but please note the word preliminary used above:

Ian Barton