ABSTRACT Conversion of ATSR-2 Sea Surface Skin to Bulk Temperature-USB迷|专注于互联网分享

2024年4月1日发(作者：宝星儿)

Conversion of ATSR-2 Sea Surface Skin to Bulk Temperature

for Use in Climate Studies

Anne O'CARROLL, Brett CANDY, Roger SAUNDERS

The Met. Office, London Road, Bracknell, RG12 2SZ, UK

Email: agocarroll@, bcandy@

, rwsaunders@

ABSTRACT

Over three years of bulk SSTs were derived using ATSR-2 cloud-cleared skin SSTs (averaged at a resolution of 0.5

degrees), using a skin to bulk temperature model. The algorithm is based on a physical skin model and covers all

surface wind speed ranges. The near-surface diurnal thermocline is also modelled with the aim of rejecting those

observations likely to have a strong thermocline.

This paper presents the results from comparisons of bulk SST with a climate dataset containing SSTs at 1 degree

resolution. The dataset is maintained by the Hadley Centre at the Met. Office, and contains SST observations gathered

from a combination of data including in-situ and Advanced Very High Resolution Radiometer (AVHRR) observations.

INTRODUCTION

The high accuracy and almost ten year time period of sea surface skin temperatures retrieved from the Along Track

Scanning Radiometer (ATSR) series of instruments make them potentially useful observations for climate studies.

However, for incorporation in current sea surface temperature (SST) analyses they need to be converted to represent

bulk SST at a depth of around one metre in order to be consistent with other SST observations used such as from ships

and buoys.

Cloud-cleared skin SSTs derived from ATSR-2 data were obtained from the Rutherford Appleton Laboratory (RAL) for

June 1995 to February 1999. They were processed by RAL into a half degree product using the SADIST-2 scheme [1].

No data is available from January 1996 to June 1996 due to scan-mirror problems on the ATSR-2 instrument.

The skin SSTs for the whole period have been converted into bulk SST using a processing scheme based on the Fairall

skin model [2] and the Kraus and Turner [3] thermocline model. A comparison can be made to check the behaviour /

accuracy of these bulk SSTs by comparing them with SSTs from buoys, which measure the bulk temperature of the

ocean under the surface skin layer.

In this study, comparisons of the ATSR-2 bulk SSTs have been made with SSTs from buoys at fixed moorings which

report on the GTS, and additionally, buoys from the Tropical Atmosphere Ocean (TAO) project in the tropical Pacific

Ocean [4] which are also at fixed positions. No comparisons have been made using drifting buoys as experience has

shown these measurements are more likely to be subject to larger errors. Additionally, global comparisons have been

made between ATSR-2 bulk SST and the Global Ice and Sea Surface Temperature dataset (GISST) [5]. GISST is a

climate dataset created and maintained by the Hadley Centre. It is based on in-situ and AVHRR SST data, is at a

resolution of 1 degree and covers the period from 1961 to the present.

The aims of this study are to look at the characteristics and accuracy of the ATSR-2 bulk SSTs with respect to buoy sea

surface temperatures and to examine any patterns emerging. The comparisons against the GISST dataset provide

interesting insights due to the higher resolution and high spatial coverage of the satellite based observations. It is also

instructive to look at the behaviour of the ATSR-2 SSTs over several years duration compared to the GISST dataset.

METHODOLOGY

Outline of How the Skin SSTs were Produced

The ATSR-2 skin SSTs used in this study were retrieved by the Rutherford Appleton Laboratory (RAL) from top-of-

atmosphere radiances and coefficients determined by linear regression from brightness temperatures simulated using a

radiative transfer model [6]. The coefficients are chosen to minimise s. difference between the prescribed skin

SSTs and retrieved skin SSTs in a diverse atmospheric profile dataset. During the day radiances from the nadir and

forward views of the 11 and 12 micron channels were used to derive the skin SST and at night the 3.7 micron channel

radiances are also used from both views with a different set of coefficients. The latter channel is more transparent to

water vapour and so can be used to infer more accurate skin SSTs in the tropics.

Details on the Production of the Bulk Product

The ATSR-2 skin SSTs have been used to generate a global product of bulk SST using a scheme developed at the Met.

Office. The scheme involves a series of operations. First NWP model fields such as radiative surface flux and surface

wind data are interpolated to provide an estimate of surface conditions for each observation. Then, the conversion of

skin to bulk temperature can be performed using a skin-layer model. The skin model attempts to estimate the

temperature difference across the ocean skin layer using the model surface conditions and the retrieved satellite SST.

Various models have been proposed to describe the temperature difference across the oceanic skin layer and for our

purpose we require one which is applicable across all ocean conditions. Saunders [7] derived a theoretical model for

forced convection conditions by treating the skin layer as a fixed conduction layer whose depth depends on windspeed.

It has the following form:

∆T=

κρ

where Q represents the heat flux out of the ocean, ν

∗

is the friction velocity in water, κ is the thermal diffusivity, ρ is the

density of water, c

is the specific heat capacity of water,

is the kinematic viscosity, and λ is a dimensionless

constant. Fairall [6] extended the model to cover low windspeed conditions by redefining λ as a function of flux and

friction velocity:

f(v

,Q)

where λ

is the value that λ

will tend to as the free convection regime is approached. λ

has been estimated by various

authors to be between 5 & 10. For the night-time a value of 4 was found to be optimal, and for the day-time a value of

7.5. Different values of λ

are use for day and night as different SST retrieval coefficients are used for the different day

regimes. We can use λ

as a tuning constant to remove the residual bias between in-situ buoy SSTs and converted

satellite bulk SSTs.

Fig. 1a shows the behaviour of the skin-layer model over a range of windspeeds. The example is for night-time and has

been compiled using 1000 matchups of buoy SSTs with satellite SSTs. The main points to be emphasized are:

•

A reduction in the observed delta T is observed with windspeed which follows the Fairall model.

Fig. 1b displays day-time results compiled from 800 matchups and shows that the same pattern as seen in the

night-time plot can also be seen but to a greater extent.

A subset of quality-controlled drifters were also used, where both the mean and standard deviation of the observed buoy

SST minus background SST values for a buoy are less than 1K, and are thus considered to have good sensor calibration.

These drifting buoys measure the temperature closer to the surface and gave better results in terms of the correlation of

the observed delta T to predicted delta T. A thermocline model is used under day-time conditions, to predict the effects

of diurnal heating. High insolation and low windspeed conditions may lead to the possibility of a thermocline

developing, where a large temperature difference (delta T) below the skin layer up to a few deg K may occur.