The majority of climate change research using AATSR data will be performed by major meteorological users such as the Hadley Centre and the BoM. The Hadley Centre for Climate Prediction and Research, located within the Met Office with financial support from DECC, provides a focus in the United Kingdom for the scientific issues associated with climate change.

The main aims of the Hadley Centre are (from http://www.met-office.gov.uk/research/hadleycentre/):

• To understand physical, chemical and biological processes within the climate system and develop state-of-the-art climate models which represent them
• To use climate models to simulate global and regional climate variability and change over the last 100 years and to predict changes over the next 100 years
• To monitor global and national climate variability and change
• To attribute recent changes in climate to specific factors
• To understand, with the aim of predicting, the natural inter-annual to decadal variability of climate
  

The Hadley Centre activities can be separated into three parts, namely:

• Climate monitoring and climate datasets
  • The Hadley Centre receives, quality controls, and archives large amounts of observed climate data. These are used for monitoring the climate, in studies of the causes of climate change, and in climate modelling.
• Climate models used in the Hadley Centre
  • Detailed three-dimensional representations of major components of the climate system are mostly run on the Met Office's Cray T3E supercomputers in various configurations.
• Climate change projections
  • Key results are obtained from climate-change experiments conducted using Hadley Centre computer models of the climate system.

It is expected that AATSR data will be useful in all three of these areas as they are mutually dependent. So far only the usability of AATSR SST data has been investigated; the usability of AATSR LST data (and indeed other climate products such as cloud parameters) will be the subject of ongoing discussions with the Hadley Centre.

There are a number of ways in which AATSR data are relevant to the work of the Hadley Centre. An increasing strength of the AATSR data is the long term nature of the record when considered as a long term data set in conjunction with the observations of sea surface temperature (SST) by ATSR-1 and ATSR-2.

Discussions with Hadley Centre personnel have identified the following key areas which are of importance to the centre and where accurate SST data, such as that available from AATSR, can contribute to significant aspects of climate research.

The Hadley Centre sea ice and sea surface temperature climatology, HadISST1 (version 1), is unique among available integrated SST and sea ice analyses in being globally complete whilst spanning well over a century [RD2]. The dataset currently includes, in terms of satellite data, bias-adjusted SSTs from the Advanced Very High Resolution Radiometer (AVHRR). As the (A)ATSR data set emerges as a long term dataset, possible benefits of this data set include:

• Validation of HadISST (as a first step)
• Combination with AVHRR as an input data set, taking advantage of the complementary qualities of the two data sets
• Information on the accuracy of satellite data sets and determining the rationale for any potential bias adjustments
• Improved information on sea ice/sea surface temperature relationships, which would involve further research into new retrieval schemes.

The HadISST1 data set itself is primarily employed as a tool for validating atmosphere-ocean models, as a tool for atmospheric model forcing and as a data set for climate variability studies. AATSR data are currently being evaluated within the Hadley Centre, with a view to incorporating them systematically into HadISST1

It was mentioned above that some climate change research using AATSR data will be carried out in responsive-mode or otherwise supported. A very relevant example of this type of independent climate change research is being carried out by Lawrence et al at the University of Leicester. An initial analysis of the existing ATSR-1 and ATSR-2 data-sets, using the currently available versions of their global SST fields (designated ‘version’ 1) has shown that, by removing the annual cycle and some of the natural variability (the El Niño signal) from the time-series, it is possible, with little more than 8 years of ATSR-standard data, to detect a rate of temperature increase that is compatible with model predictions. This methodology, differs from the approach used by the HC in that it isolates clearly identifiable natural variability from the data before examining it for trends, has great potential and it will be of great interest to apply it to the 15-year ‘version 2’ data-set that will be available when the new consolidated archive will be complete. It is also of interest to apply the techniques developed by Lawrence to other parameters such as NDVI, because Peter Cox from the Hadley Centre has recently shown interesting anomaly fields with significant correlation with the El Niño and the Lawrence methodology explicitly extracts the El Niño-correlated components of any time series.

Trends in regional sea surface temperature patterns are fundamental to the climate system not only because regional patterns can possibly provide early indications of climate change which is occurring (so-called “fingerprinting”) but also because evolving SST changes in certain regions can influence the pattern of climate change experienced in a different region (teleconnections). Parameters that are significant include both mean values and patterns of SST values in the region. In effect, gradients of SST in the pattern can be significant.

The teleconnections aspect is clearly of significance and bears directly on the issues of seasonal, inter-annual and decadal forecasting. An excellent example is the relation of rainfall to SSTs in the tropical Atlantic. Rowell and colleagues have shown in a number of papers [RD3-7] that by using coupled ocean-atmosphere models forced with global SSTs, they can simulate the magnitude and patterns of seasonal rainfall across a number of regions of the globe, including the Sahel, Soudan and Guinea Coast regions on the Atlantic coast of North Africa thus confirming the suggestion that global SST patterns are closely linked to much of the variability in seasonal rainfall. In addition, they have also shown that the likelihood of a drought in the Sahel region in particular, is increased during a classic ENSO period owing in part to direct atmospheric teleconnections between the east Pacific Ocean and the tropical Atlantic. Further they noted that the reduction, during an ENSO event, of the large-scale SST gradient between the west Pacific and the eastern Indian Ocean enhanced the likelihood of a drought in the Sahel.

Regional SST patterns may influence our knowledge of climate change through a number of factors including:

1. Observations of trends in regional SST and SST patterns (regional gradients) leading to improved characterisation of the patterns of climate change and temperature trends.
2. Improvements in the accuracy of SST measurements which may lead to better understanding of teleconnection patterns related to climate change (although other sources of variability may be limiting factors)
3. Monitoring of changes in regional SST patterns with high teleconnection significance leading to improved climate forecasts for impacts in related regions
4. Observations of changing correlations between climate impact parameters and SST, indicating, for example, a change in atmospheric circulation.

It has been shown that oceanic forcing by SST/sea ice has a statistically significant, and therefore predictable, influence on inter-annual variability over 80%-90% of the globe, as characterised by mean sea level pressure and rainfall in model studies. Many regions also have a substantial annual cycle of predictability. Accurate monitoring of SST may therefore be of significance in improving seasonal forecasting of climate.

Studies have shown that decadal climate predictability is largely restricted to the tropics and some extra tropical regions. Knowledge of SST trends and variability on the tropics is probably most significant for this area both in terms of research characterising the tropical sea surface state and for developing realistic models of trends in SST patterns for decadal predictions. In the extra-tropics, the exact importance of oceanic forcing is less clear and further research is probably necessary. The extent of research into relations between sea ice/polar SST changes and regional climate is less clear.

A second use for SST in decadal predictability is in exploiting SST measurements to accurately prescribe the sea surface state and to constrain the behaviour of the mixed layer at depth in the ocean. Ideally, direct observations of the vertical distribution of temperature with depth in the ocean would be available throughout the globe but this is not the case. Therefore recent research has concentrated on applying ocean surface data, via model correlations, to infer and characterise components of sub-surface anomalies which contribute to surface variability. In this way, important ocean dynamical signals can be obtained.

Operational oceanography would exploit accurate sea surface temperature through data assimilation techniques applied to near real-time data. At the current time, only limited research has been performed on the exploitation of sea surface temperature data for these systems. However, it is expected that on-going systems such as the Forecasting Ocean Assimilation Model (FOAM) will incorporate AATSR SST data in the future. Further consideration of this area is recommended.

A further development is the incorporation of ocean biology into some oceanographic systems such as the Hadley Centre Oceanic Model (HadOc). Biological sensitivity to SST can be of prime importance and indeed was one of the key early identifiers of El Niño’s impact on regional economies. Both near real-time and longer term forecasting systems may have significant requirements for accurate and consistent SST data sets.

If SST is to be used as a true indicator of climate change, it is important that ocean processes that have a strong SST signature are well understood. In particularly the natural variability associated with such processes needs to be quantified. Major processes with the potential to perturb the global SST signature include El Niño, the Somali upwelling, the Gulf Stream, the Kuroshio Current and the Agoulhas Current. Research into the behaviour of such phenomena, particularly with respect to quantifying their intensity and geographical extent, should receive high priority. A highly relevant scientific question concerns the relationship between global SST and heat content of the oceans. In particular, as the oceans warm, is it appropriate to assume that the relationship between SST and heat content remains constant? Intuitively, increased heat input to the oceans should lead to increased vertical mixing and a changed relationship between SST and heat content.

This section describes scientific research into land surface processes that may benfit from using AATSR data. The AATSR can provide high-quality data on Land Surface Temperature (LST), on the reflective visible and emitted infrared properties of the land surface and it produces a state-of-the-art vegetation index product which will provide information on vegetation dynamics. The following sections highlight specific land processes where exploitation of data from the AATSR can make a significant contribution.

Information on the reflective visible properties of the land surface are obtained from the three visible channels on AATSR, at 0.56 μm, 0.67 μm and 0.85 μm. In addition, a state-of-the-art vegetation index product is produced from the ratio of the 0.67 μm and 0.85 μm reflectances. The following sections highlight specific land processes where exploitation of data from the AATSR can make a significant contribution to vegetation monitoring.

The Earth's forests play an important role in absorbing carbon dioxide from the planet's atmosphere, and their destruction will contribute to the Greenhouse Effect. On a local scale, deforestation has had dramatic effects on the climate with the combination of reduced rainfall and soil erosion compromising attempts at agricultural use of the cleared land. Clear evidence of the anthropogenic origin of this destruction is the regularity and linearity of the inroads that have been made into the forest. Many remarkably straight tracks are clearly seen in ATSR-2 images. These cleared areas show up brighter than the surrounding vegetation because they have different thermal properties and can be up to 4K warmer during the day. Ground-based estimates of the scale and rate of deforestation are notoriously inaccurate. However satellite images provide a reliable and convenient method for long-term global monitoring of this phenomenon.

The 1980's were the worst decade for volcanic disasters in the twentieth century, with 24000 - 28000 fatalities each associated with two particularly devastating eruptions, that of El Chichón (Mexico, 1982) and Nevado del Ruiz (Columbia, 1985). Such tragedy clearly shows that active volcanoes continue to represent extreme hazards, despite advances in the technology available for ground-based monitoring of pre-eruptive volcanic phenomena. In addition, there is the problem of the sheer number of potentially active volcanoes whose monitoring needs to be addressed. The Catalogue of Active Volcanoes (CAVW, 1951-1975) documents over 500 volcanoes that have had recently dated eruptions and, on average, more than 50 eruptions occur annually. With such a huge number of potentially active volcanoes, traditional monitoring techniques such as seismic and microgravity require assistance if all these targets are to be kept under surveillance.

The AATSR series is well placed to assist in the development of techniques for volcano monitoring since it views all of Earth's terrestrial volcanoes once every three days under night-time conditions. AATSR measures the amount of thermal energy being emitted from the Earth's surface, including all terrestrial volcanoes, and these volcanic measurements can be related to the amount of high temperature activity occurring at centres of known eruptive activity.

During night-time observations (i.e. in the absence of sunlight) AATSR measures the amount of thermal radiation arriving at the instrument in four different wavebands (1.6, 3.7, 11 and 12 µm), making measurements on a 1 x 1 km grid over the entire Earth surface. The relationship between the temperature of the Earth's surface and the amount of emitted radiant energy is governed by a theoretical relationship known as the Planck Equation. Understood simply this relationship indicates that the wavelength of peak energy emission decreases as the temperature of the surface increases.

So, whilst AATSR's longer wavelength channels are good for looking at energy emitted from ambient temperature surfaces such as the ocean, AATSR's short wavelength channel (1.6 µm) can be used to observe the significant amounts of energy emitted from very hot surfaces such as those found at active volcanoes. Since surfaces at temperatures less than around 300 °C do not emit significantly at 1.6 µm, any night-time signal at this wavelength is evidence of high temperature activity at the location of interest. By monitoring this volcanic signal over time these measurements can be used to deduce changes in the nature of the high temperature activity and these inferences used to assist monitoring of the eruptive and pre-eruptive phenomena.

The (A)ATSR series of instrument provide frequent observations of the Arctic and Antarctic region. Information from these observations can be used for qualitative physical monitoring of icebergs, providing information on their spatial distribution, of their calving, breakage and melt rates, and of their movement. These icebergs represent a major component of the mass discharge from the both the Greenland and Antarctic ice sheets, and hence of the overall mass budget of the ice sheet. In addition, images from the (A)ATSR series can be used for long term monitoring of sea ice extent. Such information is an important input to the many climate change scenarios.

Clouds and aerosols are an important consideration in AATSR retrievals of SST with consequent requirements for cloud masking, and for the correction of aerosol/thin cloud contributions to the observed brightness temperatures. Understanding of these effects is therefore an important consideration for SST retrieval.

The importance of clouds and aerosols in moderating or amplifying radiative forcing is generally accepted. It is also generally accepted that our knowledge of cloud dynamic and radiative properties falls well short of that required by modern climate analyses and prediction schemes. Once the practical priority of identifying the presence of clouds in order to retrieve surface temperature has been satisfied there is much scope for using AATSR’s multi-angle multi-wavelength viewing geometry to characterise and investigate the properties of clouds.

The sources of aerosols are diverse, ranging from large-scale natural events such as volcanic eruptions to desert storms, biomass burning and anthropogenic sources associated with industrial pollution and agriculture. The AATSR, on account of its unique dual angle viewing geometry, is especially sensitive to atmospheric aerosol and there is great potential for using AATSR data, generally in combination with data from other sources, to examine and quantify the radiative properties of atmospheric aerosols.
Aerosols and clouds are also related through the potential of aerosols to modify cloud droplet growth and hence affect both cloudiness and precipitation in magnitude and location. This aerosol-cloud interaction can play a significant role in both the direct radiative balance of the atmosphere and in the hydrological cycle. The implications for both the understanding and prediction of future climate change are significant.

Aerosols influence both the visible and infra-red channels of the (A)ATSR instruments, with larger effects in the forward than in the nadir view. The challenges of aerosol retrieval are such that current global algorithms tend to use nadir-only views of the visible channels. More restricted applications on regional scales have also shown the benefits of utilising the dual and forward views but considerable care is required over land. Further possibilities to differentiate large particle sources, such as Saharan dust, using the infra-red channels increase the likely success of differentiating aerosol types with AATSR data.

The importance of clouds to the radiation balance of the terrestrial climate is well known. Future and current climate simulations, using state-of-the-art models, are very sensitive to changes in the current cloud parameterisation schemes. Indeed, models can even introduce compensating errors that hide additional sensitivities to certain parameters. Accordingly, the scientific community has put an imperative on the validation of these cloud parameterisations by confronting the model simulations with observations. While high-resolution measurements obtained in experimental campaigns are necessary to develop the parameterisations, the evaluation of model cloudiness requires comparison with global climatological data. Such climatological comparisons can highlight specific areas of disagreement, but do not always explain the reasons why the observations and models disagree as other model problems can manifest in the simulated cloudiness. For comparison with climate models, observational studies that are restricted seasonally and/or spatially to identify specific synoptic regimes are becoming more important.

Despite the apparent proliferation of cloud data, the information about cloud properties is often limited to frequency information and optical thickness along with environmental data (e.g. cloud top temperature and pressure). While there is relatively good agreement in some cases between sensors for gross measures of cloudiness (e.g. seasonal and zonal means) there is still considerable disagreement in detail. Long term global data sets of cloud optical and physical parameters are essential. The required long term cloud data can be obtained from the long term data set of visible reflectances from 1995 until the present day, from ATSR-2 and AATSR. Several different algorithms have been developed over the years and one of these forms the basis of the GRAPE project, led by the University of Oxford, that will initially analyse the ATSR-2 dataset with future data from the AATSR added at a later stage.

The GRAPE project intends to produce a new cloud database which will include the following parameters, along with associated error measurements (enabling the use of this data in some form of data assimilation at a later date):

• Cloud Optical Depth
• Cloud Phase
• Cloud Particle Size
• Cloud Top Pressure
• Cloud Fraction
• Cloud Water Path