Atmosphere and Ocean

Shona Mackie

Cloud Screening and Classification in Satellite Imagery

Clouds are important to both weather and climate, through their controlling influence on the planet's energy balance and through their role in transporting large volumes of water around the Earth. Different cloud types can affect weather and climate systems differently, depending on properties such as height and optical thickness. Such factors have to be taken into account in any model of climate or weather, for example in that used by the MET Office to produce a numerical weather forecast.

There are two problems associated with the treatment of clouds within such models. Firstly, the appearance of clouds must agree spatially and temporally with reality, and secondly, sub-grid scale processes within clouds must be represented. These are addressed to some degree using satellite imagery, visual inspection of which is currently used to make small amendments to the model's representation of clouds, bringing their appearance into line with reality. Work is currently being carried out into the automated assimilation of cloud fields from satellite imagery, and this is what my research will contribute towards. Currently, standard cloud detection is performed on imagery by threshold testing, which produces a binary mask of clear and cloudy pixels. Inaccuracies in the product are thought to stem from the method's lack of a sound physical foundation, and are thought to be both spatially and temporally variable.

The aim of my project is to expand the cloud detection technique developed by Dr. Chris Old, which currently produces a probability of cloud contamination for individual pixels based on the assimilation of background information from climatology and, where available, forecast fields, using Bayes Theorem. It is therefore physically based and exploits all the available information. I will expand the clear and cloudy classes produced by this technique to include cloud types, and will develop it to be as effective over land as it is over the ocean. To do this, I will look at various methods of clustering pixels based on their spectral characteristics and assigning probabilities to clusters rather than to individual pixels. This should address the problems associated with textural parameters within the method. It is hoped that the project will lead to cloud types being accurately and automatically identified within satellite imagery, and will be of operational use to the MET Office.

Shona's presentation

Christopher Nankervis

Water Vapour and Cloud from Satellite and the Earth's Radiation Balance

The Earth Observing System (EOS) comprises of a series of satellites in formation which follow precisely the same orbital path (often referred to as the A-Train). For the first time, global remote measurements of water vapour and cloud are now available in conjunction with outgoing radiation on a daily basis. This allows the Earth’s radiation budget to be examined in detail anywhere on the globe where the microwave limb sounder makes its measurements. The Microwave Limb sounder (MLS) is an instrument onboard the satellite Aura, which measures trace atmospheric constituents in the atmosphere’s edge with high vertical resolution and precision. This project makes use of data from instruments onboard two of the EOS satellites, Aqua and Aura, which make measurements just 7 minutes apart. This firstly involves matching the satellite measurements of same time and location. Water vapour is the most important greenhouse gas due to its broad infra-red absorption range and high concentration. Clouds also have a huge impact on the Earth’s radiation budget. Both water vapour and clouds are poorly understood and consequently remain a great uncertainty in radiation models. A warmer atmosphere has a far greater potential for maintaining water in the atmosphere as vapour, and it is the exponential relationship of water vapour to temperature which has led to reports of a possible run-away greenhouse effect.

Radiation models are used to simulate the radiances in infra-red water vapour bands. This is then compared to the radiances directly observed from collocated satellite measurements onboard Aqua instruments. Model inputs include 7 atmospheric gases and a temperature profile, all of which may be obtained from MLS retrievals. Initially fast models are used at set frequencies bands in the IR. For a detailed and more accurate analysis, line-by-line models are then manipulated, which are far slower at performing calculations.

Aqua instruments, MODIS and AIRS, make readings of thermal emission over the infrared region. These are both nadir viewing instruments onboard Aqua and therefore look directly down at the Earth’s surface below. There are just 16 MODIS bands in the IR, limiting the modelling of water vapour. However, AIRS has 2378 bands in the same region, which may be used for comparison with the modelled radiances. MLS makes measurements best above 6km where far – infrared water vapour bands dominate. Bands will therefore be selected which are relatively opaque to the outgoing radiation, with a low brightness temperature. This will allow a better understanding of the water vapour forcing in the higher reaches of the atmosphere. A cloud ice input may be used later in the project to gain a better understanding of the effect of clouds on the radiation budget.

Chris's presentation

Ana Lucia Viegas de Barros

Vegetation fires and Earth’s Atmosphere

Global warming since the Industrial Revolution has been mainly caused by the radiative forcing due to the accumulation of greenhouse gases in the atmosphere, which, in turn, come mainly from burning of vegetation and of fossil fuel.

Estimates of vegetation fire emissions require complex calculations and statistical methods applied to the analysis of data from satellite sensors, vegetation maps, bio-geo-chemical models and laboratory experiments.

The accuracy of future climatic predictions will greatly depend on the improvement of satellite sensors and statistical methods of satellite data analysis that will allow shorter time resolutions in global fire emissions inventories, on the development of fire models that reproduce the broad characteristics of wildland fires in different ecosystems --- their trend patterns, temporal and spatial variability, and on the establishment of statistical correlations between vegetation fires and natural climate forcings, such as lightning, the Southern Oscillation in the tropics, the Atlantic Oscillation in temperate woodlands and the Arctic Oscillation in boreal forests.

The resulting data will be essential for the future global chemistry-climate models to simulate the atmospheric chemical composition, explain the causes for the observed trends and variability in pollutant background concentrations, and predict the evolution of greenhouse gases concentration in the atmosphere.

Ana Lucia's presentation

Vivian Scott

Assessing Uncertainty in modelled estimates of the Ocean Carbon Sink

The ocean is largest of the dynamic carbon reservoirs and the only net carbon sink for anthropogenic CO2. Estimates of the size of this ocean carbon sink and hence its role in slowing global temperature rise are highly uncertain- between 1.5-2.1 PgC net annual uptake. The flux of carbon between the atmosphere and ocean is a complex two-way exchange arising from both physical and biological processes requiring extensive parametrisation to model. Understanding the relative importance of model parameters and their interactions is essential to focus further research in development of a new climatology at the Hadley Centre in the next two years. Beginning with simplistic models of phytoplankton growth sensitivity and uncertainty analysis techniques are applied to answer these questions

Vivian's presentation