Met Office: Satellite Microwave Radiances

	Satellite microwave radiances

Advantages of microwave wavelengths for atmospheric sounding

The microwave spectrum is the region where the wavelength of light is a few millemetres, compared to micrometres in the infrared and hundreds of nanometres in the visible. These long wavelengths have one key advantage over the shorter wavelengths: clouds are semi-transparent. For wavelengths in excess of 3 cm clouds are almost completely transparent, but unfortunately there are no useful spectral lines between 3 and 10 cm wavelength to allow atmospheric sounding, and as wavelength increases so does the size of antenna required on the spacecraft to obtain high-resolution images. Cloud opacity increases with decreasing wavelength below 3 cm. The first spectral band which can be used for temperature sounding is reached at approximately 5 mm, a complex band of oxygen spectral lines. At this wavelength the effect of ice cloud can still be largely ignored, with the exception of very deep cumulonimbus clouds. However, the effects of water drops can not be ignored. Nonetheless the situation is far better than in the infrared, where any cloud, no matter how thin or high, makes the data difficult and in many cases impossible to exploit.

Use of microwave sounding data at the Met Office

The ATOVS series of instruments carries a 20-channel microwave sounder which exploits the 5 mm oxygen band for temperature, and also water vapour lines at 12 mm and 1.6 mm. When we refer to a channel we mean the measurement of the electromagnetic spectrum in a narrow wavelength region. By measuring where oxygen absorption is strong we are sensitive to the upper-atmospheric temperature. By measuring where it is weak we are sensitive to lower atmospheric temperature. By measuring in a number of discrete wavelength bands, or channels, we gain information on the vertical profile of temperature. The same technique can be applied to measuring water vapour, and indeed other atmospheric constituents such as ozone, carbon dioxide, CFCs etc. The 20 channel microwave instrument we use is known as the Advanced Microwave Sounding Unit (AMSU), and was first launched in 1998 and used operationally as a key component of the global observing system just 8 months later. It is "Advanced" with respect to its predecessor which measured only 4 channels, all close to the 5 mm oxygen band.

In the early days of atmospheric sounding, the radiance measurements were processed to produce an estimate of the atmospheric temperature profile. This is known as a retrieval (of temperature). These retrievals were then used as if they were temperature profile measurements made by instruments carried by balloons (radiosondes). However this was far from the best way to use the data. The measurements are very different. The satellite data measures radiance. These radiances are sensitive to the atmospheric temperature in deep layers. Typically the fields of view are 50 km wide, and a single radiance measurement can be affected by temperature variations over depths of 3-5 km, and thus can not uniquely resolve features which are thinner vertically than this depth. By contrast radiosondes can resolve very fine structure in the vertical, but are measuring the temperature of a very small sample of air, which may not necessarily be representative of a wider area. Both are accurate measurements but if the data is used in way which does not take into account what was actually measured in each case then some, and perhaps even most, of the information in the measurement can be lost. So the differences between different types of measurement need to be recognised in the way the data is used, because both have strengths and weaknesses. A method has been developed known as direct radiance assimilation, whereby we compare the radiance measurement with what we would have expected to measure, given everything we already know about the atmosphere. Our knowledge of the atmospheric state is then updated only where the two disagree, and only on the scales which the satellite measurement can actually resolve. This approach has very significantly increased the value of satellite data to weather forecasting.

Impact of microwave sounding data

We regularly measure the impact of the microwave sounding data on the accuracy of our weather forecasts. As would be expected satellite data has the largest impact where there are few other observations. Forecast errors would be doubled in the southern hemisphere for example if satellite radiances were unavailable. However, in recent years the importance of satellite data in regions where other data are available has increased. This is to be expected given the differences in the type of information provided by satellite radiances and other observations is now recognised in the way we use the data. It is now found that the satellite data has an impact comparable or even larger than the radiosondes even in the northern hemisphere, where other observations are often in plentiful supply (although even the northern hemisphere has data voids, especially over the oceans).

Research activities in assimilation of microwave radiances

Precipitation assimilation

Improved radiative transfer models