How are the images created ?

 In the beginning, when images from satellites were first thought about, the procedure consisted of installing a camera in the satellite from where the pictures were taken. The satellite (or at least part of it) would then be returned to Earth, hopefully reasonably intact, bringing the camera & film with it from which the pictures would be obtained. Not only was this a bit hit & miss (often miss in the early days), but clearly quite impractical from the weather observation point of view, and it remained confined mainly to military applications.

The first steps to useful weather observation came with the launch by NASA in the 1960s of the TIROS series [Television InfraRed Observational Satellite]. The first of these satellites incorporated a television camera and indeed it became possible to view weather developments as they occurred. However, though successful in its way, this method suffered from several drawbacks. One was the vulnerability of the television cameras themselves, being prone to solar radiation damage after a relatively short operational period with consequent degredation of the pictures.

More serious though was the limitation on the kind of data obtainable from the images themselves. For weather monitoring purposes, it's not sufficient just to view cloud patterns and evolution, useful though this is - temperature data both of cloud and surface under different conditions and a means of estimation water vapour content is also needed. All of this is quite difficult to obtain from a single television camera, quite apart from the limits of resolution imposed by the scanning procedures and the demands on the bandwidth of the downward link. A better system was required.

To understand how the solution was arrived at, it's worthwhile considering for a moment just what sort of information is needed for weather monitoring. Some requirements can be identified without difficulty:  cloud cover - its depth, area & water vapour content, wind directions and velocity at differing altitutes, air, sea and land surface temperatures for example. Not all of these are immediately apparent by simply looking. Clouds we can see by visible light when it's there, but temperature assessment for instance requires more subtlety.

All bodies in the universe radiate electromagnetic radiation of some sort by virtue of motion of charged particles in the atoms and molecules of which they are made up. Such radiation is sometimes termed black body radiation. The electromagnetic radiation is a wave motion and is characterised by its wavelength. It spreads out from all radiating bodies like ripples in a pond into which a stone has been dropped. The distance between one crest and the next is the wavelength. The radiation carries energy with it, other things being equal, increasing with decreasing wavelength. Furthermore, the more quickly the particles of matter are moving, the more energetic the radiation - we perceive this as heat, itself indicated by what we measure as the body temperature. The energy distribution of a black body radiator with respect to wavelength looks something like Figure I. The distribution rises to a peak at some wavelength which moves to ever shorter values as the temperature rises.

We can put this into some perspective looking at Figure II which displays a wider gamut of electromagnetic radiation and where the various types fit in. At the metre length we have radio waves familiarly used to transmit broadcast radio, television and other telecommunications (including the telemetry from the satellites...) This gives way to microwaves (familiar in the kitchen - the energy content is now increasing) and also used for radar and other ranging applications. It should be noted too that as the wavelength decreases, the radiation tends to become more directional in that if generated as a directed beam, it tends to stay that way and not spread out with increasing distance. After microwaves, we get to the interesting bit. Here we have the well known region of the infra-red, visible & ultra-violet spectrum. This is the part of most value in weather monitoring. The regions beyond belong to increasingly energetic radiation - X-rays and gamma rays characteristic of atomic nuclear processes; important they certainly are but not here.

Figure III gives a closer look at the visible region i.e. the light by which we see the world. If we wish to record visible features, this is the part of the spectrum we need to use.

However, we can only do this by virtue of reflected light from the sun since the Earth is far too cold to emit light on its own account in any significant amount. The characteristic tempeature of the Sun is of the order of 5500°K whilst that of the Earth is around 300°K. [NB °K = °C + 273.1] At that temperature however, the Earth does emit significant amounts of infra-red radiation, the strength of which depends on the local temperature on its surface. Thus by measuring infra-red radiation intensity, we can remotely obtain the surface temperature at any point. This principle has been used industrially for many years to obtain temperatures of dangerously hot bodies, but equally it's the same general idea behind how weather satellites obtain some of their information.

Modern satellites use a radiometer of some kind which incorporates sensors receptive to the particular wavelength of interest. The modern US NOAA series of satellites use an Advanced Very High Resolution Radiometer {AVHRR] for this purpose. It looks [before installation into the satellite) something like this.

Figure IV should give some idea how it works. A Cassegrian reflecting telescope is mounted with its axis along the satellite direction of travel. The Earth beneath is viewed via a mirror mounted at 45° to the axis about which it rotates at six times per second. The telescope view is relatively narrow so that as the mirror rotates, the field is swept across a very narrow strip perpendicular to the direction of travel. The satellite is moving so successive strips lie adjacent to one another and thus an image is built up, very much as the scanning spot builds up a picture on a television screen. The dimensions and timing are so chosen such that the strips are just about 1km in width. The radiation emerging from the telescope passes through a series of partially reflecting mirrors which divert part of the beam to different sensors each receptive to a different wavelength. In this way a set of images each formed from radiation at different wavelengths is simultaneously built up. If you refer back to the picture of the AVHRR, the rotating mirror assembly can be seen at the front, the reflecting telescope mirror behind it, and the sensor arrays are contained within the main body.

Earlier NOAA satellites carried five sensors giving five channels of data. These respond respectively to the wavelength ranges:

Channel Wavelength m
10.58 - 0.68
20.725 - 1.1
31.58 - 1.64
410.3 - 11.3
511.5 - 12.5

Comparing these figures with those in the figures above, we can see that Channel 1 corresponds to the visible region, Channel 2 to the near infra-red. These channels produce images by virtue of reflected radiation (from the sun) and consequently there is only anything to see when in daylight. Channels 4 and 5 on the other hand respond to emitted radiation from the Earth so give images at all times. Channel 3 is intermediate between these and may give images either by reflected or emitted radiation, according to the time of day. Later NOAA satellites added an additional sensor so that Channel 3 became 3a and 3b. The former operates in the region 1.58 - 1.64 mM, the latter 3.55 - 3.93mM respectively. This channel was designed for cloud discrimination and the additional sensor was to improve performance in this respect. The output of the sensors, after some on-board processing, is digitised and transmitted down to Earth line by line - the lines of data from each sensor are transmitted consecutively and reassembled into separate images.

Geostationary satellites create similar images in an analogous way, though they may have different numbers and types of sensors from the polar orbiters. The European Meteosat 7 has three sensors for visible, infra-red radiation and an intermediate one for estimating water vapour content. The satellite rotates at about 100 rpm and a moving mirror assembly enables scanning of the entire Earth surface.

Back to Introduction

© Stuart Hill [Updated November 2002]