SumbandilaSat: turn images into useful evidence

SumbandilaSat launched from Baikonur on 17 September 2009. The mission linked satellite development with South African ground operations and practical Earth observation. SANSA’s retrospective describes an engineering demonstration and a training effort as well as an imaging mission. Building a camera is only part of that chain: the spacecraft must point it, store its measurements, transmit them and support processing on the ground. Its archive card uses the name Sumbandila-OSCAR-67.

SANSA’s data portfolio lists Sumbandila observations from 2009–2011 with 6.25-metre sampling. A sampling interval — open definition describes the spacing of measurements on the ground; it is not a promise that every object that size can be identified. Clouds, contrast, illumination, motion and processing affect what a reader can infer. A useful image product therefore needs its date and measurement context alongside its pixels.

For an idealized exercise, consider a square area 100 metres on each side sampled on a 6.25-metre grid. There are sixteen samples along each side, making 256 square grid cells. The area is 10,000 square metres. Multiplying a cell’s area, 39.0625 square metres, by 256 gives the same total. This example illustrates a regular grid; it is not a reconstruction of a particular Sumbandila scene.

Now suppose clouds hide sixty-four cells. Only 192 cells are available for a clear-sky land comparison. A map that marks twenty-four visible cells as flooded has identified flooding in one eighth of the visible cells, or 12.5 percent. It has not established the condition beneath the clouds. Reporting the denominator prevents an attractive colored map from suggesting more complete knowledge than the observations provide.

Count the grid in two dimensions

SANSA’s 2021 account distinguishes the end of primary-function data in 2011 from later engineering information. Its discussion of an expected atmospheric reentry is a forecast made at that time, not an observation we should silently turn into a confirmed date. Different parts of a mission can stop at different times, so a single undated active label is often unhelpful.

Historical imagery remains useful when its limits are understood. A reader could compare an old coastline or settlement boundary with a later, suitably aligned observation, then check whether seasonal changes or different sensors explain part of the difference. The satellite does not need to be transmitting today for an earlier image to contribute to that comparison. SumbandilaSat’s place in this course is therefore both technical and practical: it connects local engineering experience with a data record that can be studied after flight operations change.