Experimental statistics Smart business cycle statistics based on satellite data


As at 30 December 2019

Benefit of satellite data for official statistics

Satellites can provide images of large areas of the earth’s surface at short intervals. Technological progress has substantially influenced and improved the availability and analysis of satellite data so that they are coming into play also in official statistics (see also the WISTA article entitled "Neue Wege der Geodatennutzung"). For this reason, the Federal Statistical Office is undertaking feasibility studies to analyse the potential of satellite data for official statistics in co-operation with the German Aerospace Center (DLR) and the Federal Agency for Cartography and Geodesy (BKG).

The suitability of satellite imagery for preparing business cycles was studied in the Eurostat-funded "Smart Business Cycle Statistics" (SBCS) project. The basic idea is that some economic activities leave traces which can be seen on satellite images and thus be quantified. The number of ships and containers in harbours, for example, can be used as indicators of trading activities and production figures. The occupancy of parking spaces adjacent to shops could be indicative of current sales figures. The advantage of satellite images is their rapid availability. In addition, satellite imagery enables analyses across administrative borders, and cross-border economic areas can be covered without methodological breaks. The idea is to prepare a nowcasting indicator with the help of the satellite images which can identify comprehensive changes at an early stage.

Data source

The SBCS project is based on satellite images. The Copernicus Programme managed by the European Union provides free and comprehensive access to satellite data. As high resolution optical systems were developed, a commercial market emerged as well where businesses operate satellites and sell their data. This means that users of satellite images can draw upon data at different resolutions and frequencies (cf. Table 1).

Table 1: Selection of satellite images
SatelliteResolutionPrice per km2 1 Frequency2
Worldview 0.3 m$ 25daily
Pleiades0.7 m$ 25daily
Spot-71.5 m $ 4.75daily
Rapid Eye6.5 m $ 1.28daily
Sentinel-210 m free of chargeevery 5 days
Landsat 830 m free of chargeevery 16 days

1: Prices as benchmarks for Apollo Mapping
2: Theoretical frequency to be achieved under optimal conditions; sources: Apollo Mapping and ESA

A high quality of the satellite images is essential for the calculation of business cycles to be valid. To obtain a data series without gaps, images must be taken at short intervals from the same position. The fact that 55% of the European territory is covered by clouds on an annual average basis presents problems for the use of many satellite images (King et al. 2013). There are other negative consequences of unfavourable shadow casting and diffuse light.

Another decisive factor is the resolution of the image data. The spatial resolution of satellite data is given as metres per pixel. To enable computer-based automated object recognition, the images must be available with sufficient resolution so that the respective objects can be identified correctly. Container ships can easily be identified on account of their size, the large distance between the individual ships and the single-colour background (water), they can be detected also at resolutions of 10 m. However, a resolution of under one metre is required for containers and cars as they are smaller in size and more tightly packed in harbours and on parking sites (cf.Figure 2).

Satellite images of commercial providers with a much higher resolution must be drawn upon for an analysis of these objects. However, high resolution satellite images are not taken continuously but have to be ordered from the operators (tasking). Consequently, costs are higher and it takes longer until the data are provided if there are competing tasks. When satellite images with sufficient resolution are available, processing them requires considerable storage and computing capacity, which go up as the resolution increases.

Project description

At the start of the project, objects had to be selected from which the economic situation of a region can be inferred and which are suited as indicators. In particular when there are significant changes, these indicators are to provide information on the economic development of a region at an early point in time. For example, the number of ships in inland waters can be used as an indicator and shown as an auxiliary variable for trade. The total number of cars parked alongside retail shops is indicative, at an early stage, of the way in which proceeds and purchasing power develop in a region, especially in the case of significant changes.

Then satellite images with an adequately high time and spatial resolution have to be procured for the area studied. The free images from the Sentinel-2 mission operated under the Copernicus programme, which have been tested, are only partly suited for the purpose. Their resolution is not high enough for an algorithm used to automatically detect small objects such as cars and containers.

The SpaceKnow start-up was commissioned to analyse a test area with high-resolution images. In the long run, however, such analyses are to be made in the Federal Statistical Office.

Then the data series obtained have to be weighted and linked to mirror economic activity. This cannot yet be done as data have been purchased for the test area only and intervals between time series data are sometimes very long. This is due to weather conditions and the fact that high resolution images of sufficient quality are only seldom available.


The fact that there often are large intervals between usable images is currently the biggest challenge with the analysis of economic activities on the basis of satellite images. So far these gaps can only be closed to a limited extent by using satellite data of commercial providers. As satellite technology is advancing, considerably lower costs and a higher frequency of images can however be expected in the near term. A key element in this process is the development of small satellites. These comparatively low-cost satellites weighing not more than 30 kg allow for markedly shorter orbital periods and provide high-resolution images thanks to modern camera lenses. It can therefore be expected that it will be easier in the near future to cope with the above challenges and that satellite images will be used for the analysis of economic activities also in official statistics.


Behncke, N. (2019): Der Konjunkturzyklus und seine Wirtschaftsauswirkungen (thinkaboutgeny.com)

Funke, T. et al. (2016): Eigenschaften und Entwicklung von Kleinstsatelliten. Deutscher Luft- und Raumfahrtkongress. Berlin.

King, M. et al. (2013): Spatial and temporal distribution of clouds observed by MODIS onboard the Terra and Aqua satellites. In IEEE Transactions on Geoscience and Remote Sensing 51 (7), S. 3826–3852.


More studies from our EXSTAT series