Information about the dimensions of urban expansion, and its spatial, temporal and thematic patterns, is crucial for metropolitan governance - a city’s ability to actually manage this process. Earth Observation (EO) services are increasingly used in this context and have become a valuable tool to help achieve the mission of a number of organisations across various sectors, which can obtain significant benefits from the stream of comprehensive geospatial information for more informed decision-making. In particular, long-term trends, up-to-date situation and future forecast of such expansion in regional context (within and outside city jurisdiction) play therefore important role in setting their investment and analytical and operational priorities in support of the urban development.
Urban change detection provides important information for a number of sectors but operational monitoring is missing. At present there is no global source of information to monitor urban expansion and population change. Existing sources of data are not comparable because countries define urban extents differently and because redefinition of areas from rural to urban frequently lags actual change in land use and population density. Administrative unit boundaries are set very differently and fail to capture de facto urban extents.
Value currently available
Currently the operational techniques of urban change detection are limited to semi-automated pixel-based or object-based change estimation and visual editing. These are typically bi-temporal unsupervised techniques coupled with supervised labeling of selected classes. Contextual approach is possibly applied at polygon (object) level counting of e.g. length of common border.
- bi-temporal analysis for two points in time – only few maps (typically two) for long time period,
- land cover / land use change is currently mostly supported by programs with 6-10 year period - sufficient for the back-casting application, but has apparent limitations supporting now-casting or even forecasting ones,
- semi-automated pixel-based or object-based change estimation and visual editing - time-consuming and expensive.
Value provided by UD processor
The proposed approaches for continuous urban change detection provide solutions to some currently existing constraints that are objective reasons for low automation of the change detection procedure, besides temporal limits.
Novelty content and improved technical features:
- continuous monitoring – new information are available annually;
- large volume data processing – processing of more cities, new satellite system as ESA Sentinel-2 or NASA LDCM - will be open and free;
- contextual pixel-based approach, avoiding problems of vector object-based approaches;
- adaptive modeling;
- accounting for uncertainty;
- fully automatic;
- incorporating the current knowledge about the urban changes into the models.