Improving Urban Population Distribution Models with Very-High Resolution Satellite Information

摘要

Built-up layers derived from medium resolution (MR) satellite information have proventheir contribution to dasymetric mapping, but suffer from important limitations when working at theintra-urban level, mainly due to their difficulty in capturing the whole range of variation in terms ofbuilt-up densities. In this regard, very-high resolution (VHR) remote sensing is known for its abilityto better capture small variations in built-up densities and to derive detailed urban land use, whichplead in favor of its use when mapping urban populations. In this paper, we compare the addedvalue of various combinations of VHR data sets, compared to a MR one. A top-down dasymetricmapping strategy is applied to reallocate population counts from administrative units into a regular100 × 100 m grid, according to different weighting layers. These weighting layers are created fromMR and/or VHR input data, using simple built-up proportion or reallocation “weights”, obtainedfrom a set of multiple ancillary data used to train a Random Forest regression model. The resultsreveal that (1) a built-up mask derived from VHR can improve the accuracy of the reallocation byroughly 13%, compared to MR; (2) using VHR land-use information alone results in lower accuracythan using a MR built-up mask; and (3) there is a clear complementarity between VHR land coverand land use.