| Popis: |
Densification of urban land use, demand for new residential, commercial and industrial areas, and the corresponding construction of (new) infrastructure results in increased soil sealing. As a consequence, more surface runoff is generated and the natural infiltration of water is reduced, affecting the recharge of the groundwater table. The study area of this research, which extends approximately from the axis Antwerp-Brussels-Charleroi eastwards till Liège, corresponding to the Brulandkrijt groundwater system, is for the larger part located within Flanders, and is characterized by extensive urban sprawl. The high density of sealed surface cover in this area (13.5% within Flanders) is far above the European average (1.8%). As such, assessment of environmental impacts of natural and human-induced changes in the water balance requires detailed knowledge, in terms of resolution and spatial distribution, about the presence of sealed surfaces, both now and in the future. Much work has been done on sealed surface mapping, mostly using remotely sensed data, yet so far limited research has been conducted on the assessment of future sealed surface cover. Traditional forecasting models relate sealed surface cover to population densities, or apply a so called factor model, assigning a fixed fraction of soil sealing to each type of land use. In previous research, a stepwise regression model relating sealed surface proportions to a number of neighborhood and land use descriptors has been proposed, demonstrating a strong correlation with e.g. total employment and building unit density, at the level of municipalities. Within Flanders, an alternative approach has been used, subdividing residential/commercial areas in different classes according to the density of urban land use in their vicinity, and then assigning a corresponding average sealed surface fraction to each class, based on topographical map observations. The use of topographical maps as reference data, however, introduces a systematic underestimation of the actual degree of sealing, due to not taking into account elements smaller than housing units (e.g. driveways, terraces, garden houses, etc.), which are not present on the topographical map. The method also does not account for within-class variations in sealed surface density. Being responsible for three quarters of all land-use transformations by 2030, as predicted by the Flanders land use change model, growing residential/commercial areas (15.3% relative increase) can be considered as the main driver of the future increase of soil sealing. The research presented aims to develop an explanatory model that can be used for accurate estimation of current and future sealed surface cover within residential/commercial areas at high-resolution. Therefore, neighborhood characteristics that are considered to influence sealing degree, e.g. surrounding land-use composition, spatial positioning with respect to the transport network and urban centres, are included in the modelling, together with other explanatory variables thatcan be linked to the pattern of sealing. Traditional factor models are compared to global stepwise multiple regression models and more complex spatial partitioning multiple regression models, which allow improving sealed surface density estimation based on spatial characteristics. Remote sensing imagery is used for detailed mapping of present sealed surface cover, which is used as an input for defining the various spatial explanatory models. To estimate future sealed surface cover the models developed are applied in combination with forecasted land-use produced by a cellular-automata based urban growth model for Flanders . |
