• Aucun résultat trouvé

Compléments sur les landscape metrics

B. Compléments sur les landscape metrics

162 B. Compléments sur les landscape metrics

Hydrologically-relevant metric of urbanization

In the studies cited herebefore, the term ‘urban’ is used for different things and it appears that some “urban” catchments drawn from the literature review present rather low urban areas in percentage of the whole catchment area. Most of past studies refer to total impervious area or population density to investigate urbanization effect on catchment response (Pomeroy, 2007).

Imperviousness is apparently a metric more relevant for hydrological studies (Schueler, 1994;

Bosch et al., 2003) but it can be computed in many ways. The population density as a proxy of urbanization is also widely used (e.g. DeWalle et al., 2000) since this information is relatively easily available for quite a long-term period compared to precise land use information.

A recurrent hypothesis brought up when evaluating the impact of urbanization on streamflow is the existence of a threshold on the impervious area. Schueler (1994) used a 10% threshold of total impervious cover for each urban area, assuming less than 10% urbanization has a negligible impact on the hydrological response of the catchment. Other thresholds have been proposed: impervious surface of 25% (DNR, 2005), 20% (Brun et Band, 2000) and 3-5%

(Yang et al., 2010).

Many past studies on the issue of the impacts of urbanization on catchment response used quantitative metrics whereas some recent studies tend to show that qualitative metrics might be relevant as well (McMahon et al., 2003; Weng, 2007; Buyantuyev et al., 2010). Roesner and Bledsoe (2003) indicated that gross measures of urbanization such as total imperviousness do not bring enough information and that the hydrologic impacts due to land transformation from rural to urban shall be dealt with precise understanding of landscape metrics process (McMahon et al., 2003; Weng, 2007). These qualitative metrics were used in the past mainly for geographic characterisation (Narumalani et al., 2004; Jarvis et Young, 2005; Weng, 2007; Buyantuyev et al., 2010; Paudel et Yuan, 2012) but also applied in water quality assessment (e.g. Johnston et al., 1990; Rodríguez-Martínez et al., 2004) or for studying erosion processes (Uzun et Gultekin, 2011).

Due to the different hydrological response to several types of urbanization (i.e. habitat fragmentation, type of development, etc), we believe that these qualitative metrics may provide an interesting additional value for studying the impact of urbanization on catchment response, compared to classical quantitative metrics.

Landscape structures 163 Landscape structures

The spatial mosaic of landscape elements determines to a great extent, the visual appearance and the human perception of a landscape (Figure B-1). The spatial configuration and composition of landscape elements are likely to play an important role in the catchment hydrological behavior.

Figure B-1. Example of land cover map (Whippany River, New Jersey) the entire area under investigation is composed of a mosaic of patches of two land cover classes, urban (red) and non-urban area (white).

There are negative effects of pressures on landscape and species living in the area, which are especially vulnerable, more intense to the human effected landscapes. Landscape metrics investigate landscape structure and modification occurring in the landscape over time.

Landscape structure evaluates land mosaic as measure, number, size and shape (Goyker 2010).

In this context numerous indices have been developed that allow the objective description of different aspects of landscapes structures and patterns (McMahon et al., 2003; Eiden et al., 2010):

Patch Density (PD)

164 B. Compléments sur les landscape metrics A patch represents an area, which is covered by one single land cover class . The different patch densities for urban catchment with fragmented and concentrate urban area is

demonstrated in Figure B-2.

Where PD is the patch density (per 100 ha), is the number of patches and is the area (ha).

Figure B-2. The different patch densities and Edge density for urban catchment with concentrate (left) and fragmented (right) urban area is demonstrated

Edge density (ED)

Edge density (in m/ha) is the length (in m) of all borders between different patch types (classes) in a reference area divided by the total area of the reference unit. Edge density standardizes edge to a per unit area basis that facilitates comparisons among landscapes of varying size. It is calculated as:

Landscape structures 165

Where ED is the patch density, E is the total edge (m) of patches and A is the total area (ha).

Like patch density, edge density is a function of the size of the smallest mapping unit defined (grain size): the smaller the mapping unit the better the spatial delineation is measured, resulting in an increase of the edge length (Figure B-2).

Tableau B-1 and Tableau B-2 summarized the numerous mathematical indices which have been developed to describe of different aspects of landscapes structures and patterns.

166 B. Compléments sur les landscape metrics Tableau B-1. Summarized the indices to describe of different aspects of landscapes structures and patterns

Land scape metrics Equation Description

Patch density (PD)

Edge density (ED) The total length of all edge segments per ha for the land-cover class or landscape of consideration (unit: m/ha)

Standardize edge to a per unit area

Percent Class Area (%CA) Relative area of a specific patch type in a Landscape types, pi = proportion of the landscape occupied by patch that number of patch types. It measures the degree of evenness of area distribution among patch types

Largest Patch Index (LPI) Percent of the landscape occupied by the largest patch (unit: %)

Landscape structures 167

Tableau B-2. Summarized the indices to describe of different aspects of landscapes structures and patterns

Land scape metrics Equation Description

area-weighted mean shape index (AWMSI)

Sum of shape value for each patch multiplied by

Mean Patch Size (MPS) The average area of all patches in the landscape

168