Mogens Humlekrog Greve

Implementing GIS regression trees for generating the spatial distribution of copper in Mediterranean environments: the case study of Lebanon

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

  • Rania Bou Kheir, Danmark
  • Mogens Humlekrog Greve
  • Jean-Paul Deroin, Université de Reims Champagne-Ardenne, Frankrig
  • Noamen Rebai, Université El Tunis Manar, Tunesien
Soil contamination by heavy metals has become a widespread dangerous problem in many parts of the world, including the Mediterranean environments. This is closely related to the increase irrigation by waste waters, to the uncontrolled application of sewage sludge, industrial effluents, pesticides and fertilizers, to the rapid urbanization, to the atmospheric deposition of dust and aerosols, to the vehicular emissions and to many other negative human activities. In this context, this paper predicts the spatial distribution and concentration level of copper (Cu) in the 195km2 of Nahr el-Jawz watershed coastal area situated in northern Lebanon using a geographic information system (GIS) and regression-tree analysis. The chosen area represents a typical case study of Mediterranean coastal landscape with deteriorating environment. Fifteen environmental parameters (parent material, soil type, pH, hydraulical conductivity, organic matter, stoniness ratio, soil depth, slope gradient, slope aspect, slope curvature, land cover/use, distance to drainage line, proximity to roads, nearness to cities, and surroundings to waste areas) were generated from satellite imageries, Digital
Elevation Models (DEMs), ancillary data and/or field observations to statistically
explain Cu laboratory measurements. A large number of tree-based regression
models (214) were developed using (1) all parameters, (2) all soil parameters only, and (3) selected pairs of parameters. The best regression tree model (with thelowest number of terminal nodes) combined soil pH and surroundings to waste
areas, and explained 77% of the variability in Cu laboratory measurements.
The overall accuracy of the predictive quantitative copper map produced using
this model (at 1 : 50,000 cartographic scale) was estimated to be ca. 80%.
Applying the proposed tree model is relatively simple, and may be used in other
coastal areas. It is certainly of significant interest to local governments
and municipalities. It will serve several development projects concerned
with improving the environmental conditions and the quality of living in coastal
TidsskriftInternational Journal of Environmental Analytical Chemistry
Sider (fra-til)75-92
StatusUdgivet - 2013

Se relationer på Aarhus Universitet Citationsformater

ID: 40717003