TITLE:
Investigating the Use of a Modified NSGA-II Solution for Land-Use Planning in Mediterranean Islands
AUTHORS:
Miltiades Lazoglou, Polychronis Kolokoussis, Efi Dimopoulou
KEYWORDS:
Land-Use Planning, Multiobjective Optimization, Genetic Algorithm, NSGA-II, Landsat, OBIA
JOURNAL NAME:
Journal of Geographic Information System,
Vol.8 No.3,
June
17,
2016
ABSTRACT: This paper explores the potential application of a modified version of the
Non-dominated Sorting Genetic Algorithm (NSGA)-II for land-use planning in Mediterranean
islands that constitute a geographical entity with similar characteristics. Study
area is the island of Naxos, which is a typical Mediterranean island. In order to
monitor the land-use changes of the island for the period 1987-2010, object-based
classification of three Landsat images has been carried out. The 1987 land-use classification
defined the initial population for the Genetic Algorithm (GA) and the aim was to
provide the optimal development scenario for Naxos island taking into consideration
legislation, geological characteristics and environmental parameters. The GA was
used in order to introduce land use changes while maximizing transformation suitability,
compactness, economic return, and minimizing soil erosion. The output of the GA
was compared to the actual development of the island. The outcomes confirmed the
proposed algorithm’s convergence process, while the GA solutions eventually formed
a Pareto Front and performed adequately across all objectives. The GA algorithm has proposed reduction of
Irrigated farming land by 16%, increase of Dry farming land by 131%, and the
maximum allowed by the defined constraints increase of Urban land (100%),
mostly on the eastern and central part of Naxos. These changes significantly
differ from the actual development of the island. Economic return after optimization
increased by 18%, while soil erosion decreased from 1948 t/y to 1843 t/y.