therefore, a geology map has been used in order to estimate K. An experienced researcher-geologist, who has repeatedly worked on this particular island, taking into consideration certain classes of the geological map, and considering the soil existing above, has proposed 5 classes for K factor as displayed in Table 4.

Thus, a raster GIS layer for K factor has been created and entered as input to the algorithm (Figure 2).

The LS factor has been created with SAGA GIS using a method proposed by Moore et al. [44] and was used as another raster GIS layer into the algorithm.

Cultivation terraces, is one of the oldest means to reduce soil loss [45] . This cultivation practice is extremely popular in islands due to their efficacy to stop or reduce soil loss. Taking this into consideration, C and P factors have been defined according to relevant literature [46] [47] (Table 5).

4.4.6. Constraints

In order to obtain a balanced land use allocation pattern, six hard constraints were introduced to the algorithm:

・ The first three constraints were related to allowable slopes. The maximum acceptable slopes for Dry farming land, Irrigated farming land, and Urban land were 15%, 8%, and 25% respectively [5] . For this constraint to be realized, a slope map has been created as raster GIS layer which was used as input to the algorithm (Figure 3). This constraint can be defined with Formula (9):

(9)

where: Slopei is the slope of land-use parcel I;

is the maximum allowable slope for land-use.

Figure 2. The K factor raster GIS layer.

Table 4. Estimation of soil erodibility factor (K) based on geological map classes.

Table 5. The C and P factor values which have been used.

・ A fourth constraint regarding the maximum land that can be transformed into Urban land was implemented. Since the model was applied to a Greek island, the relative results of the Greek Censuses held in the last 40 years were considered. Based on these data an upper limit of 100% increase was set for the Urban land parcels.

(10)

where: Urbi shows the number of land-use parcels which are transformed to urban;

Urbinit is the initial urban land-use parcels;

q is the desired percent for maximum urban increase;

P stands for the total population of the units.

Figure 3. Slope map of Naxos island.

As the model developed concerns a Greek island, the relevant provisions of the Greek legislation were also considered. Urban land was not allowed to be allocated to wildlife sanctuaries, small island wetlands, and Natural 2000 network areas. Furthermore, certain constraints related to Urban land were introduced as shown in Table 6. All these partial constraints have been combined as the fifth constraint by creating a mask GIS raster layer that defined areas where Urban land class category should not be introduced (Figure 4) because of institutional provisions.

The following Formula (11) applies the urban restriction constraint:

(11)

where: Urban_maski is the urban restriction mask value for land-use parcel i. In restricted areas Uran_mask equals 0.

・ Finally, a sixth constraint has been applied for the “Other” category, not allowing this category to be generated in areas where it did not already exist. For this constraint to be realized, the Landsat TM classification of 1987 has also been used as GIS raster layer reference and used in the problem definition of the algorithm. The following Formula (12) defines this restriction:

(12)

where: lui is the class assigned to land-use unit I;

Classi is the class value of the reference raster;

K is an auxiliary variable which should be greater or equal to 0.

5. Results and Discussion

The proposed NSGA-II algorithm was set to run for 15,000 generations in order to handle the described objectives, variables, and constraints. The first feasible solution has been achieved in generation 166 and Figure 5 presents four graphs which show the progress of this iteration process for certain combinations of objective functions. In these graphs blue points represent the 170th generation solutions, cerise points represent the 350th generation solutions, orange points represent the 2500th generation solutions, and grey points represent the final solutions belonging to the last generation of the algorithm. The graphs in Figure 5 reveal that better solutions are progressively achieved with respect to the objectives while the scatter of the solutions among the objectives constantly becomes narrower and ultimately forms a Pareto Front. This stepwise graph clearly demonstrates the

Figure 4. Urban restriction mask.

(a) (b)(c) (d)

Figure 5. NSGA-II algorithm convergence process.

Table 6. Institutional constraints related to urban land.

converging process of the proposed algorithm.

The examination of the objectives in two (Figure 5(a), Figure 5(b)) or three (Figure 5(c), Figure 5(d)) dimensions reveals that the algorithm performs sufficiently across all objectives. This trend is also demonstrated in Figure 6 that relates the per pixel average value of the objective functions to the generations created. Since, the problem has been formulated in such way that all objective functions are being minimized the average value of global solutions should decrease as the generation number increases. It can be noticed that a vast decrease has been achieved by the 4000th generation, and the algorithm could have been terminated in generation 5000 providing good solutions. Nevertheless, somewhat better solutions seem to have been achieved around the 10,000th generation while no further improvement was achieved after that point.

A large number of solutions (1279 solutions) lie on the Pareto front and “can be used to derive a suitable solution when considered against the qualitative requirements of different users” [29] . Different scenarios can be drawn as to which solution would be best for a specific area and whether the user should prefer an equally weighted solution or a solution that would favor certain objective function extremes. The “equal weight preferred solution possibly has the most balanced land use distribution” [29] and is the one that we have selected as the best solution in this study.

Table 7 summarizes land-use changes between the initial classification (1987) and the NSGA-II best solution. The modified NSGA-II model proposed to reduce the Irrigated farming land by 16.2% (552 ha decrease) mainly by transforming it into Dry farming land, and to increase Dry farming land by 131.5% (2003 ha increase). The reduction of the irrigated farming land seems to be caused by the slope constraints and the soil erosion objective. With this change, the multiobjective optimization algorithm managed to decrease the soil erosion from 1948 t/y to 1843 t/y. Besides the slope and erosion factors, since water reserves in Cyclades islands are rather low, the reduction of the irrigated farming land could be useful although it would decrease economic return. An impressive increase is proposed by the algorithm regarding Dry farming, and this is mainly caused by the economic return objective. The NSGA-II model also proposed the maximum allowed by the constraints increase for Urban land (100%) mostly on the eastern and central part of the island. Theoretically, the more urban areas are built, the higher economic return is achieved, taking into account an upper limit, above which degradation of the environment will occur. Within this framework, as mentioned in the constraints section, Urban land increase constraint was set to 100%, close to reality (Table 1). It has to be noted that executing the algorithm with different parameters (i.e. 5000 generations) or different number of locked categories (i.e. Urban land, Irrigated farming land and Dry farming land) the proposed new Urban land was more or less located in the same areas. Economic return after optimization increased by 18%.

Concerning agricultural land, the NSGA-II algorithm, given the defined objectives and constraints, proposed (Figure 7) that some parts of Naxos island used for Dry farming in the central part of the island in the classification of 1987 should be turned into Grassland, due to incompatible slopes. According to the NSGA-II best solution the Dry farming practices should be allocated mainly to the central and western part of the island because of its geomorphology.

Moreover, according to the NSGA-II best solution, many land parcels used for Irrigated farming land in the eastern part of Naxos island should be turned into Dry farming land because of the sharp slopes of the area. Farmers though, prefer irrigated farming which is more profitable, despite water deficiency, higher soil erosion and incompatible slopes.

Figure 6. The average per pixel objective function value of each generation.

Figure 7. The initial classification (1987) and the optimum result of the NSGA-II algorithm (after 15,000 generations).

Cyclades islands are touristic, and high pressure for Urban land increase occurs, which was not seriously considered during the NSGA-II model definition. This can be clearly seen in the 2010 classification, as the Urban land has increased by 112%, in comparison to the 1987 classification, and most of it have been constructed at the southwest part close to popular beaches of Naxos island. Thus, in order for the multiobjective optimization algorithm to provide more realistic results, the need for touristic development, which constitutes the main income of Naxos population, should also have been formulated as another objective function.

This analysis reveals that the optimal development of Naxos island should be based on agriculture, especially dry farming, together with balanced and not unilateral urban/touristic development, uniformly over the island

Table 7. Land use/cover changes between the initial classification (1987) and the best NSGA-II result generated within 15,000 generations.

and not just located around the coastline. Focusing on the word “balanced”, different scenarios and weights for every objective function should be evaluated during the best solution selection, while in this paper all objectives were equally weighted. In a later stage, weights could also be imposed to the current objectives in order to improve the proposed model and produce more realistic results.

6. Conclusions

Resource allocation problems necessitate addressing various and usually conflicting objectives. Multi-objective heuristic Pareto-front-based methodologies provide the mechanism to resolve this challenge. In land-use allocation, these methodologies aim to allocate certain land-uses to each land unit.

A heuristic algorithm which aims to support land-use allocation problems on Mediterranean islands has been presented in this paper. The proposed algorithm is a modified version of the NSGA-II, adjusted to allocate land-uses in Mediterranean islands taking into consideration legislation, geological characteristics, economic and environmental parameters.

The effectiveness of the modified NSGA-II algorithm was validated in a land-use allocation problem, which included four objectives, namely economic return growth, transformation suitability, maximum compactness, and least possible soil erosion, as well as six constraints, concerning geomorphology, legislation and urban development limitations. The results showed that the algorithm performed sufficiently, and could be possibly used to address even more objectives and constraints.

Comparing the development proposed by the GA with the actual development of the island, it should be noted that, in order to get more realistic results for land use planning in Mediterranean islands, the multi-objective optimization algorithm should also involve the interaction of touristic development, existing infrastructure, land ownership and environmental resources. Moreover, aiming at a more balanced development, different scenarios and weights for every objective function should be evaluated during the best solution selection, while in this paper all objectives were equally weighted.

NSGA-II is a widely used algorithm and a comparison standard, thus the authors use this algorithm as a starting point for their research on land-use allocation in Mediterranean islands. This research will be extended to other relevant multi-objective optimization algorithms for land-use planning, some of which can better handle larger number of objectives. Weights, interdependencies and uncertainties could also be imposed to the objective functions and constraints in order to improve the multi-objective optimization model and produce more balanced results.

Acknowledgements

This research was generously supported by ΙΚΥ Fellowships of Excellence for Postgraduate Studies in Greece-Siemens Program. The authors would like to acknowledge gratefully Dr Irene Galanos for her support on geological issues, and the National Cadastre and Mapping Agency of Greece for providing the preliminary borderline of the coastline of Naxos island.

NOTES

*Corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1] Cao, K., Huang, B., Wang, S. and Lin, H. (2012) Sustainable Land Use Optimization Using Boundary-Based Fast Genetic Algorithm. Computers, Environment and Urban Systems, 36, 257-269.
http://dx.doi.org/10.1016/j.compenvurbsys.2011.08.001
[2] Datta, D., Fonseca, C.M. and Deb, K. (2008) A Multi-Objective Evolutionary Algorithm to Exploit the Similarities of Resource Allocation Problems. Journal of Scheduling, 11, 405-419.
http://dx.doi.org/10.1007/s10951-008-0073-9
[3] Huang, K., Liu, X., Li, X. and Liang, J. (2012) An Improved Artificial Immune System for Seeking the Pareto Front of Land-Use Allocation Problem in Large Areas. International Journal of Geographical Information Science, 27, 922-946.
http://dx.doi.org/10.1080/13658816.2012.730147
[4] Malczewski, J. (1999) GIS and Multicriteria Decision Analysis. John Wiley & Sons Ltd., Sussex.
[5] Shayga Dn, M., Alimohammadi, A., Mansourian, A., Govara, Z.S. and Kalami, S.M. (2013) Spatial Multi-Objective Optimization Approach for Land Use Allocation Using NSGA-II. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7, 906-916.
http://dx.doi.org/10.1109/JSTARS.2013.2280697
[6] Pareto, V. (1971) Manual of Political Economy. Macmillan Publishers, New York.
[7] Aerts, J.C., Eisinger, E., Heuvelink, G. and Stewart, T.J. (2003) Using Linear Integer Programming for Multi-Site Land-Use Allocation. Geographical Analysis, 35, 148-169.
http://dx.doi.org/10.1111/j.1538-4632.2003.tb01106.x
[8] Balling, R.J., Taber, J.T., Brown, M.R. and Day, K. (1999) Multiobjective Urban Planning Using Genetic Algorithm. Journal of Urban Planning and Development, 125, 86-99.
http://dx.doi.org/10.1061/(ASCE)0733-9488(1999)125:2(86)
[9] Goldberg, D.E. (1989) Genetic Algorithms in Search Optimization and Machine Learning. Addison-Wesley, Reading.
[10] Holland, J.H. (1975) Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence. University of Michigan Press, Michigan.
[11] Duh, J.D. and Brown, D.G. (2007) Knowledge-Informed Pareto Simulated Annealing for Multi-Objective Spatial Allocation. Computers, Environment and Urban Systems, 31, 253-281.
http://dx.doi.org/10.1016/j.compenvurbsys.2006.08.002
[12] Eastman, J.R., Jiang, H. and Toledano, J. (1998) Multi-Criteria and Multi-Objective Decision Making for Land Allocation Using GIS. In: Beinat, E. and Nijkamp, P., Eds., Multicriteria Analysis for Land-Use Management, Springer, Dordrecht, 227-251.
http://dx.doi.org/10.1007/978-94-015-9058-7_13
[13] Srinivas, N. and Deb, K. (1994) Multiobjective Optimization Using Nondominated Sorting in Genetic Algorithms. Evolutionary Computation, 2, 221-248.
http://dx.doi.org/10.1162/evco.1994.2.3.221
[14] Deb, K., Agrawal, S., Pratap, A. and Meyarivan, T. (2000) A Fast Elitist Non-Dominated Sorting Genetic Algorithm for Multi-Objective Optimization: NSGA-II. In: Schoenauer, M., et al., Eds., Parallel Problem Solving from Nature VI, Springer, Berlin, 849-858.
http://dx.doi.org/10.1007/3-540-45356-3_83
[15] Knowles, J.D. and Corne, D.W. (2000) Approximating the Non-Dominated Front Using the Pareto Archived Evolution Strategy. Evolutionary Computation, 8, 149-172.
http://dx.doi.org/10.1162/106365600568167
[16] Coello, C.A.C., Pulido, G.T. and Lechuga, M.S. (2004) Handling Multiple Objectives with Particle Swarm Optimization. IEEE Transactions on Evolutionary Computation, 8, 256-279.
http://dx.doi.org/10.1109/TEVC.2004.826067
[17] Coello, C.A.C. and Cortés, N.C. (2005) Solving Multiobjective Optimization Problems Using an Artificial Immune System. Genetic Programming and Evolvable Machines, 6, 163-190.
http://dx.doi.org/10.1007/s10710-005-6164-x
[18] Jones, D.F., Mirrazavi, S.K. and Tamiz, M. (2002) Multi-Objective Meta-Heuristics: An Overview of the Current State-of-the-Art. European Journal of Operational Research, 137, 1-9.
http://dx.doi.org/10.1016/S0377-2217(01)00123-0
[19] Matthews, K.B. (2001) Applying Genetic Algorithms to Multi-Objective Land-Use Planning. Doctoral Dissertation, the Robert Gordon University, Aberdeen.
[20] Niaraki, A.S. and Kim, K. (2009) Ontology Based Personalized Route Planning System Using a Multi-Criteria Decision Making Approach. Expert Systems with Applications, 36, 2250-2259.
http://dx.doi.org/10.1016/j.eswa.2007.12.053
[21] Eldrandaly, K. (2010) A GEP-Based Spatial Decision Support System for Multisite Land Use Allocation. Applied Soft Computing, 10, 694-702.
http://dx.doi.org/10.1016/j.asoc.2009.07.014
[22] Hajehforooshnia, S., Soffianian, A., Mahiny, A.S. and Fakheran, S. (2011) Multi Objective Land Allocation (MOLA) for Zoning Ghamishloo Wildlife Sanctuary in Iran. Journal for Nature Conservation, 19, 254-262.
http://dx.doi.org/10.1016/j.jnc.2011.03.001
[23] Cao, K. and Ye, X. (2013) Coarse-Grained Parallel Genetic Algorithm Applied to a Vector Based Land Use Allocation Optimization Problem: The Case Study of Tongzhou Newtown, Beijing, China. Stochastic Environmental Research and Risk Assessment, 27, 1133-1142.
http://dx.doi.org/10.1007/s00477-012-0649-y
[24] Porta, J., Parapar, J., Doallo, R., Rivera, F.F., Santé, I. and Crecente, R. (2013) High Performance Genetic Algorithm for Land Use Planning. Computers, Environment and Urban Systems, 37, 45-58.
http://dx.doi.org/10.1016/j.compenvurbsys.2012.05.003
[25] Siegmund, F., Bernedixen, J., Pehrsson, L., Amos, H.C.N. and Deb, K. (2012) Reference Point-Based Evolutionary Multi-Objective Optimization for Industrial Systems Simulation. Proceedings of the 2012 Winter Simulation Conference, Berlin, 9-12 December 2012, 1-11.
http://dx.doi.org/10.1109/wsc.2012.6465130
[26] Yuan, Y., Xu, H. and Wang, B. (2014) An Improved NSGA-III Procedure for Evolutionary Many-Objective Optimization. Proceedings of the 2014 Conference on Genetic and Evolutionary Computation, Vancouver, 12-16 July 2014, 661-668.
[27] Matthews, K.B., Buchan, K., Sibbald, A.R. and Craw, S. (2006) Combining Deliberative and Computer-Based Methods for Multi-Objective Land-Use Planning. Agricultural Systems, 87, 18-37.
http://dx.doi.org/10.1016/j.agsy.2004.11.002
[28] Deb, K. (2001) Multi-Objective Optimization Using Evolutionary Algorithms. John Wiley & Sons Ltd., New York.
[29] Cao, K., Batty, M., Huang, B., Liu, Y., Yu, L. and Chen, J. (2011) Spatial Multi-Objective Land Use Optimization: Extensions to the Non-Dominated Sorting Genetic Algorithm-II. International Journal of Geographical Information Science, 25, 1949-1969.
http://dx.doi.org/10.1080/13658816.2011.570269
[30] Liu, Y., Liu, D., Liu, Y., He, J., Jiao, L., Chen, Y. and Hong, X. (2012) Rural Land Use Spatial Allocation in the Semiarid Loess Hilly Area in China: Using a Particle Swarm Optimization Model Equipped with Multi-Objective Optimization Techniques. Science China Earth Sciences, 55, 1166-1177.
http://dx.doi.org/10.1007/s11430-011-4347-2
[31] Tapia, M.G.C. and Coello, C.A.C. (2007) Applications of Multi-Objective Evolutionary Algorithms in Economics and Finance: A Survey. Proceedings of the IEEE Congress on Evolutionary Computation, Singapore, 25-28 September 2007, 532-539.
[32] Deb, K. and Agrawal, R.B. (1994) Simulated Binary Crossover for Continuous Search Space. Complex Systems, 9, 115-148.
[33] Institute for the Management of Information Systems of the “Athena” Research and Innovation Center in Information (2016).
http://geodata.gov.gr/geodata/
[34] Baatz, M. and Schape, A. (2000) Multiresolution Segmentation—An Optimization Approach for High Quality Multi-Scale Image Segmentation. Proceedings of the Angewandte Geographische Informations Verarbeitung XII, Karlsruhe, 12-23.
[35] McFeeters, S.K. (1996) The Use of the Normalized Difference Water Index (NDWI) in the Delineation of Open Water Features. International Journal of Remote Sensing, 17, 1425-1432.
http://dx.doi.org/10.1080/01431169608948714
[36] Hudson, W.D. and Ramm, C.W. (1987) Correct Formulation of the Kappa Coefficient of Agreement, Photogrammetric Engineering and Remote Sensing, 53, 421-422.
[37] Deb, K., Pratap, A., Agarwal, S. and Meyarivan, T.A.M.T. (2002) A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6, 182-197.
http://dx.doi.org/10.1109/4235.996017
[38] Kanpur Genetic Algorithms Laboratory (2016).
http://www.iitk.ac.in/kangal/index.shtml/
[39] Williams, K. (1999) Urban Intensification Policies in England: Problems and Contradictions. Land Use Policy, 16, 167-178.
http://dx.doi.org/10.1016/S0264-8377(99)00010-1
[40] Bramley, G. and Power, S. (2009) Urban Form and Social Sustainability: The Role of Density and Housing Type. Environment and Planning B: Planning and Design, 36, 30-48.
http://dx.doi.org/10.1068/b33129
[41] Yuan, M., Liu, Y., He, J. and Liu, D. (2014) Regional Land-Use Allocation Using a Coupled MAS and GA Model: From Local Simulation to Global Optimization, a Case Study in Caidian District, Wuhan, China. Cartography and Geographic Information Science, 41, 363-378.
http://dx.doi.org/10.1080/15230406.2014.931251
[42] Verones, F., Huijbregts, M.A., Chaudhary, A., De Baan, L., Koellner, T. and Hellweg, S. (2015) Harmonizing the Assessment of Biodiversity Effects from Land and Water Use within LCA. Environmental Science & Technology, 49, 3584-3592.
http://dx.doi.org/10.1021/es504995r
[43] Yang, Q., Zhao, Z., Chow, T.L., Rees, H.W., Bourque, C.P.A. and Meng, F.R. (2009) Using GIS and a Digital Elevation Model to Assess the Effectiveness of Variable Grade Flow Diversion Terraces in Reducing Soil Erosion in Northwestern New Brunswick, Canada. Hydrological Processes, 23, 3271-3280.
http://dx.doi.org/10.1002/hyp.7436
[44] Moore, I.D., Grayson, R.B. and Ladson, A.R. (1991) Digital Terrain Modeling: A Review of Hydrological, Geomorphological, and Biological Applications. Hydrological Processes, 5, 3-30.
http://dx.doi.org/10.1002/hyp.3360050103
[45] Dorren, L. and Rey, F.A. (2016) A Review of the Effect of Terracing on Erosion.
http://www.ecorisq.org/docs/Dorren_Rey.pdf/
[46] Vahaviolos, T. (2014) Estimation of Soil Erosion, Sediment Discharge and Deposition in Reservoir Using Empirical Methods with Emphasis to the Influence of Rainfall, MSc Thesis, National Technical University of Athens, Athens.
[47] Famellos, S. (2016)
http://www.lifesos.eu/index.php/en/progress/
[48] Greek Presidential Decree 236 (1984) About the Erection of Fences in Areas within a Zone of 500 m from the Shore or Bank of Public Lakes in Accordance with Article 23 Paragraph 1 of the Law 1337 (1983) Official Government Gazette 95/A’/22-6-84.
[49] Ministerial Decision 3046/304 (1989) Building Structure Code. Official Government Gazette 59/D’/3-2-89.
[50] Law 2971 (2001) Sea Shore Legislation and Other Regulations. Official Government Gazette, 285/Α’/19.12.2001.
[51] Greek Presidential Decree 209 (1998) Measures about the Safety of Long-Distanced Transport. Official Government Gazette 169/Α’/15-7-98.

  
comments powered by Disqus
JGIS Subscription
E-Mail Alert
JGIS Most popular papers
Publication Ethics & OA Statement
JGIS News
Frequently Asked Questions
Recommend to Peers
Recommend to Library
Contact Us

Copyright © 2020 by authors and Scientific Research Publishing Inc.

Creative Commons License

This work and the related PDF file are licensed under a Creative Commons Attribution 4.0 International License.