Geoscience Reference
In-Depth Information
An integral part of automating the generation of alternative land redistribution plans is to
essentially emulate the process of meeting with each landowner so that preferences for parcels
can be specified. The ES incorporates a measure referred to as the parcel priority index (PPI)
whose role is to define two crucial land redistribution issues: the priority of each landowner-par-
cel pair in the whole project in terms of allocating a new parcel to that landowner in a certain
location and the ranking of the location preferences for each landowner's new parcels. The PPI
also contributes to enhancing equity, transparency and standardisation of the process in terms
of the location and the allocation of the new parcels. More details can be found in Demetriou
et al. (2011).
The basic data that are provided to the ES include a cadastral map of the area and the associ-
ated database tables with information about each parcel, landowner and other ownership details.
To generate alternative land redistribution plans, the user (or the land consolidation planner in this
case) determines which factors should be considered and what weights should be applied to these
factors. The system allows the planner to try out different scenarios, for example, changing the
factors and weights as well as undertaking a sensitivity analysis. The results are a set of database
tables and maps that indicate (1) those landowners taking property in the new plan and those that
do not, (2) the total area and land value of the property that each landowner receives in the new
plan, (3) the number of parcels that each landowner receives in the new plan, (4) the area and land
value of each new parcel and (5) the approximate location (i.e. centroid) of the new parcel(s) owned
by each landowner.
The integration of ES and GIS in LandSpaCES is accomplished via the no-inference engine
theory (NIET). The basic feature of NIET, which was proposed by Hicks (2007), is that the
knowledge base and inference engine are combined into a single unit and not kept separate as in
more conventional ES. This effectively transforms the traditional inference engine into a proce-
dural solution involving a sequence of if-then statements. Thus, the rules are ordered in a logical
sequence during the development stage. Where two or more rules have at least the first condition
of their premise in common, the conflict is resolved by firing the rules with the greater number
of conditions, so that these can be tested first. This conflict resolution strategy is commonly
employed and is the default for most ES products. The prototype ES was built using VBA and
ArcObjects in ArcGIS. Although VBA development has now been phased out of the latest version
of ArcGIS, the ES will be further developed in the future using another programming language
such as Python.
The system was validated through application of the system on a real-world test case taken from
an actual land consolidation project in Cyprus. Details of the test case study area can be found in
Demetriou et al. (2010). Nine performance criteria were used to evaluate the system: number of
landowners who received property (C1), number of common landowners who received property
(C2), number of landowners who received a completed parcel (C3), number of common landowners
who received a completed parcel (C4), total number of new parcels created (C5), number of new
parcels created per owners' group (C6), number of new parcels received by each landowner (C7),
number of new parcels received by each landowner in common blocks (C8) and number of new
parcels received by each landowner in a common location (C9). These validation criteria cover the
most important decisions made by land consolidation experts in Cyprus regarding land redistribu-
tion plans and were therefore used to evaluate the overall system performance when compared to
the solution generated by the human experts.
The results of the system (Figure 11.5) show very good performance based on these nine criteria
although there is clearly room for improvement in factors C3, C8 and C9. However, this result is
remarkable given that the system currently lacks significant data such as the landowner prefer-
ences, which were emulated using the PPI referred to earlier, the land use and the personal data
from the landowners, for example, residence, age and occupation. Moreover, there would have
been exceptions to the legislation that were applied in this case study but which could not have
Search WWH ::




Custom Search