NEW METHOD FOR THE DESIGN OF THE REALLOCATION PLAN IN LAND CONSOLIDATION PROJECTS
Neue Methode für den Zuteilungsentwurf in Bodenordnungsprojekten
Methode nouvelle pour l'élaboration de la remembrement rural
Ir F.B. RosmanFaculty of Civil Engineering and
Geo-sciences, Delft University of Technology
Thijsseweg 11, 2629 JA Delft, The Netherlands
Tel +31 15 2783780, E-mail: f.rosman@geo.tudelft.nl
Ir. J.K.B. Sonnenberg
Netherlands Cadastre
Waltersingel 1, 7314 NK Apeldoorn, The Netherlands
Tel +31 55 5285245, Fax +31-55-3557931, E-mail: sonnenbe@pi.net
ABSTRACT
The paper reports on research at the Delft University of Technology on a new framework for the support of the design process in land consolidation projects.Changing viewpoints on the sustainability of ecological, agricultural and spatial developments in the Netherlands have led to the introduction of explicit demands in these fields. During project implementation, these demands require preservation and improvement of existing natural elements as well as creation of new natural elements. But also a suitable layout of the farms is required in order to facilitate less intensive farming methods. Conflicting interests between land claims of different farmers and conflicts between ecology and farming are unavoidable and need to be resolved in an accountable way.
Ongoing research into a set of models and implementation of these in information systems is proving to be successful. The new method generates more alternative solutions, provides a setting for better informed design decisions and enables better information and communication.
Es wird berichtet über die Entwicklung einer neuen Rahmwerk für den Entwurf
unter den erweiterten Fragen und Bedingungen die den Bodenordnung heute gestellt sind. Untersuchungen nach Modellen und Umsetzung der Modellen in Software zeigen sich erfolgreich. Es können mehr Alternativen generiert werden und dass ergibt mehr Möglichkeiten bessere Entscheidungen zu treffen und diese zu kommunicieren.
L'objet de ce rapport est de décrire un modčle et des programmes pour l'élaboration d'un projet de remembrement rural sous les contraintes la societé demande aujourd'hui. Le projeteur peut choisir de plusieurs des alternatives et obtenir plus des informations pour communicer les résultats.
INTRODUCTION
Land development in the Netherlands has evolved from an optimisation of agricultural land use, to the improvement of several functions of a rural area. As a result, land development projects now also include measures for the improvement of the environment, ecological conditions, landscape and recreation. These measures are taken in addition to reallocation of farmland and improvement of water supply and control.
Unavoidably, where more functions need to be placed in the same area, the competing demands lead to a more difficult design problem.
Land development projects are divided in two phases. In the preparation phase, alternative development strategies for the project area are explored. This is done in relation to the national and regional policies on physical planning. Building public support for the measures with both landowners and users and with, for instance, nature preservation societies, is a crucial element of the preparation phase.
The decision to proceed with the implementation phase is taken either by ballot of the landowners or users or by decision of the provincial authorities, depending on the nature of the project. In the implementation phase works are carried out (improvement of waterways, natural elements, roads, etc.) and a land consolidation is executed.
Agricultural benefits of consolidation remain unchanged: for a farm there is considerable economic advantage in being allocated in fewer, larger parcels of better shape, closer to the farm-buildings. Larger parcels in general make it feasible for the farmer to use less intensive methods, which decreases environmental impact in the form if nitrate and phosphate production.
Because of the consolidation public bodies and nature preservation societies are able to buy available land in the project area against market prices. This avoids having to pay inflated prices for land in situ or having to resort to the unpopular instrument of expropriation. In the land consolidation this land can be allocated at the planned location.
The project is carried out under the direction of the local Land Development Committee, supported by government agencies.
The way the process of reallocation is structured and the way it can be supported by computer software are the elements we will focus on in this paper.
THE REALLOCATION PROCESS
The size of a land development project is typically between 3000 and 5000 hectares, involving several hundreds to a few thousands of landowners and users, here in general referred to as claimants. The numbers vary regionally because of different customs regarding tenancy and co-ownership. The number of actual farms and larger owners (including public bodies etc.) are in the 100-200 range.
To be able to carry out the reallocation efficiently and accurately, several automation projects have been carried out since the end of the sixties. The need to avoid errors is obvious from the high financial (and emotional) value of land. Not only is there a need to allocate land to the correct owner with the correct value, also tenancy rights and mortgages need to be transposed to the new situation.
Computer support for this administrative problem was introduced at an early stage in the seventies. In the currently available LIN system there is comprehensive support for all aspects of the administrative problem.
However, LIN can be seen as a registration of the before situation, of intermediate designs steps and of the final design. It does not actually support the designer in the puzzle he (or she) has to solve.
Design support
Although the approach to the design problem was under development even before the introduction of programmable calculators, the use of the computer instigated important research in this field in co-operation between the Delft University of Technology (DUT) and the Netherlands Cadastre (Van der Schans 1972).
This has led to the formulation of a model for the process of reallocation, consisting of three consecutive steps:
The allocation is primarily based on land value, rather than area. Because of variations in soil quality and other aspects, this means that a claimant's allocation can be larger or smaller than in the situation before the project. However the difference must be smaller than 5%.
The design of the value allocation plan is the redistribution of a claim-value over the project area under certain agricultural, ecological and other criteria. For this step alphanumerical software, named INOK, is available at the Cadastre. INOK has two allocation-algorithms (one based on a heuristic, the other on a linear programming model) each with appropriate data structures (De Vos 1991)
Once a balance is achieved between supply and demand of land (expressed in value with a general location), a layout of the new parcels is made (sketch). Because of economic as well as ecological concerns, increasingly the existing topographic boundaries are used as new parcel boundaries. Conflicts that arise out of the difference in value to be allocated according to the value allocation plan and the available area according to the topography need to be resolved locally. This can involve moving some part of the claimant's allocated value from one location to the next or moving complete parcels. This is generally not regarded as a modification of the value allocation plan. Although the LIN system provides every possibility to register these intermediate design steps and produce plots on screen or paper, there are no specialised functions to support the user's search process to solve the details of the modification.
In the third step, the LIN system is used to finalise the design by accurately placing the parcel boundaries, calculating the realised value and area and checking the claimant's allocation against his claim value and area.
Survey of system use
In 1994 a survey, commissioned by the Netherlands Cadastre, confirmed that the usability of the existing design support software INOK was insufficient. Some employees used only part of the system, others had found other ways of storing, retrieving and analysing the required data. The existing software was under-used, because benefits from the cumbersome data-entry work were no longer offset by useful design-support by the software. Even in the existing framework, under-use of the software is undesirable, because manual elaboration of value allocation plan makes it more difficult to explore allocation alternatives that result in an over-all better reallocation.
Similarly, the lack of design support in the layout design phase makes it more difficult to preserve ecological feature of the landscape. Here manual elaboration makes backtracking on design decisions into a practical impossibility, because of the amount of work involved in creating a consistent sketch of a certain area of the project area.
Here there are possibilities to integrate ideas developed at the DUT to support the sketch design. In this research by Annette Buis, the geometric design information is manipulated by a knowledge-based system to generate alternative parcel boundaries on the basis of a value allocation plan and on basic rules regarding topology, geometry, agriculture and ecology (Buis 1998).
What is required is a system that combines ease of use with appropriate feedback and which is based on a suitable design strategy.
FUNCTIONAL DESIGN CONCERNS
In the multi-functional land development projects, some aspects are addressed primarily in the preparation phase. Water management and capacity and quality of the road network are examples. If necessary, land for these functions is reserved and this serves as a framework in which the allocation of the remaining functions must take place.
New nature reserves (for which farmland is to be used), or areas designated for special ecological management are usually connected to certain soil types or groundwater conditions. These areas are therefore also fixed in location, although the exact delineation may change with the reallocation of agriculture land.
In general, functions that require a certain area at a specific location are fixed in the preparation phase.
Agricultural concerns
As mentioned before, agriculture goals for the reallocation are the allocation of as few as possible parcels for each farm and the reduction of the distance of fields relative to the farm buildings. These are goals for a single farm, but the design needs to optimise these goals for all farms, or at least improve the situation. It is obvious that optimising the allocation for one farm can be negative for the economic benefits of its neighbours. Although this problem was recognised many years ago, no attempt had been made to model the effects for the implementation phase.
However, in the Netherlands there are models to calculate the economic conditions of the farms in the project area as part of the preparation phase. This model is used to evaluate the necessity of the project. The same model can be used afterwards to measure the net-benefits of the project. During the allocation design it is difficult to use the model for the following reasons:
It must be noted that economic concerns at present are not ignored, but the designers use them in the form of guidelines, with generally satisfactory results. The reason to stress this point is the introduction of competing goals e.g. in terms of ecology. Also there is the desire to have an accountable design, for which it is necessary to argue more explicitly for or against certain allocation alternatives.
The importance of the degree of consolidation and of reduction of distance varies with the type of farm the farmer operates. For instance, a dairy farm requires more than 60% of the land to be located adjacent to the farm building with the milking machine, in order to operate efficiently. On the remaining 40% the farmer can grow feed or use the land in other ways that contribute to his income.
Both approaches have in common that the way the individual farmer uses the land is not important, since individual conditions, agricultural knowledge etc. may vary and therefore it would be both impossible and undesirable to evaluate on a level of detail lower than farm-type.
Work is continuing at the DUT to modify the existing models for use in the allocation design.
Apart from consolidation and distance, it is required that each allocated parcel is adjacent to the road (this is to minimise legal relationships between different claimants, since otherwise access is only possible across parcels of a neighbour). This access requirement is also a design criterion that can only be checked marginally in the value allocation plan and that needs to be addressed in the layout design.
Ecological concerns
In the Netherlands over 90% of the land is either in agricultural, urban or other intensive use. The remaining land is sometimes referred to as 'nature' but this is a relative term, given the (negative) influence that is present from the surrounding land usage. There is a growing awareness that the reduction of natural areas in the Netherlands has reached a critical level. The introduction of the national policy document on nature protection and development in 1989 has been a start to improve the situation. Many land development projects have been affected by the new policies.
The policy document is based on the desire to preserve the bio-diversity of Dutch flora and fauna. A main ecological network through the Netherlands is being established, linking up with similar networks in neighbouring countries. To this end measures are focused on restoring habitats for certain animal species, usually at the top of their food chain, e.g. the badger. The measures also benefit other species in the same food chain.
The ecological networks are based on the concept of the metapopulation. Groups of individuals belonging to a certain species, living on spatially separated landscape elements, form a metapopulation. Each group has a certain probability of extinction, depending on the characteristics of the species and of the size and quality of the living area. For normal migration, movement between the habitats must be possible. Also, in case of extinction there is a certain probability of resettlement by neighbouring population, depending on the properties of the fields separating the suitable habitats.
It is clear that reclamation, intensified land use, road improvement, widening of canals etc. both have reduced living areas (thus increasing the probability of local extinction) and have reduced chances of resettlement.
It is believed that with carefully planned measures, it is possible to stabilise and even increase populations of many species. It should be possible to combine this with improved agricultural conditions.
In land development projects, land for living areas can be reserved as nature reserves in the preparation phase. However the all-important links between the areas must be safeguarded in the layout design. Especially existing topographic elements are ecologically valuable and cannot easily be replaced by newly planted elements. E.g. the continued existence of a hedgerow can best be served by allocating different land users at each side, if at all possible. This is an example of ecological restrictions on agricultural goals.
In general conflicting interests between land claims of different farmers and conflicts between ecology and farming are unavoidable and need to be resolved in an accountable way.
MODELLING THE DESIGN PROCESS
A design process such as the reallocation of land consists of solving a design problem. Problem solving in general starts with choosing a strategy that is appropriate to the structure of the problem. Simple, repetitive problems with a well-defined goal can be solved using specific recipes that dictate both method and required resources. Cooking recipes are a simple example. A well-defined goal means that the question whether an attempt to solve the problem was successful or not can be answered to a satisfying degree. Often it is not enough that all restrictions on each step as well as all possible end-conditions are well defined. It can still be impractical to optimise the result due to the enormous number of possible outcomes (e.g. playing chess).
Design problems are generally considered to be ill-defined (Zandi-Nia 1992). This holds for both goals and restrictions. This means that it is difficult to conclude that a given design is optimal or even satisfactory.
Solving a real-world design problem will always consist of making goals and restrictions more explicit during the process. This often results in a layering of the process, where decisions of a more global nature are taken on one level and then serve as restrictions on the next. The division of the land development project in a preparation and an implementation phase is one example, and this division is an important one, since is separated by a go/no-go decision. The three-step model mentioned before is another with separation that are easily crossed in both direction.
Within one level, design is a step-by-step process with local decisions that may result in conflicting restrictions after a certain number of steps. In that case backtracking on the partial decisions may become necessary. This also means that a deadlock on a certain level will have to result in a return to the previous level to reconsider some of the restrictions and goals (unless that is not acceptable, in which case there is no feasible solution).
Translation to land reallocation
The brief exploration of some of the agricultural and ecological demands in the previous paragraph has shown that not all goals and restrictions are clear or that there will not always be a clear way to resolve competing claims.
In the three-step model, it is necessary to first complete the value allocation plan before proceeding to the next level, the layout design. Indeed it is often observed that conflicting restrictions on the layout necessitates the transfer of some of the claims to other parts of the project. This in fact constitutes backtracking to a previous level.
Simply because the software makes this impractical, until now this has not been seen as an important event. The three-step model over-emphasises the global nature of each step, since design is a series of local decisions that can be backtracked upon.
DEVELOPMENT OF A DESIGN SUPPORT SYSTEM
The goal is to develop a design support system that is based on the way a designer works (or will be working). In this system both user and computer will perform those actions in which each is best.
A design support system is very well suited to track degree of completion, check adherence to restrictions and is able to perform global optimisations on a the basis of computational models (which function better of a limited number of element-types).
A designer is better able to oversee the spatial relationships of elements of the design and is better in optimising a detailed allocation situation of design elements of different types in a limited area. It must always be recognised that not all aspects of a design can be quantified, so that the designer must be able to take the final decisions on the basis of information available to him.
Transfer
Although an implementation of an integrated design support system that incorporates the three-step model is uncertain, an important step has been taken by the development of the system called 'Transfer'. This implements an improved data model that is based on the way the claimants express their allocation preferences at the preference hearing and on the way the design decision should be made.
At the start, the project area is divided into areas called blocks. These areas are homogeneous with regard to access from the roads, soil type and condition and groundwater level. Because of the relatively dense road network in the Netherlands, the majority of the block boundaries usually coincide with roads or waterways wider than 2 meter.
A block has a certain value and area which can be allocated.
A Farm or claimant is a combination of rights of ownership and use of land, which requires allocation in the project area. Nature preservation organisations and municipalities also own and lease land and such organisations also come under this header, but can distinguished by the attribute type. Their allocation rights are structurally (legally, technically) similar to real farms but they require little design during the phase of the value allocation plan.
A claimant has several attributes, of which value and area to be allocated are the most important for checking the completeness of the allocation. The attributes village and type are important for quality control with aggregated overviews and for the algorithmic commands automated structuring and the transfer algorithm (discussed below).
Preferences and alternatives
Usually the claimant wants 100% of his (or her) land allocated adjacent to the farm building as a first preference. A second preference might be to have 60% adjacent and 40% in a field at some distance of the buildings. Preferences can be mapped to allocation alternatives, consisting of claimparts. Each claimpart is connected to one or more blocks and this connection is called a placement. One of the alternatives is always the current alternative and for each of the claimparts, one of the placements is always the current placement. The system checks the completeness of the input. In this way the allocation is always complete, but because of possibly overlapping or competing placements the allocation may not be correct. The summation of currently placed claimparts in a block may show a positive or negative residual. Especially in blocks with many farm buildings, large positive residuals are difficult to resolve.
Depending one one's point of view, the set of alternatives constitutes a solution space or problem space in which we must try to balance the supply of land (the blocks) with the demand for land (the alternatives). The way we move through this space can be determined by the residuals and by weights that can be attached to the placements - and through aggregation - to the alternatives.
Creating an alternative means specifying a structure of the farm. A certain structure will have consequence for the economic benefits for the farm. This is where an economic model will be able to provide a steering-mechanism for choosing between competing alternatives.
The new data model provides a means of hierarchical choice. First we have to choose from the available structures. Within the current structure we may choose from available placements. From the economic models mentioned before it is clear the choice of placements will have a lesser impact on the benefits than choosing another structure.
Editing
All editing of the database can be done from a list of claimants or by accessing the data from a map. In this map the claimparts are shown as circles as indication of the value/area. The blocks are coloured according to their residual. The user can pick up the claimpart with the mouse and move them about within the block, or transfer the claimpart to another block if allowed by the available placements.
The location within in the block is at first irrelevant, but as the work progresses this can change. The user can start to use the location of the circles to assess layout possibilities, especially when a cadastral or topographic map is available as reference.
Balancing supply and demand can be tried manually or more easily by the use of the transfer-algorithm that available. This algorithm is able to search for alternatives and placements that reduce the block residuals starting with any given state of the problem space.
The system provides feedback on the state of the design as an aid to the designer. E.g. a block is able to report on the amount of free space, which is the difference between the block value and the sum of claimpart values for which there are no alternative placements available in the database. A low percentage free space in a block can be an important starting point for the designer when he is attempting to reduce extreme residuals.
Functionality
An approach unknown to previous software is allowing the user to fix any part of the allocation. A block can set to fixed, indicating that the allocation algorithm should not change any of the allocation in the block. A user, attempting to transfer any claim part from a fixed block (or change the VAP's value), is warned but may proceed if he wants.
The IsFixed attribute of a claimpart or farm can also be set with similar effects. This enables the user to work on the design, solve local allocation problems and finalise the decisions.
This approach is an attempt to solve a drawback of the previous systems. After the (integral) recalculation of the value allocation plan, all allocations could change. The only way to prevent this was removing the information concerning the alternative. Fixing the allocation of parts of the project was therefore both laborious and 'destructive'. With this approach it is possible to preserve the information that was gathered during the process and enable the user to backtrack if unexpected consequences become apparent.
Automated structuring
To gain some insight in the problems that may be expected in a project, as far as competing claims is concerned, it is possible to let the system generate alternatives. This process is called automated structuring. The user can provide templates for the alternatives for each farmtype. The system will match templates to claimants and apply them. The structuring process provides a complete solution space in which initially all alternatives with the highest individual benefit is chosen as the current alternative.
Layout generation
In a PhD research project Annette Buis proved it possible to generate layout designs on the basis of basic rules, such as shown in the example below.
The prototype she developed, finds two possibilities to achieve this: parallel to the existing boundary or perpendicular to it. The system is able to work in two modes: automatic or manual. In the manual mode the user decides which of the layout alternatives is used. In automatic mode the system uses another set of rules to choose from the alternatives it creates.
Example: According to the value allocation plan, the parcel of claimant A (containing the farm buildings) has to be expanded. The system would choose alternative 1, since the shape of the resulting parcel is better than the use of the perpendicular intersection (Buis 1998)
If the user enters classification codes on the quality of the existing boundaries, the system is able to take these into account when generating or evaluating alternatives. This means that ecologically valuable boundaries will not be crossed when expanding parcels and that no new boundaries are chosen that are closer than a tolerance to an existing one (which is important from a cost-reduction point of view since moving a fence or ditch can be expensive). The prototype is not yet able to backtrack in case of conflicting restrictions, but with the method operational as it is, this is primarily a data management problem.
As a prototype the system works satisfactorily in most cases. The system has been tested on several real-world situations. Work is continuing on the rule-base and the classification method. A next step would be to integrate the layout design into the Transfer-system.
Within 'Transfer', the method can be used in several ways:
CONCLUSIONS
An integrated model for the design of a reallocation plan gives new possibilities to the designer. The integrated also uses the value allocation plan and the layout design, but stresses the concurrency of the design work rather than the consecutive execution of these steps.
The 'Transfer' system enables the user to generate more alternative solutions for local allocation problems although at present the layout design is not possible. But 'Transfer' already enables the user to try out alternatives in larger areas where before this was too time-consuming. The system provides a setting for better informed design decisions and enables better information and communication.
The 'Transfer' system should be expanded with layout design capabilities. When this can be realised, the designer will have an integrated set of tools to design a reallocation plan. It will be up to the user to choose the tool that is appropriate to the local and current situation of the design work. Instead of an artificial separation between the value allocation plan and the layout design, it will be a continuum, where in some parts of the project area some allocations can already be fixed and a layout created, whereas in another parts competing claims still need to be resolved. The user will have the benefit of a system that supports a trial-and-error approach, which is appropriate for the design problem that land reallocation is.
REFERENCES
Buis, A., (to be published) 1998, Support of the layout design in land development projects (in Dutch), Delft University of Technology, Delft.
De Vos W. and F.B. Rosman, 1991, The development of instruments for land consolidation projects, in: Proceedings of the 2nd Dutch-Polish Symposium on Geodesy, Delft.
Van der Schans R., 1972, Waardefuncties voor de toedeling (in Dutch), Ruilverkavelingsbode, 1972-2, Netherlands Cadastre.
Zandi-Nia, 1992, Topgene: an artificial intelligent approach to a design process, Delft University of Technology, Delft.
BIOGRAPHY
Freek B. Rosman
Date of birth: 22 July 1962
Geodetic Engineer (MSc.), Delft University of Technology, 1990
Career: land management (land consolidation)
Present position:
Lecturer, Faculty of Civil Engineering and Geo-sciences, Delft
University of Technology
Office address:
Thijsseweg 11,
2629 JA Delft, The Netherlands
Tel.: +31 15 278 3780
E-mail: f.rosman@geo.tudelft.nl
Jan K.B. Sonnenberg
Date of birth: 27 September 1938
Geodetic Engineer (MSc.), Delft University of Technology, 1966
Career: land management (land consolidation)
Present position:
Director of Land Information and Land Consolidation in the
Netherlands Cadastre
Delegate of the Netherlands Association for Geodesy to FIG
Commission 7.
Office address:
Netherlands Cadastre
PO Box 9046
7300 GH APELDOORN, The Netherlands
Tel.: + 31 55 5285 245
Fax: + 31 55 3559 731
E-mail: sonnenbe@pi.net