EGIS (1994), copyright EGIS Foundation.
The researching of housing markets for strategic planning purposes involves the clear understanding of population behaviour, when an attempt is made to understand the factors that influence a particular pattern. Planners have to rely on information available at real time and try to predict physical and social demands in the future. The availability of housing units and the ever changing lifecycle, are the contributing factors to the redrawing of city boundaries. When needs expand to the point where they can no longer be met by the unit occupied because of changes in either the household or the house, the household attempts to move to another unit. The result is the development of intra-urban migration patterns, directly connected to spatial references and generating a fundamental question: can the migration behaviour of households that move be predicted from information about their origins? Is it the migration process that remodels the urban structure, or is it the structure itself that controls the migration process? (McLennan 1992).
The Committee of Enquiry into the Handling of Geographic Information (1987), defined in very simple terms " ...geographic information is information which can be related to specific locations on the Earth". Many aspects of our every day behaviour and most of the decisions we make, are restricted, guided or controlled by geographic information. We choose a particular route for our travel between home and work based on our knowledge of local traffic conditions. We open centres to care for the elderly at strategic points, based on our knowledge of the elderly distribution in a particular locality. When the family outgrows its current dwelling, another home is seeked possibly within a particular school catchment area, or because it is overlooking a particular beauty spot or for its proximity to a fast road network.
The event of moving a house and the behavioural patterns which dictate that move out of one geographical area and possibly into another one, is something which every planner would love to have the opportunity to study thoroughly. The Registrar General has noted that at the local level of a District and particularly in areas with small populations, ".....the migration assumptions are more critical than the fertility and mortality assumptions" (General Register Office, 1991). Assumptions are normally based on recent migration trends, but they also take into account the economic prospects, policies and plans which may affect the population distribution (Scottish Office Statistics Bulletin, March 1991).
Depending on the size of the geographical area of study, the distribution of the migration within the geographical area has also to be considered. In the absence of detailed data it is normally assumed that migration is equally distributed throughout each District and the net migration, i.e. the balance between those moving in and those moving out, is calculated on that basis. For the purposes of housing demand assessments, however, regional planning authorities are likely to examine more closely the distribution of migration within local housing markets. It is necessary that long distance (inter-regional) migration is distinguished from local (intra-regional) one. While the latter is more likely to be influenced by availability and choice of housing and by life cycle factors (such as increase of family size), the former is normally associated with change of employment. (McLennan, 1992).
[End Page 190]
Population migration in search of suitable housing arises simply from the localisation of housing demands of choosing housing within accessible distance from the workplace, of a given quality and comfort and within the financial reach of the household as a whole. As a result, a self contained housing market area is created in which the majority of those moving house (without changing job) will stay.
Despite the fact that migration may also be the result from a change in the socio-economic profile of the household (move to a more affluent area following a job promotion), it has an undisputed spatial reference to it. No matter the reason, a household will move from point A to point B. If the geographical co-ordinates, in any projection system, of points A and B are known, then it is very easy to calculate the distance travelled and also the direction. The difficulty is, where that information can actually be obtained from.
As it is explained in a later section, this possibility exists within a well organised cadastral system, such as the Scottish Land Registers, provided that adequate spatial references are attached to both the destination and originating addresses, as recorded in the title deeds. The British postcoding system with its own particular geography and the spatial inferences attached to it, serves also very well this purpose. While everybody's understanding of a postcode is that it is simply a means to speeding the sorting and delivery of mail, the postcoding system used in this country has also some other attractive attributes.
Scotland was the first country in modem times to establish a national system of registration giving to the individual the power to have his rights recorded in an official register and thus protected. This register was set up by the Registration Act of 1617 and is called the Register of Sasines, which while it has evolved over the years to a high degree of perfection offering security and public access, it is an expensive system to maintain due to the excessive length of deeds and of the expenditure of time involved in the examination of title. In addition, the registration of the deed is not map or plan based. The Land Registration (Scotland) Act 1979, introduced a new system known as Registration of Title, operating on the same principles as the Register of Sasine but based on large scale 1:1250 Ordnance Survey maps. Both systems offer a fiscal function (which is used for taxation purposes) and a legal one (which is used for establishing ownership).
As each property comes to be registered (usually by transfer), the Keeper makes a title sheet, entering on it a description of the land including a description based on the official map. The title sheet will also state the name and designation of the person entitled to the property and heritable securities effecting the property. Appropriate changes will be made on the title sheet as occasion demands, e.g. when the property is sold to another person or where a heritable security is discharged. When the registration process is complete and all the certificates have been prepared, the deeds are microfilmed for retention in the relevant Department. All the information which appears on the Land and Charge Certificates, including the digital maps, is stored on computer.
After careful examination of each presented writ, a short summary of it is made, which is then printed and bound in yearly volumes, known as the Minute Book. The Minute Book provides, therefore, an abstract of all the writs presented in the county, in their order of presentation. Scotland is divided for the purposes of registration, into 33 counties the geographical boundaries of which do not unfortunately, correspond with any of the 9 Regional and 55 District areas of administrative responsibility. An extract from the 1991 Minute Book for the county of Midlothian, is shown below:
47488 (No.227) 26 Aug. 1991 Disp. by Executors of MARGARET MURIAL CARSON or TULLY, latterly, 61 Craigleith Hill Gardens, Edinburgh - To KEVIN JAMES SMITH and ANDREA LITTLE or SMITH, spouses, 40 Comley Bank Avenue Edinburgh, equally and
[End Page 191]
survivor. - dominium directum and dominium utile of ground with house &c. 61 CRAIGLEITH HILL GARDENS, EDINBURGH thereon, described in Disp to Huddersfield Building Society, recorded 31 Dec. 1931; which subjects were last vestedin said Margret Murial Carson or Tully; from which said Granters acquired rights by Confirmation. Dated 17 and 20 Aug. 1991 (£87795) 47489 (No. 228) 26 Aug. 1991 STANDARD SECURITY for (£54000 and further sums by KEVIN JAMES SMITH and ANDREA LITTLE or spouses, 40 Comely Bank Avenue, Edinburgh - To ABBEY NATIONAL PLC, - over the subjects in the preceding minute. Dated 20 Aug. 1991.
The Royal Mail the national British postal organization has historically been maintaining in a summary form a list of postally correct addresses, so that mail can be delivered accurately and efficiently. The resulting postcode is simply that summary of each address, in a format that can be easily memorized by humans and read by OCR systems. There are about 1.7 million postcodes covering approximately 24 million addresses in the UK.
The largest postal unit within the whole country, is the Postcode Area and it is denoted in the postcode by a one or two letter abbreviation mnemonic of a large conurbation such as G for Glasgow, EH for Edinburgh, AB for Aberdeen etc; There are 120 such postal areas. Each of these postcode areas is divided into Postcode Districts. denoted by a number ranging from 0 - 99; There are 2,679 such postal districts. Both the Postcode Area and District comprise the outward postcode and for Royal mail purposes it identifies the town mail is to be sent to for further sorting or forwarding. Fig.1 shows two adjacent postcode areas and their corresponding postcode districts.
Each of the postcode districts is divided into a Postcode Sector. which is a single digit numerical indicator; there are 8820 such postcode sectors. Finally, the full postcode is produced by adding two alphabetic characters. Both the postcode sector and the additional alphabetic characters, comprise the inward postcode. which is used for the actual delivery of mail.
[End Page 192]
For delivery purposes, it is not necessary that each and every house is identified individually. Instead, the unit postcode is designed in such a way, so that it contains on average 17 delivery points. The definition of a delivery point is really a letterbox, so that a detached villa is a singe delivery point, but a block of 8 flats, each one with its own letterbox is 8 delivery points.
The resulting database of all these 1.7 million postcodes is maintained centrally and is called Postcode Address File (PAF). It is available on a 12cm compact disk (PAF on CD), which provides for each and every one of the postcodes in the UK:
Fig. 2 shows the complete entry for the Gl3 INF postcode as extracted from PAF on CD, 1993/4 edition.
Some of the attributes which are contained in the PAF on CD file, require some further explanation, so that the inter-relationship of these area units can be clearly understood. The number of postcode sectors is more or less the same as that of electoral wards, but the defined areas do not exactly match. While there are 651 parliamentary constituencies (electing the equivalent number of MPs in Parliament), the number of postcode districts is about 5 times as numerous as these constituencies. The finest set of a spatial unit is that of the enumeration district (ED), used for collecting information in the Census of Population. The number of postcode sectors is approximately ten times that of the enumeration districts.
For each unit postcode, Eastings and Northings of the National Grid are also recorded. While the National Grid is explained in a separate section, some points on its usage and usefulness are necessary at this point. While the Ordnance Survey has developed a grid whose co-ordinates are alphanumeric, PAF contains a conversion to numeric representation of the easting and northing Since the origin of the grid is west of the Scilly Isles, the co-ordinates of the Gl3 1NF postcode are 254.67 kilometres east of that point and 668.48 north. These co-ordinates represent the position of some place within the group of delivery points for Gl3 1NF, offering an average resolution of around 10m in Scotland and 100m in the rest of the country. The origins of these co-ordinates, as held in PAF on CD, go back to the early 1970's and their method of collection and digitisation makes them quite inaccurate and out-dated since they are only updated when new postcodes are created or old ones deleted. This problem is however about to be resolved with the Ordnance Survey's "Address Point" product, which is a database of National Grid co-ordination for each and every property in Britain.
[End Page 193]
In their attempt to provide a consistent way of specifying the position of any point in Britain, Ordnance Survey, the British national mapping organisation, have devised the National Grid. For the purposes of this application, the National Grid can be thought of as a complementary element to the Postcode as a way of describing position in space. The following short description is based upon that given in Harley (1975).
The National Grid can be thought of as a sheet of the paper covering the whole of British with its origin just west of the Isles of Scilly. The small longitudinal extent of the UK, prevents any serious divergence from the Transverse Mercator projection graticule (representing lines of latitude and longitude). however, an exact match of the two systems is impossible to achieve. Rectangular co-ordinates are related to a working origin, approximately Latitude 49° 46'; Longitude 7°33'W of Greenwich, slightly south west of the Isles of Scilly. Co-ordinates are expressed in metres and parts of a metre, according to the standard of precision required. The grid references do not give the actual distances from the reference point, but a system of combination of letters and numbers has been employed to denote location, thus avoiding the lengthy numerical representation of a 12 digit (one metre resolution) grid reference. The grid consists of fifty five 100km squares. each referenced by a 2-letter abbreviation, as shown in Fig 3. Each of the digits shown in brackets, represents the distance from the origin in 100km units, of the south west corner of each square. Thus, square NS(26) denotes a distance of 2xl00 = 200km east; 6 x 100 = 600km north from the reference point.
Each of those 100km squares, is further divided into smaller grid squares. of 10km 1km and 100m side length, each denoting the distance from the south-west corner of the 100km square a particular point lies. These squares provide the number with 2, 4 or 6 digits, which identifies each particular grid square. Fig. 4 shows the grid reference of my house to various degrees of precision.
The grid reference, therefore. of my house. with 10m resolution, is NS5467 6848. Remembering that the numerical equivalent of the NS square is 26. it means that my house lies 254670 metres east and 668480 metres north of the Isles of Scilly. These are the co-ordinates recorded on PAF, for postcode Gl3 INF.
[End Page 194]
The following table shows the grid references of this location, at various scales.
Distance from
Map scale Grid interval(m) Grid reference origin(m) Precision(m)
1:625000 10000 NS54 68 254000 668000 1000
1:150000 1000 NS546 684 254600 668400 100
1:1250 100 NS5467 6848 254670 668480 10
- 10 NS54673 68484 254673 668484 1
While the primary purpose of the Scottish Registers, the speedy and unambiguous establishment of title, is served well by the systems currently in place, the extraction of information out of their raw data contents, is practically impossible. None of the schemes adopted allows easy data abstraction and manipulation, such as sampling, data aggregation and sorting, basic operations which are necessary for research purposes.
This research potential was realised as early as 1977 by the Department of Land Economics at the University of Paisley and as a result, the Land Value Information Unit (LVIU) was established which has grown from a departmental resources laboratory to a unique commercial and research centre. Computerised data processing of the records, introduced as early as 1978, has allowed the extensive databanks to be used for undergraduate teaching, research and consultancy.
The Minute Book which is publicly available, for each of the 33 counties is received by the Unit and after extensive processing, is turned into a database, using software developed under Ingres on a Hewlett Packard 9000/800 server computer. While most of the raw data content is recorded, the records are enhanced with in-house added codes and with spatial information, via an interface with the PAF on CD database. While a section of the database is offered as a service to the surveying profession in Scotland, the full contents of it are used for carrying out research into the housing market, in a variety of ways.
For each property transaction recorded in the Registers, the LVIU database shows:
(1) Characteristics
Internal characteristics of the property (i.e. number of apartments, internal area and other features) are not recorded during the registration process. This information is collected from other sources and is independently processed, added automatically to the core database, at a later stage, employing address and property unique number matching processes.
2) Classification
An in-house evaluated code, which determines the usage of the property, such as residential, office, shop etc.
(3) Address
The full postal address is extracted directly from the raw data, thus providing the basic information of where exactly on the earth is the property which has been sold.
(4) Postcode
This is not recorded during the registration process and is obtained by interfacing the core database with PAF on CD, by matching the address, for the extraction of not only the postcode, but some of the spatial information associated with it.
(5) Registration Title Number
This is an administrative number, provided by the Land Register, and serves as a unique property identifier. Every property acquires a unique property identification number, which remains with it and never changes, no matter the number of transactions. Subsequent searches, therefore, are becoming very easy to instigate.
(6) Date and Price
The fiscal purpose of the Registers is served, recording the date of the transaction taking place (month and year) and the price paid.
(7) Type of Sale
This is another of the in-house added codes, describing in a coded form the nature of the transaction, e.g. a purchase under the terms of a particular legislation, purchases of newly built houses as opposed to traditional
[End Page 195]
ones etc. Them codes are the inherent strength of the databank, since they provide the opportunity for data abstraction, aggregation and analysis.
(8) Previous residence of purchaser(s)
While for the purposes of the surveying profession, only the town is released, the research version of the databank contains the full postal address of the purchaser's previous residence. Similar to the postcode field the postcode and the spatial information of that address are also obtained from PAF and recorded.
(9) Name(s) of seller(s) and buyer(s)
The recorded names provide further supportive evidence, when there is a doubt of the recorded address. They can also be used to identify areas of investment activity by corporate purchasers and/or sellers. Fig. 5 shows a complete extract from the research version of the database. The highlighted fields show the data as entered manually by the data processing staff, while the remaining are obtained automatically, via purpose written software. Information relative to the property being purchased has been marked as destination, while information relative to the purchaser's previous residence, has been marked as origin.
While as it has been reported earlier, this database is maintained using Ingres application software, it also exists in a PC format under the dBaseIII+ format, which makes it accessible by both DOS and Windows-based software packages, such as Microsoft Access and Borland's Paradox.
Bartholomew provides a digital map database of the UK, at 1:250000 scale, which is updated annually. The maps are produced in Arc/Info Export and Ungen formats and they occupy around 250 Mbytes of disk space. They are structured in the form of a map library which potentially has 18 layers for each of the fifty five 100 x 100 km squares comprising the National Grid (see Fig. 3). The data are structured in such a way, so that there is a workspace for each square and up to eighteen coverages in each workspace. The coverages provided for each square, may comprise some or all of the following:
IDRISI is a raster-based geographic analysis system developed at Clark University, that is designed to provide inexpensive access to computer assisted geographic analysis technology. It is not a simple computer program,
[End Page 196]
but a collection of almost 100 program modules that may be linked by a unified menu system. These independent modules are written in Pascal or Basic and since they are linked by a simple data structure, the system allows users to develop their own modules with minimal regard for the internal workings of the core set modules.
These modules fall into one of the broad groups:
While IDRISI runs on any Intel based microcomputer, it should be used ideally on a 80486/20MHz+ machine, with 2Mb+ RAM, a VGA display and around 4Mb of free hard disk space. It is my understanding that while the latest available version under DOS is 4.1, a Windows version is scheduled for release towards mid-1994. The DOS version supports a limited number of output peripherals, and any digitiser capable of ASCII output is supported via TOSCA.
While this package is not best suited for the particular application it was chosen for some very good "strategic" reasons
(1) only recently has approval been given for the creation of a small GIS laboratory at Paisley; therefore, costs had to be maintained at minimum levels, by using software on existing hardware (exclusively MSDOS/IBM-compatible PCs).
(2) familiarity with the operating system, made the learning curve of the package quite short, reaching quickly acceptable levels of proficiency. Version 4.0 was used for this particular project.
(3) availability of raster-based digital maps for the whole of the UK, under extremely favourable terms agreed between British academics and the publisher Batholomews, was well suited to the vector-to-raster conversion modules available in IDRISI. However, image limitation in v4.0 used, resulted in some images having to be digitized by hand. It is claimed that v4. 1 would have avoided this problem.
(4) its capability to interface directly with dBase format files, was also a particularly favourable point. The module DBIDRIS provides the interface between IDRISI and dBaseIII+/dBaselV data files. It allows the extraction of any field of data into an IDRISI attribute values file and the import of an attribute values file into a dBase field by means of a relational lookup.
While DBIDRIS and most of the modules are small enough to run from within dBaseIII+. by using the RUN command at the dot prompt, dBaseIV does not leave enough free memory to do this.
[End Page 197]
Fig 6, above provides the lineage of the application, where 2 distinct paths can be clearly identified independent from each other. The left hand side path can be carried out as a stand-alone application, producing an efficient Information System (IS). Only the Interfacing with the processes shown in the right hand side path, permit the IS to be turned into a true GIS. When this project was initiated boundaries in digital format existed only for the 9 Scottish Regions. The output shown in Fig. 7 demonstrates the principle at that level of spatial reference, but it could be readily expanded, should the postcode boundaries were available in digital form, to produce results in postcode sector zonal resolution.
[End Page 198]
Colpaert A, Rusamen J and Naukkarinen A, (1992) The square kilometre grid system: a new database for Regional Research in Finland, Proceedings of EGIS 92, vol II. ppl570-1571, Munich, Germany.
Department of the Environment, (1987) Handling Geographic Information, Report of the Committee of Enquiry chaired by Lord Chorley.
Douglas P, (1992) A quality standard for a postcode address file, Proceedings of Mapping Awareness, London, England
Flowerdew R, Green M and Lucas S, (1991) Analysing local house price variations with GIS, Proceedings of Association for Geographic Information Conference, Birmingham, England
Harley J B, (1975) Ordnance Survey maps - a descriptive manual, Ordnance Survey. McLennan D, (1992) Analysis of needs and demand in the local housing system, Centre for Housing Research University of Glasgow.
McLennan D, (1992) Housing search and choice in a regional housing system: new housing in Strathclyde, Centre for Housing Research, University of Glasgow.
Pitticas N, (1992) Land Value Information Unit, The Scottish National Land Information System?, Proceedings of Association of Geographic Information Conference, Birmingham, England
Pitticas N, (1992) The Scottish Land Registers, Proceedings of URSANET Forum, Patras, Greece.
Raper J, Rhind D and Shepherd J, (1992) Postcodes, the new geography, Longman. Rideout T, (1993) Bartholomew Digital map Data 1993, GB 1:250000, Bartholomew.
Rowlingson B, Pooley C and Gatrell T, (1993) Analysing migration paths in a GIS framework, Proceedings of EGIS 93, vol. I, pp 125-133, Genoa, Italy.
Weber C, Esnault D and Hirsch J, (1992) Remotely sensed indices for Urban Growth using low and high resolution sensors: comparison in an urban growth study with ARC/INFO, Proceedings of EGIS, vol II, pp 940-944, Munich, Germany.
[End Page 199]