This paper introduces a new theme... the relationship between heritage data
GIS and the O.S. Do the doubts and uncertainties characteristic of our data
pose particular challenges to the emerging O.S. strategy? I'll post this in
two parts as its quite long...

Edmund Lee



From:   Peter McKeague, RCAHMS [mailto:[log in to unmask]]

**Where we are:  Understanding data and positional accuracy issues at the
RCAHMS**

GIS plays a fundamental role in the organisation and management of a wide
range of data within the Royal Commission on the Ancient and Historical
Monuments of Scotland.  Data held by the RCAHMS includes transcriptions of
oblique aerial photographs, digitally gathered survey plans (line data),
Historic Landuse Assessment (area or polygon data) and a geospatial
expression of our core data set representing the National Monuments Record
of Scotland database (point data), familiar to many through the CANMORE
interface on our website (www.rcahms.gov.uk).


*Positional Accuracy
GIS is no more than a tool to display, search and analyse data from a
variety of sources, captured to varying degrees of accuracy and precision.
GIS confers upon data an apparent degree of accuracy often far beyond the
integrity of the source material.  Data is often shared within an office
network or exchanged between organisations, but knowledge about how that
data was accrued is not.  Point data can be particularly deceptive - 'X'
marks the spot.  But does it?

To fully comprehend the potential problems of a particular data set, the
user must understand the data gathering process.

*Transfer of site location from existing databases to GIS
Since its inception, as a card-index maintained by the former OS Archaeology
Division to manage their work in mapping and publishing antiquities, the
NMRS database has grown exponentially to include information reported, to
different levels of precision, from a wide variety of sources and abilities.


Within the NMRS database the locations of most records are expressed to
either the nearest 100m (OS 100km map square letter followed by a six digit
coordinate or to the nearest centimetre) or the nearest 10m (OS 100km map
square letter followed by an eight digit coordinate or nearest millimetre).
Using the OS 1:10,000 chart map as a base, the average human cannot achieve
a higher level of accuracy when plotting or reading a grid reference.
Additionally point data does not convey area extent.  Ideally, site area
extents should be digitised from the chart copy record maps but this is
currently beyond our resources.  Applying a buffer zone around the point to
reflect the predicted extent of a site, on the basis of a pre-determined
value, is one possible solution but sites are rarely symmetrical and the
point data not centrally placed.

At the other end of the scale, there are also records, located to the
nearest 1km (OS 100km map square letter followed by a four digit coordinate)
or even 10km (OS 100km map square letter followed by an two digit
coordinate).  Many of these coordinates are spurious, generated to attach an
antiquarian find spot to a particular generalised location, such as a
parish.

Within a database, variances in the NGR are readily apparent and understood
by the user as they are expressed in a familiar format.  Even if the
accuracy of any given reference is not instantly apparent, the precision of
a coordinate is relatively clear to see, whether it is expressed to the
nearest 10m, 100m or 1km.

Transferring this data into a meaningful geospatial coordinate instantly
muddies the waters.  At the RCAHMS, conversion is achieved through scripting
a numeric substitution of the OS 100km letters to the front of the relevant
parts of the NGR together with padding out the tail with zero values (though
this could be set to any value).  The resultant twelve digit values (or
thirteen for those in OS 100km squares commencing with the letter H) appear
precise.  They are not.  The precision level can, however, be displayed
graphically to distinguish between those coordinates given to the nearest
10m from those precise to 100m or 1km.  Ideally, although part of the same
database, each level of precision should form a separate layer within the
GIS so that the precision level is instantly apparent.  It may even be more
appropriate to attach these records to an appropriate administrative area
map such as a parish map and dispense with displaying them altogether.


*Site creation in a GIS
Traditionally sites were created and plotted against the 1:10,000 chart maps
and, as described above, the centre points could be created to a precision
level of about 1m.  Creating new records on screen against a GIS backdrop
raises concerns about positional accuracy, particularly when working against
the backdrop of a raster map.  A vector map overlying a raster background
demonstrates the different levels of accuracy between the two datasets.
Data created against the background of a vector map, particularly in an
urban environment, is more accurate than that created against a raster map.
Raster maps are imagws, they are not scaleable and their useful viewing
range is not that flexible.  In this regard data created may be of a similar
level of accuracy as that created on the chart maps, but gives the
impression of being much motre reliable than it actually is.

*Survey detail
Modern surveying techniques enable site location to be expressed to the
nearest metre (OS 100km map square letter followed by a ten digit
coordinate) if not several hundredths of a metre.

Within the RCAHMS, field survey mapping techniques have evolved from
detailed plane table surveys of individual monuments to mapping monuments
and landscapes through, firstly, EDM survey and, latterly, differential GPS.
Although the results of such surveys may look very similar, the information
is recorded to two very different standards.  The positional accuracy of EDM
surveys is established through locally surveyed reference points; a
methodology reliant on the accuracy of the mapped detail in relation to the
precision of the surveying equipment, whereas GPS surveys are controlled by
fixing the base station into the active GPS network.  Additional local
control may be mapped to ensure standards within the GPS survey.  Overhead
satellite coverage produces an absolute position expressed to the GPS
standard coordinate system (WGS84).  Through the national GPS network
website (www.gps.gov.uk), surveyors can obtain precise coordinates in the
European standard GPS coordinate system ETRS89, and instantly convert these
to the British National Grid (OS GB36) using Ordnance Survey national
standard high-precision transformation models.  Such surveys are deemed to
have absolute accuracy (in contrast to the results generated by hand-held
GPS kits which apply a spurious level of accuracy to their results).

Within the GIS, distinction must be made between information gathered
through GPS, EDM or other survey techniques.  With GPS derived data,
accompanying metadata should document the transformation process applied (if
any).

...part 2 to follow.