Survey Mapping and Database Errors
SURVEY AND DATABASE ERRORS
The purpose of this section is to describe, in non-technical terms, the sources and types of errors likely to be encountered in survey and mapping digital databases and in overlaying or integrating these data bases. These are often manifested as apparent displacements between the cadastral, topographic and image location of the same feature, such as a formed road or track or boundary fence.
In summary there could be displacement of up to 40 meters (in extreme cases 100 meters), in rural areas where there haven’t been recent surveys conducted. This includes the displacement between the aerial imagery, topographic maps and the cadastre as well as displacement between WAMS coordinates and the physical position on the ground. For more detail, please see below:
WAMS Database Differences and Disrepancies
Users will sometimes notice differences in positions of features between the cadastral, topographic and imagery layers of the WAMS data base. For example; tracks or roads shown on the topo or aerial photography may be offset from the road boundaries shown on the cadastral layer; or the cadastral boundaries of a river may be in quite a different position to that shown on the topo and may also be different to that shown on the aerial photography. These are due mainly to the way in which the source mapping was complied and then converted to digital data. Where they occur will more usually be in rural and mountainous areas where differences of tens of metres may be apparent.
These differences can occur for several reasons:
- The topographic maps have a designed accuracy of +/- 20 to 30m, relating to the scale at which the mapping was done, 1: 50 000; where I mm on the map represents 50m on the ground. The scale of the mapping also means that sometimes adjoining narrow or linear features, such as roads or streams, have to be displaced to allow them to be shown in the correct relationship to each other.
- The imagery (corrected aerial photographs) will vary in accuracy depending on the scale or height at which the photos were taken, from around 1 metre to tens of metres in rugged terrain.
- The cadastral boundaries will vary in accuracy, depending on the scale and age of the original mapping and frequency of surveys and updates. In urban areas the accuracy will generally be better than 1 metre, increasing to a few meters in closely settled rural areas. In more rugged and remote terrain, such as the high country of the South Island discrepancies could be more than 100 metres.
- Another discrepancy can occur in the relative positions of water boundaries (rivers, lakes, and coastline) between the cadastral, topographic and imagery layers. These are usually due to changes in these boundaries (such as river banks) since they were mapped or photographed and different dates on which they may have been mapped or photographed.
Users may also notice a systematic or consistent displacement between the cadastral boundaries and the equivalent features on the topo and imagery layers. For example, the cadastral property boundaries may appear ‘offset’ for fences on boundaries depicted in the aerial photographs. This will mainly be due to the way the earlier cadastral surveys were joined up and mapped and to the lower accuracies of the time. This is not because the original survey was wrong. If the original survey pegs could be found or resurveyed they would very likely be close to the boundary fences shown on the imagery.
In this situation a practical approach is to accept that the legal boundary or road or track is where it is shown on the imagery. If necessary, the cadastral data can adjusted by identifying common key points, such as trig stations, and correcting the displacement in the data bases. In these situations there is no loss or gain of land. It should be realised that transfers and purchases of rural land are almost always made accepting the original surveys and fence lines, without any requirement for resurvey.
There may also be situations where a road formation or track may diverge from the surveyed road line. This will usually be for practical and terrain reasons and, again, the practical approach is to accept that the formation follows the legal road line.
Early surveys, generally up to around the 1870s, were of significantly lower accuracy than more recently. The level of accuracy were typically about 1-2 metres per km and would often propagate and accumulate extensively through the very large surveys of those times. These were due to a number of factors, such as equipment, surveyor competencies, scale and difficulty of terrain, lack of other surveys, records, or trig stations to check errors and the use of reconnaissance or sketch methods of survey. Accuracies since those early times were typically around 0.2 to 0.4 m per km, although these can still accumulate significantly in large surveys of difficult terrain. Modern surveys will be better than 0.1m/km.
As well as the standard errors of the time there is also the occurrence of blunders, being very large errors or mistakes. These are random and infrequent and more likely in the older surveys, due to the above factors, and carelessness, misidentification of features or survey marks, calculation errors etc.
Surveys of very extensive areas often employed lower accuracy (and hence quicker and cheaper) methods of survey, on the basis of low land values and low intensity of land use, such as large forested areas, pastoral leases and some forms of easements, including Walkways, and covenants. In these types of surveys extensive use was and is made of natural features, such as rivers, lakes, coastline, ridges and bush edges, with generalized drawing of these features on the survey plan. These days much use is made of aerial photography.
Another example is the use of dashed burnt sienna lines on survey plans or cadastral record maps to indicate the intention to ensure public access over pastoral lease areas, but without any attempt to survey or to define an actual route. These would have been plotted on the survey plans in a much generalized way and could easily be many hundreds of metres away from what over time became the recognised and used route. This practice would have been derived from the general public policy underlying land administration and settlement in NZ of ensuring public access to waterways, mountains, coasts, etc.
The digital cadastral data base was principally derived by digitising maps drawn at a variety of scales and standards, at different times and using different reference systems. Some of the maps were very old and the materials on which they had been drawn had been damaged and distorted by use and storage over many years. A variety of methods were used to correct for these error sources in the digitising process but some errors were quite random and difficult to identify and correct. These types of errors are more likely in more remote areas, due to the general lack of survey activity and need to redraw or update such maps.
There will be errors in the actual digitising process due to the precision at which the digitising can be done and the scale of the original maps. This error represents 2m at a scale of 1:10 000.
In the survey and title automation process carried out by LINZ (Landonline) the cadastral data base in urban and peri-urban areas was reconstituted using the survey information from the original survey plans and recalculating coordinates within a network of trig stations and GPS surveys. The data base accuracy in most of these areas will be better than +/- 0.5 m.
The topo data base was captured by scanning the original drawings maps. As this mapping was done to a common standard nationwide it will have a high degree of consistency of plotting at about the 10-20 m level. This accuracy will diminish for generalised features such as roads, and in mountainous terrain.
These are generally due to differences in the way the survey or mapping information was initially gathered, the relative scale of the source maps and differences in reference systems used. Differences in the structure of the data and descriptions used for features may also cause some discrepancies, although these can usually be corrected.
In its raw form aerial photography and imagery will contain quite large errors, due to perspective, camera orientation and terrain. These can be corrected to provide rectified imagery or orthophotos. These corrections are relatively straight forward in gentle or flat terrain. In mountainous terrain it is more difficult to fully correct for all these errors, resulting in some residual distortion and displacement, possibly of tens or hundreds of metres in extreme terrain.
Much public access land is defined by reference to water margins, which are liable to movement. The law pertaining to water boundaries and the effect on adjoining land titles is complex and usually can only be applied on a case-by-case basis. For "Queen's Chains" laid out before the Conservation Act ( 1990) the landward boundary remains fixed so the "Queens chain" may become completely eroded away or increased in width with accretion. Even if the land is eroded away legal roads still exist and may reappear on the other side of the water body. Marginal strips laid out after 1990 are ambulatory, that is they move with the water margin and hence remain at the original width. As a general rule the landward boundaries of Esplanade reserves do not move, but Esplanade strips do move as the water margin moves.
There are some situations where water margins have been defined by stable rocky terrain unlikely to be affected by erosion or accretion. In early surveys these features may have been surveyed to low accuracies or just sketched in. Sometimes this would be on the basis that it was sufficient, for the purposes of the survey and the grant or title, to provide a generalized outline of the natural feature defining the boundary, as it would be very stable and capable of identification and location at any time in the future. The situation arises where later more accurate surveys, mapping, or imagery indicates significant movement of the feature. However it can be assumed that the boundary feature has always been where it is and the public land (Queen's Chain) remains defined by the current position of the water margin. In these cases there is no error or actual displacement; the feature has just been positioned more accurately.