Keelhauled: Problems and Errors

All models are generalizations of real world phenomena and can never include all the complexities of the real world. This means that analyses using a GIS will have a number of errors inherent to them. However, results can be improved by following the 'garbage in, garbage out' principle. This analysis also employed the 'dead men tell no tales' principle which is of standard practice to pirates. So as to avoid being the victim of the implications inherent to this principle, effort was taken to obtain the highest quality of available data. Even still, various problems and errors with the analytical process can be identified. Furthermore, the importance of considering potential problems and errors cannot be understated as they could undermine the results of the analysis.

Sources of error:

Errors can be broadly grouped into two categories: spatial errors and temporal errors.

Spatial error:

Spatial errors refer to the possibility that data in the layers used for analysis are not in the same location as they are in the real world. As most of the data used in the multi-criteria evaluation undertaken for Bloody Henriques the Englishman was from secondary sources, it is possible that some layers contained errors that were then propagated throughout the whole analysis. For example, while care was taken to accurately record the postal codes for each BC Liquor Store in the GVRD, these codes could have been recorded wrong or entered into the web site wrong in the first place. Similarly, some could have been overlooked all together leaving the dataset incomplete.

Temporal error:

Temporal errors reflect the fact that geographical entities are often not stable through time. The date of data collection is, therefore, an important consideration in terms of the usefulness of the results generated from a particular analysis. In the analysis undertaken for Bloody Henriques the Englishman, the most recent data available for both landuse characteristics and census data was 2001. Given the rapid pace of growth and development in many areas of the GVRD it can be asked how accurate the landuse characteristics and population values are five years since initial collection?

The use of distances from oceans and navigable rivers is another potential source of temporal error. Shorelines are unstable by definition. Ocean shorelines are subject to daily tidal fluctuations while river shorelines are subject to periodic flooding and recession related to the amount of received precipitation and snowmelt. The dynamic nature of shorelines could affect the distances calculated from oceans and navigable rivers. This could, in turn, mean that some areas deemed suitable will actually be unsuitable when Bloody Henriques and his crew arrive in the area while others that were reported as suitable will actually be unsuitable.

Problems with the analysis:

Various problems can be identified with the analysis carried out for Bloody Henriques the Englishman. The most significant problems with this analysis are thought to involve data choices and semantic interpretations.

Data choices:

While care was taken to collect quality data relevant to the problem being considered, some potentially important sources were ultimately left out. When obtaining data on liquor store locations in the GVRD, only BC Liquor Stores were considered. There are, however, a number of cold beer and wine stores in the region that could have been included as well. It is felt, though, that since many of these stores do not carry the desired 'grog' (rum in this case) or do not offer it at a reasonable price this omission will not significantly affect the outcome.

Data on soil characteristics could have been useful for analysis as well. In order to dig a large enough hole in which to place the treasure chest, soils must be relatively deep and dry. However, while soil data was not included, it is felt that this omission is compensated for in the consideration of distance from navigable rivers and oceans and of constraining slope parameters. This is because dry soils will usually be located at distance from water bodies while deep soils will generally occur on relatively flat slopes.

Semantic interpretations:

The appendix containing pirate terminology demonstrates some of the interpretation problems experienced when considering appropriate constraints and factors for analysis. Research into the pirate language greatly improved the language barriers, however, problems still arose. The most significant problem involved whether or not some areas under consideration would indeed be navigable to the Marauding Sea Monkey, the schooner captained by Bloody Henriques the Englishman. Descriptions of the area on my part and of the schooner in question on the part of Bloody Henriques were attempts taken to reconcile the issue. However, it will remain to be seen whether or not these attempts were in vain.

There was also no mention of whether or not sailing past areas of relative high population density or of commercial zoning would decrease suitability. As the schooner in question (pictured below) is quite an unusual vessel for the waters around the GVRD, there is the potential for it to be noticed as it sails past highly populated areas. It was not determined whether or not the schooner had any sort of disguising techniques at its disposal such as submarine sailing which is a common characteristic of cursed ships. On the other hand, navigability can be overcome by the use of 'long boats' carried on the ship.