|
|
|
ERRORS DATA STORAGE AND EDITING There is a potential for error in every coverage produced and this increases with multiple coverages in a database. It is important that the attributes are accurate in the database. Within a GIS, the storage and editing section includes multiple tools to store and maintain the map being produced. When beginning a map, it is beneficial to recognize errors that are continually being made and to fix them promptly. Also, become familiar enough with the software and data being used to avoid making errors in the first place. Grid cell positions are relative to the locations of other grid cells. Their attributes are generally what is stored within a raster system and editing is used to ensure the locations are correct. They are catalogued by these positions relative to their column/row order. Vector data is stored as a table within a database or as separate databases. Within the database, entities and attributes are separated so it is necessary to edit them differently. Tiling the data allows you to reduce the overall volume by storing portions of the database separately. This makes editing easier within these sections because you can open up only the part you want to work on. Typing errors during data input is the main cause of errors so it is beneficial to be able to focus on one small section at a time. There are three more specific types of errors found: entity errors, attribute errors and entity-attribute agreement errors. An entity error is a positional error found in vector systems. This error can be found in three different forms: missing entities, incorrectly placed entities or disordered entities. Attribute errors are found both in raster and vector systems. Typing in many attributes results in a large percentage of these errors found. Another common error is placing the attribute in the wrong grid location. Due to incorrectly placed attributes, entity-attribute agreement errors occur. To identify the misplaced grid cells display the map and compare it with the original copy. Edgematching uses rubbersheeting to align features along the edge of the current edit coverage to features along the edges of an adjoining snap coverage. This involves operating on more than one tile at a time to ensure that there is a correct match between the two tiles for entities that extend across the boundaries. Entity errors should be easily identified by referring to the topology tables because they store the information about the relationships of entities in the database. There are some basic rules to follow to help find these problems: make sure all entities that should have been entered are there, no extra entities have been digitized, the entities are in the right place and are the accurate size and shape, all entities are connected that should be, all polygons only have one label point to identify them and all are within the boundary for registration purposes. An example of an entity error is a psuedonode which occurs when a line connects with itself or where two lines intersect along a parallel path rather than crossing. These are not always errors so it is necessary to compare with original map. Dangling nodes, also called dangles, are also common errors. This is when one node is connected to a single line entity because a polygon has not been closed or the node was not connected to the object it was supposed to connect to. Attribute errors are often simply missing attributes which are easy to find. Incorrect attribute values are hard to detect though, especially in a vector system because the computer cannot identify which ones are wrong. Typing errors can occur when entering attribute data and they become matched with the wrong entities. Projection changes cause errors when coordinates do not match up. A common difference is found in latitude/longitude coordinates which are found in either degrees, minutes and seconds (DMS) or decimal degrees (DD). An easy conversion from DMS to DD is: degrees + minutes/60 + seconds/3600 This value is now in geographic coordinates and it is necessary that these coordinates match up with the original map. User commands or software can complete this transformation, but errors are still likely to occur. When working with two edgematched, adjacent coverages with the same theme it is possible to view a larger area. There are two main errors associated with edgematching. The two maps (coverages) have been created, registered and referenced separately so they will show different entity errors. Mismatches between individual objects are a common result. To ensure a good match between the two, connect all lines and polygons. Another problem is when the coverages are created with different projections because map projection is not exact. An example of a map projection error in a raster system is with the use of remotely sensed data products, such as LANSAT MSS images. "Because horizontally adjacent scenes are sensed at different times, often days apart, there is the possibility that the satellite will not be located at exactly the same latitudinal coordinates. This often results in a skew between the two images of one to several pixels (grid cells)." (DeMers, 1997) To correct the error, it is possible to move the datasets to match up neighboring pixels. Overlaying coverages of the same area, taken at different times can cause other problems. Aerial photographs commonly have this error because all of the photographs are captured at different times and the coordinates system is made from the first coverage. These coordinates will not exactly match the other photos because the view is altered due to an unsteady plane where the pictures are taken from. For an accurate match, georeference not only the corners, but also objects within the photo. "You need to be able to physically tack down objects that are in the correct place while moving the remainder to more accurately indicate their locations in space. Such a process is sometimes called rubber sheeting, in comparison to stretching a map as if it were made of rubber, but the more technical term is conflation. Conflation is an interactive process that involves making decisions about which of the coverages you wish to adjust with respect to the others to induce them to coincide." (DeMers, 1997) Use the most precise coverage to work from and select objects to tack down that have accurate locations. Do not delete the original coverages because the new map might have multiple errors. Due to the many possible errors mentioned previously, a template of the most accurate coverage is beneficial to use when overlaying multiple coverages. If the coverages have the same boundary it is possible to match up the coverages, they will often extend beyond the template's coordinates. This makes it necessary to use the coordinates from the template for all the coverages so that when the coverages are overlaid they will all have the same size, shape and coordinates. Finally, you will be left with only minor errors resulting from computational rounding and GIS calculations, which cannot be fixed by editing. At this point, the map is ready for analysis. |