Errors in GIS GIS is like children s art..? Errors in GIS False: If its digital, it must be correct, or at least more correct than a manual product Data errors are common in GIS: Errors from (pre-digital) map processes New errors from digital processes / GIS 1
The promise of GIS / digital mapping (1970s) Digital updating is easier and quicker than manually maps and data could be regularly and routinely updated, but only if resources are available. In practice, it works for small countries and regions but large areas (e.g. northern Canada) are not updated for decades Reduced data quality BC TRIM data 1:20,000 tiles.. BC topographic data TRIM I : air photos 1979-1989 data 7027 tiles ready by 1995 TRIM II: air photos 1996 - > 2
Whenever you work with spatial data you will deal with errors due to the steps involved in its creation. There are several ways to categorise errors: Spatial data are an abstraction of reality. We expect error due to: Traditional and digital mapping How we conceptualize the data How we collect the data How we present the data Types and sources of errors in GIS 1. Types of error: a. spatial or b. attributes 2. Sources of error: a. Instruments b. Human error [c. Change /updating data quality] 3
Types and sources of errors in GIS 3. The 'errors' that occur during the four steps of GIS: a. Input b. Database management c. Analysis d. Output 1. Types of Errors (spatial or attribute): a. Positional accuracy Varies according to scale /equipment e.g. BC TRIM (1:20,000): positional accuracy = 10 metres b. Attribute accuracy Classification and measurement may be incorrect through: - human error (misinterpretation) - vague contrast or high heterogeneity (variability) 4
TRIM II data and orthophoto near Trutch (Alaska Highway) Alaska Hwy (paved, but coded as gravel) 2. Instrument inaccuracies and human processing? different agencies using different methods? 1888 1:40,000 1982 NTDB 1:50,000 Contour Interval 200 Interval 40 metres Note the Lack of gulleys on north side of Bourgeau Lake 5
The same area 1999 ETM+ 1999 GPS errors Types of Error Ionosphere Clock Ephemeris Troposphere Receiver Multipath 4.0 metres 2.1 m 2.1 m 0.7 m 0.5 m 1.0 m Total 10.4 m ( Selective availability ~ 100m) Pre-2001 Corrected by differential GPS (DFGPS) 6
Selective Availability (SA) The random error, added to GPS signals before 2000.. up to 100 metres error by scrambling last 3 decimals of time signal Turned off May 1, 2000 at midnight; No intent to ever use it again May 1, 2000 Selective Availability on May 3, 2000 Selective Availability Off Differential Correction industry solved the problem of SA Uncorrected GPS ~10m Corrected (DGPS) ~1m 7
3. GIS errors a. Input: Digitizing: human error and the width of a line Dangling nodes (connected to only one arc): permissible only in line themes Topology is needed for GIS analysis [connectivity, adjacency, containment] TRIM data: initially were spaghetti Input -Digitizing errors Results: loss of topology Disconnected lines Incomplete polygons Non-adjacency Extra line lengths Additional small polygons 8
Input -Digitizing errors pseudo-nodes b. Database management Mistyping entry values Missing (null) values Metadata how / when were data gathered etc.. Units of measurement e.g. feet versus metres Data updating e.g. ongoing tree growth 9
c. Data analysis Overlay of layers, digitized separately from different sources or scales, e.g. soils and vegetation. They have common borders, but slight differences cause 'slivers'. When we overlay polygons, the resulting polygon has many small polygons that are due to the representation of the polygon boundaries. These smaller, or sliver polygons, represent spatial errors in the data Data analysis The compounding effects of overlay of multiple layers: e.g., if two layers each have 80% accuracy, the accuracy of the resulting overlay is ~ 64% Inappropriate or inadequate inputs for models Dubious classifications and processing Inconsistent processing methods 10
Inappropriate or inadequate inputs for models (colours are good ) Locations theft or where the car was found? UNBC? legend and scale - units Dubious classes 11
Processing always examine hillshade for errors in DEM Striping due to projection error raster reprojection (error) ASTER DEM (Chile) with data holes from clouds (data quality) DEM error in Google Earth Mt. Robson; error = ~100 (30m) AB 30m BC 12
Errors in GIS/GPS data - connectivity (e.g. the road connecting X to H does not exist) H x Google Maps screenshot Cranbrook Hill, Prince George Human error based on misinterpretation of GPS data When GPS goes awry: Globe and Mail, Oct 7 2010 13
Prince George 93G015 2000 / 2010 roads updated from 1980, but nothing else..need for documentation of date and collection methods in metadata www.bing.com roads that don t yet exist! : Cranbrook Hill X ** All roads in the western third of the map except those overprinted do not exist 14
GIS errors: Output Agricultural lands drought conditions Colour palettes: colours don't match from screen to Printer (same file) 15
Colour - RGB (monitor) versus CMYK (printer) Translucent (monitor) versus opaque (printer) RGB: additive colours CMYK: subtractive colours Cyan, Magenta, Yellow, BlacK Review Things to consider from this lecture and also data quality : GIS data overall can have more errors than printed maps Lack of documentation - the need for metadata Age of GIS data (relative to rate of change) The implications of combining data layers Effect of area jurisdictions - e.g. provincial differences The challenges of mapping a large province and country The impact of mapping scales Geographer s Craft: http://www.colorado.edu/geography/gcraft/notes/error/error_f.html 16
http://gis.nic.in/gisprimer/errors.html Cascading errors 17