I have been using the Natural Resources Canada (NRC) data for various generalized maps but it is not until now when i overlaid higher resolution features obtained from the provincial government (Ministry of Natural Resources Land Information Ontario) that I notice a shift in the the NRC Hydrology (watercourse, water bodies) or approximately 800m.

What is strange is that all other data from the same source such as transportation, places, etc. match perfectly with when overlaid with the higher resolution provincial data.

The Provincial data is in NAD83 UTM16 and the Atlas Canada data in WGS84.

ArcMap is projecting on the fly which is usually not an issue (other than performance).

alt text

3 Answers 3


If anyone is interested here is a reply from the data supplier regarding the issue: Hi Jakub Thank you for contacting GeoGratis about data precision issues found with the National Framework data collection for Hydrography at 1:1M scale. As you may be aware, most data collection offered on GeoGratis are static and we are unable to provide product support for this data. In this case, if we view the meta data information document that accompanies your 1:1M data file you will note that the precision for the Horizontal accuracy should be within 1000 metres. ( http://www.geogratis.gc.ca/download/frameworkdata/hydrology/cartographic/1M_HYDRO_NETWORK_EN_2008.doc ) ''Horizontal_Positional_Accuracy_Report: Positional accuracy of the data is approximately 1000 metres. This average positional accuracy estimate is based on tests performed on source data by the source agencies. As a result of the generalization and integration process the accuracy is no longer absolute, but relative to other 1:1,000,000 scale frameworks layers.' Therefore 800 metres shift falls within the product specified horizontal accuracy As I have shown the Framework Hydrography data is not that accurate to begin with and should be used for viewing purpose only. Therefore I can suggest you download the 1:250K NTDB data for 042M hydrography. At least your accuracy would improve to 150 metres horizontally. See attached image. Hoping this information explained why the Framework data is shifted. Thank you in advance for your understanding

  • Thanks for taking the time to pass on the response from NRC Jakub. I suggest marking this as the accepted answer (yes it's perfectly acceptable to change the accepted answer when a better one comes along, even if it's your own). Commented Dec 21, 2010 at 18:59

As Melita Kennedy notes, this is not a datum issue, and unlikely even to be a projection issue. My hunch is that the data layers have a different heritage. The hydrology likely came from the venerable gross scale Digital Chart of the World while places and transportation were generalised from the newer and higher resolution provincial data. In other words it's not a mistake, just a common problem when combining data from different scales and sources. I'd ask NRC to be sure though.

A way to fix it yourself is to use the Spatial Adjustment tools. You'll have better luck using the centroids of waterbodies and islands when picking the to/from link points than things like intersections of creeks. Not sure why that that's the case, but it's true in our experience.

  • Good point (different heritage). I contacted NRC about this and am awaiting response. I was hoping not to have to use spatial adjustment as these large provincial datasets. Commented Dec 16, 2010 at 18:59

GSRUG http://www.geod.nrcan.gc.ca/tools-outils/gsrug_e.php

It is different with every province in Canada http://www.fmepedia.com/index.php/Canadian_Spatial_Reference_System_%28CSRS%29_High_Precision_Datum

lots of info here: http://www.geod.nrcan.gc.ca/faq_e.php#faq1

  • is there a simple way to do this in ArcGIS? How can i easily transform the data from GCS_North_American_1983 to WGS84? Commented Dec 16, 2010 at 15:46
  • 2
    800 meters is not a NAD83 / WGS84 difference. They shouldn't be offset more than 1-2 meters. So, none of the existing NAD83-WGS84 transformations in ArcGIS is going to help. 800 meters is too big even for NAD27 / NAD83.
    – mkennedy
    Commented Dec 16, 2010 at 17:08
  • What scale is the generalized data at?
    – mkennedy
    Commented Dec 16, 2010 at 17:20
  • the data id generalized for 1:1,000,000 Commented Dec 16, 2010 at 18:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.