Stratified sampling using a polygon layer
- RivEX+ can stratify sampling based upon a user supplied polygon layer. This tool is ideal for generating a layer of random points across a network grouped by polygon or a sub-set identified by an existing selection on the polygon layer. Such output could feed into further river analysis or compliment a survey strategy based upon catchment or administrative boundaries.
- You have a variety of options that can influence (bias) the selection process and it is important that you understand these. Click here to review these options.
- The output is written to the default directory C:\RivEX\output. Each point stored will contain the following information:
- Unique ID
- ID of intersecting polyline
- XY coordinates
- % along polyline
- Polygon ID the point is within
- Once you have chosen your river network and polyline ID, click on the Polygon Sampling tab. You must first select the polygon layer, this must be pre-loaded into ArcMap. If a selection exists on the polygon layer you can choose to use these instead of the entire dataset.
- Select the polygon ID field. Note for technical reasons the OID field is not offered in the drop down. If your Polygon layer does not contain a field of unique ID's, you must build one. Click here on how to create a field of unique ID's.
- RivEX+ will attempt to locate a random point but if it fails it will try up to 200 times. If after the 200th attempt it is still unsuccessful it will bail out. This means polygons could generate less than the specified number of points.
- If your polygon layer contains a pre-calculated field, holding the required number of points per polygon you can select this.
- If you are sampling a percentage of network then you need to provide a footprint size ( in meters) and percentage cover of network. The footprint is a linear length that the sampling point represents. In the image below a user has entered a footprint of 1000m and they want to sample 10% of a network.
- The number of points created is based upon the footprint size and the desired percentage of network cover within the polygon. For example a polygon identifying 89Km of river would generate 9 sampling points if each point had a footprint of 1Km and the user wanted 10% coverage. This calculation treats each footprint as non-overlapping when determining the number of points required but the output does not displace points when choosing the random XY coordinates. Therefore large sample sizes will create the required number of points to sample a network with non-overlapping footprints but can actually sample less network because the placement of the point does not consider existing footprints (unless you choose the exclusion zone option).
