As I was exploring the Spatial Statistics Resources web-page, I quickly realized most of the spatial statistical tools offered by ESRI are not applicable to my project. My project explores spatial and temporal variations of water quality (dissolved organic carbon sources to be precise) in rivers of the Willamette River Basin. Those ESRI spatial statistical tools are not applicable to my project because 1) points are not representing actual observation points of organisms or diseases for my project but rather representing water quality sampling locations that were selected by me and 2) not only Euclidean distance but also in-stream distances, flow directions, and stream networks affect statistical significance.

I found add-in toolboxes for SSN & STARS and FLoWS that address those two issues mentioned above. These toolboxes were developed by the U.S. Forest Service (USFS). Unfortunately the currently available toolboxes are for ArcGIS 9.3, but the USFS states they are planning to publish new toolboxes for ArcGIS10 later this year.

 

http://webcache.googleusercontent.com/search?q=cache:5SIzWb38eREJ:blogs.esri.com/esri/arcgis/2013/01/29/ssn-stars-tools-for-spatial-statistical-modeling-on-stream-networks/+spatial+statistics+arcgis+water&cd=1&hl=en&ct=clnk&gl=us

 

Things I would like to accomplish by the next class period are to 1) download those two toolboxes and 2) see if they seem to work with ArcGIS10. Note, I am not planning on publishing data modified using those toolboxes developed for ArcGIS 9.3; however, these goals will help me explore what kinds of tools are available through these toolboxes and learn the concept of tools that I am interested in using.

Print Friendly, PDF & Email

2 thoughts on “Spatial Statistics for Stream Network

  1. Hi Peggy!
    I’m guessing there are other tools specifically designed to work with stream flow samples and you should definitely investigate! I was thinking of an analysis, though, that might be interesting for your data. Many of the tools will work fine with sampled data. Please see the reply to Sophie’s post for a bit more about that. Also, you are not limited to Euclidean distance. Here is a strategy I hope might be helpful:
    1) Use the Generate Spatial Weights Matrix or Generate Network Spatial Weights matrix to create a swm file to reflect stream relationships. While a bit tedious, you can create a table that defines which samples are “neighbors” and which are not. For example, if sample A is upstream from sample B it would not influence B and so shouldn’t be included as a neighbor. To reflect the one way relationships on stream networks I would:
    A) Create a road network from the line feature class representing streams… make these one-way roads. Use Generate Network Spatial Weights to create your relationships. OR
    B) Use Generate Spatial Weights Matrix, then Convert Spatial Weights Matrix to Table, then edit the table so that only the valid relationships remain, then use Generate Spatial Weights Matrix to import the edited table.

    2) Okay, so once you have a swm file with valid relationships, use that to run Hot Spot Analysis or Cluster and Outlier Analysis (for the Conceptualization of Spatial Relationships parameter, select Get Weights from File).

    Hope this helps. Please contact me if you have any questions!
    Very best wishes!
    Lauren

  2. Lauren,

    Thank you so much for your comment.
    I checked your reply to Sophie’s post.

    Note for people who do not know what “a swm file” is like I was. It is a binary matrix file created when you run Spatial Weights Matrix or Generate Network Spatial weights Matrix.

    I am looking into the help menu of tools you suggested to use, and it seems like they are going to work!

    Thank you,

    Peggy

Leave a reply