I have a number sequential GPS locations from marked animals - usually 4 to 5 locations per day for 1.5 years - and am trying to delineate seasons based on elevation and distance.

From low elevations in winter, animals generally move to higher elevations in summer. The seasonal movements span x, y and z dimensions. I can calculate the Euclidean distance between two points (or seasonal ranges) to account for change in x and y, and also calculate the change in elevation to account or z. However, I am wondering if there is a way to integrate Euclidean distance traveled and change in elevation.

The hope is to calculate a single term that accounts for both change elevation and Euclidean distance between seasons.

How can I do this in ArcGIS for Desktop?

  • Why would you not just calculate the Eucidean distance in all 3 dimensions (converting the elevation to the same unit as needed)? It's the same formula, just with the extra squared delta added in. – Evil Genius Aug 20 '15 at 18:40
  • Careful, do not use the Euclidean distance formula on latitude and longitude coordinates. You will need to convert those coordinates into projected X/Ys first. (UTM, State plane, etc) – Mintx Aug 20 '15 at 20:29
  • 1
    Why settle for a single term? A pragmatic way to put your question might be "what is gained by keeping the distance and elevation terms separate in my model"? That is answered by comparing the model with both terms to any model in which they are combined. A good start to combining them, therefore, would be to fit the full model with both terms (along with any other factors you have available). The estimated coefficients of distance and elevation will give a good starting estimates for the relative weights in the reduced model you seek. – whuber Aug 21 '15 at 13:45

Convert points to individual lines using Points to Line (Data Management). Place points along the line at regular interval. To do so you might use linear referencing, alternatively search this site, something will pop-up.

enter image description here

Calculate distance for each point along the line (Chainage in below table) . Search this site, let me know if fail, I’ll post script or field calculator solution. Sample your elevation model to assign Z values to points. Scale Chainage and Z fields using relevant ranges, so that values are in range (0.00…1.00):

enter image description here

Export this table and use add XY data (lScaled, zScaled) to convert table to points. Remove prj file if any from output points. Convert points to individual line using Points to Line (Data Management).

Play with smooth line to reveal your seasons:

enter image description here

It is likely your seasons boundaries are somewhere on steepest parts

  • The GIS technique exhibited here is good. However, the impression of boundaries in the plot strongly depends on the (arbitrary) choice of relative scaling of the distance and elevation. This approach to the analysis therefore just hides the difficulty rather than solves it. – whuber Aug 21 '15 at 13:47
  • Single term here is vertical gradient with aim to find 'spring' and 'autumn' points. For 1.5 years of observation the line shown gives the answer, I'd say spring is at point (0.2,0.4) . Next step is to average this reading for ind.head curve. I am assuming these animals aren't leaving on rock face, thus to analyze longsection of their travel we need to exagerate y-axis. Very common approach, in fact I never saw longsection of real terrain with equal scale factor for both axis. – FelixIP Aug 21 '15 at 22:03
  • Perhaps 1.5 long axis is more apprpriate here. Will it make scaling less arbitrary? Variablle normalization it is – FelixIP Aug 21 '15 at 22:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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