I am currently troubleshooting through some issues of trying to create hot zones or cold zones based on measured levels at multiple data points within a river system. The goal is to create zones of higher and lower values that can be identified within the river. I have tried using the Getis-Ord hotspot analysis technique using both a fixed distance and inverse distance methods but have found narrowness of the system allows too much interference across channel boundaries and across island barriers. Additionally I have tried the Geostatistical wizard ordinary kriging but again it does not take into account that points on one side of a island can not influence points on another side of an island due to flow direction of a river. Attached a photo of this
I have found that it is extremely difficult to work with interpolation methods and hotspot analysis in a more linear system. I do have the river system boundarys and the river system as raster data for depth, velocity vectors, etc if that helps? Does anyone have any recommendations for possibly making this task easier or a better approach that I am able to validate the results of?