https://blogs.esri.com/esri/arcgis/2010/07/13/spatial-statistics-resources/ the links don't work, something about a redirect.

Anonymous User

Spatial Statistics Resources | ArcGIS Blog 

the links don't work. Is there a website issue?

Database Error 

Anonymous User

Dan Patterson‌ Jenora D'Acosta‌ thank you!  Can email to me my contact info is SAGIS.org or I will send you a PM if you follow me (I believe that is how it works? I followed you Jenora)

And, is this the most up to date book on Spatial Analyst? http://downloads2.esri.com/support/documentation/ao_/Using_ArcGIS_Spatial_Analyst.pdf 

One question I'm trying to ascertain the answer to is how to calculate amount of features per cell from point density. It produces 'magnitude'. I just want to know how many trees per cell this actually means. Any explanation thus far I have found I do not understand. I understand if it's circular the neighborhood area will be given by pi*r2. However the examples in Ch 3 in David Allen's Spatial Analyst WorkBook (page 130) I just can't figure out simply calculating the number of trees per cell as in the example, instead of this 'density' or 'magnitude' that is a much larger number, the thousands, than what should be a few trees per cell at most (6meter).  In fact it wouldn't it make more sense to just give this value anyhow?    I also am a little hazy on the following the point of the 'cell size' guidance of 10 to 100 cells, in the accompanying Esri Guide to GIS Volume 1 in the Density Chapter. I have looked a good bit and not found much online. Thank you to anyone with any comments. 

Aaand update: I'm working through it. I realized I needed to search in Spatial Analyst forum. Threads like these appear relevant. https://community.esri.com/message/242698?commentID=242698#comment-242698  How exactly does the Point Density tool calculate cell values in SA?   it is per neighborhood area with that cell as center, I will test and confirm tomorrow with the text's examples and calc by hand to ensure I understand it.

Update 2: it appears to pan out. I backcalculated feature count at a cell (well, for the neighborhood that is centered upon the cell, as these threads above explain).If I understand correctly the cell value is the magnitude, i.e. value for the WHOLE neighborhood, with the particular pixel (cell) in question, when you look at a single cell’s value. To back-calculate actual number of features, multiple the area by the cell value, keeping area in map units with cell size of one. (circular neighborhood area    A = π × r²)  Example:  if two points overlap, value in between should be 2. Value within neighborhood not overlapping should be 1.  Value outside neighborhood should be zero.  64*0.001592 = 2   (Example from text but with neighborhood of 20, cell size 1, area in map units squared, cell magnitude of 0.001592, and generating a value of 2.0 in calculation)  I see now that density surface is not an interpolation of predicted values, but of the count of features in an entire neighborhood, centered upon each cell.  The scalar of square area unit is simply what would need to be divided by so I keep it the same by using Map unit sq. Is this all correct?

I have found calculations now to pan out as above, but the example using 'trees' in Texas in the Workbook appears to violate the assumption of random samples!

How should I interpret the output of density tools? | ArcGIS Blog