Perfroming a weighted overaly with multiple rasters....

Question

I have been doing some research, and I believe a weighted overlay is going to be my best approach, unless someone has other suggestions...

My situation....we are a crop consulting company in Iowa, and are working on developing management zones for our growers fields based on multiple layers of data. These data layers include spatial yield, soil types/topography, aerial image of bare soil, elevation data, among others. We would like to overlay all of these to produce polygon management zones for each farm. My main question is do all rasters need to have the same cell size for the analysis to work properly?

I'd also welcome any tips or suggestions on how to accomplish this. We ultimately want clean polygons of 5-6 acres each.

Thanks!
Bryan

MatthewBaker2 · Answer

My main question is do all rasters need to have the same cell size for the analysis to work properly?

Your analysis will only be as accurate as your least accurate data ... meaning if you have a mixture of low-resolution and high-resolution rasters, eventually you'll end up either generalizing your accurate data, or 'filling in' your least accurate data.

In any event, make sure you use a snap raster, so even if the pixels are of different sizes, they will line up with snap raster. Set the snap raster to your most accurate data, I'd say...

curtvprice · Answer

My recollection is they are converted to the same cell size prior to processing

Yes, this is done according to the settings of the geoprocessing environment. It does make sense to do the resampling ahead of time using tools like Resample and Aggregate so you have full control of how the resampling happens.

Once you've done your analysis, there is a nice set of tools for post-processing of areas. You can analyze contiguous areas using tools like Region Group and Zonal Geometry, and manipulate the results to get the polygon sizes you want using tools such as Aggregate and Nibble.

DanPatterson_Retired · Answer

My recollection is that if they aren't they are converted to the same cell size prior to processing...that said, it would be best to standardize all inputs before doing all analysis, that way you are sure of the inputs