I'm working with a database of land parcels and their corresponding crops per year (source)

My end goal is to create a predictive model for crop type, given land parcel. For example, if in the past 4 years a farmer grew soy, soy, corn and corn, what is the likely crop he'll be growing in 5th year.

To make the task a bit straightforward, I'm checking if the parcel polygons are the same (or similar) from year to year? If that's the case I should be able to abstract the geo info away and work directly with "parcel ID", crop type, and year data.

The gdb file for each year is quite sizeable and just loading and preprocessing in R studio takes quite a while.

What is a sane way to compare ploygon data across years and extract those exact parcels which appear in consecutive years? Would using the center of each polygon or labpt be a good way to represent the polygon?

I'm using Mac Sierra and rgdal library. An entry from 2009 gdb after converting to GeoJson looks like this:

{ \"type\": \"FeatureCollection\", \"crs\": { \"type\": \"name\", \"properties\": { \"name\": \"urn:ogc:def:crs:OGC:1.3:CRS84\" } },
\"features\": [ { \"type\": \"Feature\", \"id\": 2, \"properties\": { \"CAT_GEWASCATEGORIE\": \"Grasland\", \"GWS_GEWAS\": \"Grasland, blijvend\", \"GWS_GEWASCODE\": \"265\", \"Shape_Length\": 898.222510, \"Shape_Area\": 4885.337158 }, \"geometry\": { \"type\": \"Polygon\", \"coordinates\": [ [ [ 3.98569255494544, 51.400796253376114 ], [ 3.985792633157566, 51.400648243993814 ], [ 3.985454735105648, 51.400542942730567 ], [ 3.985419563702146, 51.400544767402884 ], [ 3.985367375532934, 51.400571777661298 ], [ 3.985239821317514, 51.400662957371289 ], [ 3.984954651582088, 51.400851050506539 ], [ 3.983832093498555, 51.401616824104352 ], [ 3.982416851757605, 51.402598958984925 ], [ 3.981780494933279, 51.403048420036697 ], [ 3.981523333282359, 51.403247946509204 ], [ 3.981396023...

This is my first geo project.


I noticed that quite a lot of polygons repeat across the years but with very slight variations in borders; so using labpt as is won't work.

2009 (red) and 2016 (green) 2009 and 2016

Only 2009 (red) only 2009

closed as too broad by Vince, whyzar, John Powell, Mapperz Jan 23 '17 at 15:44

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • I question the validity of your assumptions. Competent farmers rotate their crops, so last year's soybean field is likely to be corn or alfalfa or fallow. – Vince Jan 23 '17 at 12:03
  • @Vince yes that's true and that's what i'm going to predict. That is, if in past 4 years he grew soy, soy, corn and corn, what is the likely crop he'll be growing in 5th year. – Md Oliya Jan 23 '17 at 12:17
  • GIS SE uses a "Focused question / Best answer" model. This question does not appear to be focused enough. Data comparison ("conflation") is a huge topic, and there are many sane techniques. I suspect you will need to use GIS software, not just a statistical tool and a conversion library. – Vince Jan 23 '17 at 13:19