conversion of forestland to agriculture in hubbard county, minnesota by: henry rodman cory kimball...
TRANSCRIPT
Conversion of Forestland to Agriculture in Hubbard County, Minnesota
By: Henry RodmanCory Kimball
2013
Hubbard County
• Hubbard county is in North Central MN– 639,360 acres– 28 townships
• Agriculture is growing as part of the land use– Crop prices are increasing – Potatoes are a main crop in the County
• RDO/Lamb Weston is a large company based in Park Rapids
Why is this happening???
• CRP land grants are up• People need money so
they rent their land• Some farmers are retiring
and larger farmers are buying their land
• Cropland is more profitable than forestland
Remote Sensing as a Tool
• With remote sensing we can really see change in the landscape over the years
• Provides a ‘bird’s eye view’ on the landscape
• With access to the right orthorectified imagery and software we can do this
• We can compare maps throughout the years
Research Interest
• We are interested in this research since we are going to school for forest management
• It was a long process to get the right imagery • Conversations with the Hubbard County Land
Department– Mark Lohmeier – Hubbard County Commissioner– Kevin Trappe – GIS Coordinator for Hubbard Co.
Field Observations
• Visit to Park Rapids, MN– Snow was very deep, but Cory was still able to
document deforestation
Slash PileNew Irrigator
Software Resources
• LandSat - USGS– Imagery acquired from the GLOVIS browser
• Photos from Past Years – 2000 – 2011• Landsat 5 and 7 TM
• Arc GIS 10.1 – Used to display isolate shapefile of Hubbard
County– Shapefile
• ERDAS Imagine 2011– Used to derive Classifications of different
land uses– Accuracy assessments
Henry hard at Work
ERDAS can give a man a headache
Imagery
• Took a few different trials to decide• Decided on Landsat 5/7– 7 band– 2 different images
• Images from 2000 through 2011 were acquired– Full coverage of Hubbard County– Main focus on agriculture is in the South half
Image Classification
• Supervised Classification– 10-30 training sites per class– Merged them together into similar classes (8)
• Classes Included:– Forest land – Agriculture– Water – Clear Cut Forest land– Developed– Wetlands – Shadow– Clouds
Results of 2000 Imagery
Landsat Classified
Results of 2009 Imagery
Landsat Classified
Classified Images Compared
• 2000 compared to 2009
Difficult to see change with the whole county. The change detection slide will tell the details of the change.
Change Detection
We can see a very slight change in forest cover from 2000 to 2009
Accuracy Assessment2000 Accuracy
2009 Accuracy
Class Name
Reference
TotalsClassified
TotalsNumber Correct
Producers Accuracy
Users Accuracy
Non-forest 14 15 9 64.29% 60.00%Clouds/Shadow 12 15 12 100.00% 80.00%
Forest 19 15 12 63.16% 80.00%Wetlands 11 15 6 54.55% 40.00%
Water 19 15 15 78.95% 100.00%
Totals 75 75 54
72.00%
ACCURACY TOTALS
Overall Classification Accuracy
Problems Encountered
• Doing this project on a whole county is tough• Small issues with LandSat
– Clouds– Lines– Low resolution
• Downloading Imagery was time-consuming – Attempts at using NAIP Imagery took many trials
with ERDAS IMAGINE • Accuracy is questionable• Some areas were classified incorrectly
– It would be nice to personally see all the landscape
Conclusions
– Field observations indicate a significant change in land use, but our results do not reflect that trend
References
• http://glovis.usgs.gov/
• Mark Lomeier – Hubbard Co. Land Commissioner
• Kevin Trappe – Hubbard Co. GIS Department
• http://earthexplorer.usgs.gov/
• Labs 7-12 FR3262