Imagery for Vegetation Analysis

Question

Original User: rdg129

I am starting a project that involves classificaiton of native vs. non-native vegetation.

Can anyone tell me what imagery would work best for this application.

thx in advance.

6-12-2014 9-27-51 AM.png

6-12-2014 9-23-56 AM.jpg

Anonymous User · Answer

Great insight, Jeffrey, Very helpful and a good survey of what's out there.  WorldView-3 launched this past summer and we will be releasing a much richer set of spectral data with this sensor.  Coming soon as well is its 30cm spatial resolution which will open up the doors to new applications.  This enhanced comes at a cost of total global coverage of course as this will be an "HD satellite".  🙂

P

JeffreyEvans · Answer

You are being offered advice based on the "cult of high spatial resolution". Bigger is not always better. There are instance where high resolution data is quite appropriate and, since it is free, calibrated Red-Green-Blue-NIR, your first stop should be NAIP (http://www.fsa.usda.gov/FSA/apfoapp?area=home&subject=prog&topic=nai). Unless there is a notable difference in the photosynthetically active radiation between the vegetation types, a metric like NDVI will likely not provide adequate spectral separation.

You need to base your imagery decision on the intended classification goals. There is a bit of art invloved in selecting appropriate imagery that represents a balance between spatial, spectral and temporal resolution. There are instances where, with the addition of the SWIR band, 10m SPOT 5 will outperform 1m Quickbird because the spectral discrimination is more appropriate in the short-wave IR range. If you have vegetation the is not readily separable in the visible and NIR range then high resolution data buys you nothing. I have been very impressed with the new Landsat 8 16bit data. If your problem can be addressed with moderate resolution (30m) data then this would be the option I would aim you towards.

That said, the best satellite data I have worked with is WordView 2. The spatial resolution in the 8 multispectral bands is 1.85m and 46cm in the panchromatic. The addition of the red-edge and blue-edge bands provide amazing discrimination of many vegetation types. It is, however, quite expensive. You can acquire information on WorldView 2 as well as QuickBird, IKONOS and GeoEye on the DigitalGlobe website (http://www.digitalglobe.com/). If you have an academic affiliation DigitalGlobe has a special pricing schedule and federal partners have free access through various sources.

Anonymous User · Answer

Original User: mark909

Look up NDVI. 25cm resolution aerials should do the trick