New answers tagged remote-sensing
In the example you are referencing, NDVI is included as a predictor variable along with all of the band values. The response variable is the class (vegetation type). factor(class) ~ band1 + band2 + band3 + band4 + ndvi in your case, you could simply have a binary response (cover, or non-cover). Random forests is a very valuable machine learning ...
As per section 9.9 of the TIMESAT manual: Convert all images into flat (header less) binary format.[...] File conversion can be carried out with software such as the open source Geospatial Abstraction Data Base Library (GDAL) On GDAL Raster Format that suffices would be ENVI .hdr labelled Raster You can use gdal_translate to convert your GeoTiffs ...
In Landsat 8, when you download the geographic image it comes with a quality band (QB) that is primarily meant to show clouds. Maybe that will help?
I would recommend Landsat or MODIS for this task. You can access Landsat or MODIS data through Earth Explorer or Reverb|Echo. There are a number of publications on the subject to help you along the way: Brekke, C., & Solberg, A. H. (2005). Oil spill detection by satellite remote sensing. Remote sensing of environment, 95(1), 1-13. Hu, C., Li, X., ...
Lars Eklundh: It is possible to monitor one or several classes in TIMESAT. The option to use several classes is mainly necessary if the phenological structure is very different between the classes, i.e. different parameter settings are necessary for the different classes. In your case it might be possible to use the same settings for both classes, but ...
I use Beam-Visat for this purpose, which is a free program.
You need to select the "Show All" box in the Index Options GUI.
Sometimes distributors drop the blue band and provide only the nIR, red and green bands so that the end user can view the data as a false color composite image. Let's assume the distributor did that. There are several ways you can deduce which bands are which, especially at the red and nIR wavelengths, using basic remote sensing principles. For example, ...
The USGS phenology project might inspire you a little. http://phenology.cr.usgs.gov/other_resources.php
Just a tiny error, as far as I can see. Your substrings are incorrect. This can be seen by comparing the result from a 'which(df$bits="0000100001000100")' with a number of observed unique values, which can be seen in ArcGIS when colouring the tif-file by unique values. 00001000 01000100 = 2116, and there are 3891233 of that number in both ArcGIS and R. This ...
Top 50 recent answers are included