Learn GIS & Remote Sensing for Free
Crop LAI & Yield Sample Downloads
Wildfire Risk (Sentinel-2) Sample Downloads
Crop Agriculture

Landsat Individual Band Correlation to Corn Yield

For corn, indices using red and/or green along with NIR bands, such as MSAVI2 (red and NIR), NDVI (red and NIR), GNDVI (green and NIR), or NNIR  (red, green, and NIR) from V12 to the onset of tassel will produce sufficiently high correlations (where R² averages about 0.7). Below, based on growing degree days, tassel (VT) is between images 4 and 5. Correlation level of the different indices will vary based on reflectance amount and correlation of individual bands to yield. The indices that are correlating higher to yield and that should, therefore, be applied to a particular situation need to be determined on an image-by-image basis.

Symbology: blue = Landsat band 1 (blue band); green = Landsat band 2 (green band); red = Landsat band 3 (red band); dark red = band 4 (NIR band)

 

Landsat Individual Band Correlation to Corn Yield

Linear correlations (r) between corn yield monitor data and individual Landsat band reflectance (COST method) for fields in northwest Ohio during different times of the season from Hollinger (2011). Correlations are between reflectance and the average of clean yield monitor data within the extent of the pixel; fields are different sizes but this is not statistically important to correlation level. Above symbology is as follows: blue is Landsat band 1 (blue band), green is Landsat band 2 (green band), red is Landsat band 3 (red band), and dark red is band 4 (NIR band). Numbers correspond to images dates based on growing degree days; the darker the blue the more precipitation in the previous three days immediately prior to the image date. Image date 1 is the time when the canopy, as a rule of thumb, is overall closed enough whereby the soil influence is diminished enough for yield correlation purposes.

 

Reference

Hollinger, D. 2011. Crop Condition and yield prediction at the field scale with geospatial and artificial neural network applications. Dissertation. Kent State University.