Visible Band and NDVI Saturation & Near Infrared (NIR) Variability for Corn and Soybeans
(See Landsat Correlation to Crop Yield Folder for Correlation Levels and Patterns)
Landsat visible bands will eventually saturate (values become very similar throughout a field even though crop condition varies) for corn, soybeans, wheat, and other crops. The plots below (Hollinger [2011]) show how many different Landsat pixel values there are at different growth stages throughout seasons for different size groups of pixels (average pixels in group for corn is 79.2 [CV = 46.4 percent]; average pixels in group for soybeans is 85.1 [CV = 49.6 percent]. The plots start at V12 for corn and about R1 for soybeans because this is about when soil visibility diminishes enough to have an overall insignificant enough effect on correlation to yield. The darker the blue number the more rain there was in the three days prior to the image date. Corn growth stages are from V12 to about senescence (1 is V12 and 6 is first image with tassels); soybeans are from about R1 to beginning maturity. Band 1 is blue, band 2 is green, band 3 is red, and band 4 is NIR. Overall, the data show that, for corn or soybeans, saturation strongly affects visible bands during the reproductive stages and the NIR is much more variable during these times, particularly for soybeans.
Yellow = corn, Green = soybeans
Blue 
Green 
Red 
NIR 
Red and NDVI saturation for corn, soybeans, and wheat (Gitelson, 2004; pdf)
The graphic below from Gitelson (2004) shows red reflectance and NDVI saturation as NIR reflectance increases (as leaf area index or vegetation fraction increase) for corn, soybeans, and wheat. Even though NIR reflectance increases (which does not saturate), the strength of the red saturation outweighs the NIR variability and causes the NDVI to also saturate.
References
Gitelson, A.A. 2004. Wide Dynamic Range Vegetation Index for remote quantification of biophysical characteristics of vegetation. Journal of Plant Physiology 161: pp. 165-173.
Hollinger, D. 2011. Crop Condition and yield prediction at the field scale with geospatial and artificial neural network applications. Dissertation. Kent State University.