Image is the east view from Bootlegger Saddle in the Chiricahua Mountains, AZ

Woody vegetation NDVI vs canopy cover

We recently worked on a project using Landsat 8 and 9 imagery to compute the % canopy cover of woody vegetation.  To form a relationship between the two, we used Landsat surface reflectance bands 4 and 5 from late June 2017 to create a Normalized Difference Vegetation Index (NDVI). We then compared each pixel's NDVI value to the percent of canopy cover (from a canopy cover layer) from the same timeframe and within the corresponding pixel.  The image below shows a highlighted Landsat 30 x 30-meter NDVI pixel with the late June woody vegetation canopy.  The NDVI value for that pixel is 0.1685 and the canopy cover is 0.737 or 73.7%. 

Comparing all the pixels' NDVI with their corresponding canopy cover in a study area (~35,000 acres) results in the following scatterplot and regression:

Phenology is an important aspect of this analysis. The timing of both the canopy cover analysis and NDVI analysis are at the end of a summer rainless period. As such, any application of the regression formula would necessarily need to be at the end of a summer rainless period as well. Also, any application should be very near or within the study area (Altar Valley, Arizona) with like topography and vegetation. With this in mind, given an NDVI value for a Landsat pixel the woody vegetation canopy can be calculated by using the regression formula (shown in the graph). 

---