Continuing our series on the use of RapidEye data in geospatial projects, for this edition I expand on a discussion started in February 2012, focusing on the value of RapidEye in vegetation analysis. This month, I filter and map a Red Edge Normalized Difference Vegetation Index (NDVI) calculated from 5-meter RapidEye imagery to make more practical use of the information it can provide.
The RapidEye Constellation
RapidEye is a constellation of five 5-meter medium resolution satellites each offering five spectral bands of information. The RapidEye constellation offers a daily revisit time to every location on the planet with a huge footprint that is 77-km wide. The data is priced competitively with a starting cost of $1.28 per square kilometer for all five spectral bands – academics receive a discount on this price. RapidEye adds a fifth band, the red edge, to the 'traditional' multispectral set of blue, green, red and near-infrared (NIR). The additional spectral data available in the red edge band allows users to extract more useful land 'information' than can be extracted from traditional 4-band imagery sources.
Red Edge NDVI
Red Edge NDVI is a way to assess the health of a plant or any living creature (such as cryptograms in the desert and microbes suspended in a water column) that contains chlorophyll. In the recent months, I have spoken with several academics who have tested imagery with a red edge band. They have found that Red Edge NDVIs have higher correlations with field measurements of plant health, and are working on academic publications with this evidence. If you would like more details on Red Edge NDVI and how it is calculated, please refer to my February 2012 article.
Putting Red Edge NDVI to Work
The Band Math function of ENVI 4.8 was used to calculate a Red Edge NDVI and then to color balance a natural color 5-meter RapidEye imagery collected over the San Luis Valley, Colorado on June 19, 2011. Exporting these layers as a GeoTIFF from ENVI, I moved over to ArcMap to finish off the analysis of the valuable information we have extracted with this Red Edge NDVI calculation.
The first step I took in filtering and displaying my Red Edge NDVI in ArcMap was determining the lowest NDVI value that ties to healthy, green vegetation. NDVI has a range of possible values from -1.0 to 1.0. To decide this value, I added my color-balanced natural color imagery as the top layer and then added the NDVI layer below. By clicking on multiple pixels with the Identify Tool that appeared to be on the edge of live-dead vegetation, a Red Edge NDVI value of 0.1 was the lowest I found for green-ish plants. As such, I used a NDVI value of 0.09 (to be on the safe side) or greater as the cut off for living vegetation. This value could have easily been adjusted upward or downward as the fidelity of the analysis required.
For the next step in this analysis, I created sequential classes of Red Edge NDVI values and then color coded them accordingly. To complete this step, I navigated to the Symbology Tab under the Layer Properties of the NDVI TIFF file. Now, I switched the symbology type to Classified and calculated histograms when an ArcMap warning to do such appeared. Pressing the Classify button, I then set the Exclusion values to -1.0 to 0.09. This tells ArcMap to filter out all values in this range when classifying the data into clusters of similar values. I then classified by Natural Breaks (Jenks) with 3 classes to show a schema with less ability to separate unhealthy vegetation from the most vigorous. Finally, I created a copy of this NDVI layer, followed the same steps above but this time created a classification schema with 7 classes. For both of these schemas, I chose the same color ramp whereby unhealthy vegetation was in red; the healthiest vegetation in bright green; and vegetation with intermediate health in yellow.
Click the graphic above to see an animated comparison of a 7.5-minute USGS topo map and 10-meter NED DEM. The NED DEM has been colored so that white represents the highest elevations. (Image Source: USGS)
The results of this analysis show that there is more variability in plant health (even across a single field) than the human eye can detect. As more classes were added, the ability to determine areas that were healthier from those that were less healthy was enhanced. That said, there is certainly a limit to this fidelity that would need to be explored by comparing field measurements of plant productivity to Red Edge NDVI values. By doing this, you could determine what range of NDVI values (and at what time in the phonological development cycle) correspond to areas of less productive growth. Once identified, farmers and horticulturalists can target these locations with the necessary remedies – such as the addition of fertilizers and/or water – to assure the most abundant yield possible at the end of the growing season. What I have described here are the basic tenements of high precision agriculture which is an established method of farming in the US and many nations around the world. With RapidEye's ability to collect imagery over large areas on a daily basis at a low price point, its value in high precision agriculture cannot be understated. That said, to make RapidEye a cost-effective solution for this purpose, your fields would have to cover an area larger than many family farms typically do.
Brock Adam McCarty