For this set of tests, we added three additional variables to the short list of six variables used in http://research.esd.ornl.gov/~hnw/neon/tests2. We added solar input in the non-growing season, the "mate" to one already in the list, and we also added the number of days with measurable precipitation in both seasons. The latter layers add information about the distribution of precipitation across each season. The list of nine abiotic variables used for all of these maps is shown in the table below:

NEON2 - Nine selected abiotic variables - Potential Vegetation
10 Zones Small Map Medium Map Large Map Huge Map	15 Zones Small Map Medium Map Large Map Huge Map
20 Zones Small Map Medium Map Large Map Huge Map	25 Zones Small Map Medium Map Large Map Huge Map

Two additional maps were prepared at much higher levels of division, 100 and 200 zones. The purpose of these two extra maps, which used the same nine input layers, is to provide context for comparison, and to demonstrate the discrimination and resolving power of the Multivariate Geographic Clustering method when provided with this input data set.

100 Zones Small Map Medium Map Large Map Huge Map

200 Zones Small Map Medium Map Large Map Huge Map

Comments/Observations

• All of these represent new analyses

• These zonation maps look pretty good, and I believe that we are getting close to usable products for NEON design.

• These zones are fairly cohesive up to the spatial scale of multiple counties, but there are many examples where zones are spatially disjoint because they are broken into two or more subregions. For example:
• Montana and North Dakota are often in the same zone as southern Maine and the northeast
• Northern Idaho is often in the same zone as Michigan, Ohio, and the northern northeast
• The Sonoran desert is often in the same zone as southern Texas
• The Rocky Mountains are often spatially disjoint

• The 10-zone map still seems inadequate to capture the spatial heterogeneity present in the lower 48 United States. For example, the Rocky Mountains are not resolved into a separate zone at this coarse level of division.




• We have not included any significant information about soil depth or fertility, i.e., organic matter or nitrogen content of soil. All we have included is plant-available water, which is a soil texture property.




• We also have included no information on actual existing vegetation. So this result is about POTENTIAL VEGETATION ONLY. There can be substantial differences between existing vegetation and potential vegetation at many sites. We plan to include this information during Siting Steps 2 and 3.

Similarity Colors for NEON2

All of the zones in the above maps are shown in random colors. When colored randomly, the edges of distinct zones are easy to see. However, no information is available to show how different the mixture of included conditions is between two adjacent zones. We also have a statistical method to assign colors to each individual zone in each map such that the color shows something about the mixture of included conditions within each zone. Using this coloring method, the similarity of conditions can be compared across zones.

To help gauge the adequacy of particular levels of division, we assigned colors to zones based on the top three Principal Components. Taken together, these three PCs explained 87% of the variance in this data set.

The first Principal Component is strongly loaded with negative cold days in the non-growing season, precip in the growing season, wet days in the growing season, and solar insolation in the growing season, and negative solar insolation in the non-growing season. Hot days during the growing season also loads positively on this Factor, but loads even more strongly with an opposite sign on Factor 2. We have assigned Factor 1 to red, and we interpret it as "wetness in the growing season, and warmth/solar."

The second Principal Component is strongly loaded with negative hot days during the growing season, precipitation during the non-growing season, wet days during the non-growing season. Solar insolation during the growing season also loads negatively on this Factor, but loads even more strongly with an opposite sign on Factor 1. We have assigned Factor 2 to blue, and we interpret it as "wetness in the non-growing season, and cold."

The third Principal Component is strongly loaded with plant-available soil water, which is a measure of soil texture. We have assigned Factor 3 to green, and we interpret it as "plant-available soil water."

NEON2 Rotated Factor Loadings

All zonation maps are similar at the national scale when viewed in Similarity Colors, but more resolution and detail is present in the maps produced at higher levels of division. If additional maps were produced at even higher levels of division (i.e., 10,000 zones), such maps would be indistinguishable from the 200-zone map when colored using Similarity Colors. Thus, there is a convergence on this single map appearance at the national scale after a particular critical level of division is exceeded. No further subdivision is needed to resolve and reveal this same national pattern of gradients.

NEON2 - Nine selected abiotic variables - Similarity Colors
10 Zones Small Map Medium Map Large Map Huge Map	15 Zones Small Map Medium Map Large Map Huge Map
20 Zones Small Map Medium Map Large Map Huge Map	25 Zones Small Map Medium Map Large Map Huge Map

100 Zones Small Map Medium Map Large Map Huge Map

200 Zones Small Map Medium Map Large Map Huge Map

For additional information contact:

William W. Hargrove

Oak Ridge National Laboratory

Environmental Sciences Division

Building 1507, Room 211

Mail Stop 6407

Oak Ridge, TN 37831-6407

(865) 241-2748

(865) 574-4665 fax

hnw@fire.esd.ornl.gov