Full size [1Mb] US Soils Taxonomic Order Map After seeing my poster that described using dot maps to show soils at the AAG conference in Boston a few weeks ago, Charlie Frye suggested that I write this entry to describe my maps and the technique used to create them.  This seemed especially appropriate given the recent entries in this blog about dot maps.

The data source for the soil dot maps is the U.S. General Soil Map, previously known as STATSGO, which is available from the Natural Resources Conservation Service (http://soils.usda.gov/survey/geography/statsgo/).  The STATSGO database is a soil association map.  It was developed to depict the distribution of soil properties at small scales, and originally compiled at a scale of 1:250,000.  In a soil association map, generalization from larger scales (in this case, the county-level SSURGO data) is accomplished by grouping into a single mapping unit soil types that normally occur together on the landscape.  The digital STATSGO database retains links to the individual soil types and their properties by including information about the percentage (by area) of each component soil type represented in a mapping unit.  Most mapping units in the STATSGO database include between 10 and 20 different individual soil types.

The advantage of using the dot map over a choropleth mapping technique for this database is that it gives the map reader a sense of the variability of the soil types in each polygon. With a choropleth technique, usually the dominant condition that exists in the polygon is used to determine the pattern chosen for the entire polygon. For example, if a particular polygon were composed of 35% Alfisols, 30% Mollisols, 20% Inceptisols, and 15% Entisols, then the polygon would be assigned the symbol for Alfisols, even though the Alfisols represent less than half of the area in the polygon. With the dot map technique, there would be dots for all four soil orders randomly distributed inside the polygon: if the polygon’s area was 100 hectares, and each dot represented one hectare, there would be 35 Alfisol dots, 30 Mollisol dots, 20 Inceptisol dots, and 15 Entisol dots.

The tabular database included with STATSGO includes data about the individual component soils concerning many different soil properties. Some of the properties in the database pertain to the entire soil profile, and some to individual soil horizons. The distribution of any of these soil properties could potentially be mapped; for this map, I have chosen to use the soil order classification according to the Soil Taxonomy system.

To make the dot map, within each mapping unit I summed the areal percentage represented by each soil order, and joined the resulting table to the attribute table of the spatial data. At this point the attribute table contained a field for each soil order showing the percentage of each polygon represented by that order. Next, I calculated the area of each polygon in hectares, then multiplied the polygon area by the percent in each order. I now knew the number of hectares of each order contained in each polygon. This is the number that is used in creating the dot map.

I then made a dot density map in ArcMap. Multiple fields can be included in the dot density map; for the soil orders map I included the fields for all ten soil orders that exist in the conterminous United States. Each order was assigned a different dot color. Because the polygons were small relative to the size the of the map (there are more than 77,000 polygons in the conterminous United States), and in order to reduce the amount of clutter on the map, I chose not to draw the polygon boundaries on the map.

 

I also created smaller maps for each soil order which used a similar method to depict a single order at a time.  Here are links to PDFs for each of these:

Including these individual maps alongside the main data frame on the poster let the reader bounce back and forth between the main map and the individual soil type maps to better understand the nature of each soils type's distribution in the U.S.  Unfortunately, the main data frame was too large to link to here.