| | / home / Publications / Mattole Restoration Newsletter / Issue 19 - Winter/Spring 2002-03 /
How these Maps were Created
by Seth Zuckerman November 20, 2002
How these maps were created
The portrayal of Mattole grasslands in the 1950s is derived from the Soil and Vegetation map series published during the 1950s by the U.S. Dept of Agriculture in cooperation with the state and counties. On those maps — based on aerial photographs taken in 1947, 1948, and 1954 — the landscape is divided into tracts in which the types of vegetation are listed in order of predominance. For purposes of the current publication, we counted as “grassland” those areas where grasses were the sole or the predominant type of vegetation. We also included land marked as “cultivated” (mostly hayfields), but excluded land that had been planted in orchard. The 1950s-era maps divide the landscape into units of 10 acres or more.
Stephen Umbertis scanned these maps digitally and transferred them to the Mattole Restoration Council’s Geographic Information System, a computerized mapping and data program. He traced the outline of grass-dominated tracts by hand with the computer pointer, creating a map layer showing where prairies were located. Americorps volunteer Reid Bryson added to the map those tracts that had been mapped in the 1950s as “cultivated land,” except for land that appeared to be in orchard according to 1942 aerial photos, the set available to us that was closest in time to the dates of the photos on which the 1950s vegetation maps were based.
The 1998 map of Mattole grasslands is based on satellite imagery recorded in April 1998. Bryson used Arcview’s Image Analyst software to classify the land cover in the image into more than 200 categories. He then used 1998 aerial photography to determine that several dozen of those categories corresponded with grasslands. Bryson and Seth Zuckerman checked the accuracy of these determinations on the ground, finding a better-than-90 percent accurate correspondence between the map and the terrain, on a few dozen points on two sites in the lower Mattole. MRC GIS director Jeremy Wheeler then revised the map to aggregate areas of grass and non-grass into blocks of at least 10 acres, to correspond with the minimum mapping area used in the 1950s maps. Zuckerman then compared the map against contemporary aerial photos to catch areas that had been inaccurately classified by the automated interpretation of satellite imagery. For some areas, 1998-vintage photographs were available. Elsewhere, 1993 photos were used in tandem with photos taken in 2000.
Limitations: The resolution of the satellite imagery is 30 by 30 meters, or about 100 by 100 feet. As a result, edges where prairies meet brush or forest are difficult to categorize, and strips along those edges are often classified in error.
Some prairies that may be familiar to local residents will be absent from the map because they are smaller than 10 acres.
The maps presented today are a work in progress. The MRC welcomes corrections and suggestions from people with a personal knowledge of the landscape.
Printer Friendly
Version
Read more articles about Plants
Table of Contents for Mattole Restoration Newsletter, Issue 19 - Winter/Spring 2002-03 | |