Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
713 
The delineation and classification of inland wetlands 
utilizing fcir stereo imagery 
Stephen A.Estrin 
Mahopac, New York, USA 
ABSTRACT: The fastest and one of the most accurate methods of classifying and delineating 
inland wetlands and their adjacent upland is through the interpretation of medium-to-large- 
scale (1:12,000 and larger) fcir stereo imagery. This is accomplished by manual photo inter 
pretation of those readily recognizable properties of inland wetlands and uplands; vegetation 
types, presence of standing water, soil water content and topography-elevation and slope. Ve 
getation is one of the best indicators of water guantity, quality and permanence. Specific 
ally, in wetlands, vegetation is one of the principal factors causing differences in spectral 
reflectance; therefore, noticeable differences in vegetation, vis-a-vis their spectral ref 
lectance, depicted on fcir imagery are directly related to the presence of water, both stand 
ing and soil water content. Consequently, the delineation of the boundary between a wetland 
and its adjacent uplands can clearly be ascertained based upon these differences in spectral 
reflectance. Classification and delineation of the wetland-upland complex was found most ac 
curately determined in the northeastern and central United States from spring fcir stereo im 
agery flown at an altitude of 6,000 feet AMT (1:12,000). In comparison, the modern methods 
of field ecology survey were found to be too time consuming, difficult and costly. In parti 
cularly fragile ecosystems, comprehensive biological field studies tended to cause consider 
able physical environmental damage, while the necessary ground truth program covering only a 
small portion of the area had no adverse impact. Naturally, the subsequent imagery analysis 
causes no adverse environmental impact. 
1 INTRODUCTION 
Wetlands are an important element of the nat 
ural environment and the increased public 
concern with environmental issues has lead 
to the enactment of federal, state and local 
legislation protecting inland wetlands. 
Due to the inadequacy and inaccuracies of 
the majority of existing maps and wetland 
inventories as well as their heterogenity, 
areal extent, time and money constraints, 
many governmental agencies at the federal 
and state levels have utilized remote sens 
ing as the primary method of delineating and 
classifying wetlands. Remote sensing affords 
both a practical and economic means for their 
accurate delineation and classification. It 
differs from conventional wetland data col 
lection in that the recording methodology is 
not in direct contact with the ground. It is 
airborne. In the majority of instances, the 
choice of a remote sensing technique is a 
function of a series of interrelated fac 
tors; size of the project area, accessibil 
ity, time of year, cost, data to be obtain 
ed, accuracy, and level of detail required. 
In 1977, one of the most extensive studies 
of inland wetlands was undertaken by Carter, 
Garrett, Shima & Gannon of the U.S.G.S., The 
Great Dismal Swamp located in Virginia and 
North Carolina. These scientists/photogram- 
metrists believed that the use of conven 
tional wetland data collection methods would 
be too expensive, time consuming, and too 
difficult to interpret for such a large, di 
verse and inaccessible area. They, there 
fore, decided on the use of seasonal low- 
and high-altitude color infrared photography 
utilizing manual interpretation techniques. 
The wetland maps were prepared at a scale 
of 1:24,000. 
These analysts found that imagery obtained 
in the spring, during dormancy, allowed the 
identification of wetland boundaries, areas 
covered by water, the drainage pattern, the 
location of coniferous vegetation and its 
classification, and the classification of 
the understory vegetation. Photographs ob 
tained during the summer were used to clas 
sify deciduous vegetation. 
1.1 A real example - Longridge Corporate 
Park, New York 
In the case of the Longridge Corporate Park, 
the determination of type of remote sensing 
was never a factor because the site had been 
disturbed to such an extent from October, 
1982, that only a vague and inaccurate re 
construction of previous conditions would be 
possible. Therefore, historical remote sens 
ing data, predating October, 1982, was the 
only practical and accurate means available 
for the delineation and classification of the 
wetland. 
In March, 1973, and again in April, 1974, 
Stephen A. Estrin, Inc., as part of its con 
tract with the County of Putnam for the de 
sign of a Comprehensive Land Development 
Plan and Sewer Study for the Town of South 
east, had its Division of Photogrammetry 
contract with Grumman Ecosystems for an aer 
ial photomapping mission, to its specifica 
tions, which produced stereo color carto 
graphic mapping photography and false color 
infrared imagery at the scales of 1"=2000' 
and 1"=1000'. 
It is the April 20, 1974 fcir, 1"-1000', 
imagery that forms the basis of the delin 
eation and classification of the Longridge 
Corporate Park Wetland. The interpretation 
of this imagery was by manual means utiliz 
ing imagery interpretation and transfer e- 
quipment, ground truth and a wetland image 
analysis key developed by Stephen A. Estrin, 
Inc. in 1972, updated in 1976. In addition,