regional forest health and stream and soil chemistry using
TRANSCRIPT
Hyp-July
ALI-July
Hyp-Sept
ALI-Sept
Landsat-July
Non-forest Mask
Hemlock
<VALUE>
<20%
20-30%
31-40%
41-50%
51-60%
61-70%
71-80%
>80%
Basal Area
Regional Forest Health and Stream and Soil Chemistry Using a Multi-Scale Approach and
New Methods of Remote Sensing Interpretation, Catskill Mountains, NY
R.A. Hallett, USDA Forest Service, Durham, NHP.S. Murdoch, U.S. Geological Survey, Troy, NYJ. Pontius, USDA Forest Service, Durham, NHM.E. Martin, University of New Hampshire, Durham, NHL. Plourde, University of New Hampshire, Durham, NH
Introduction
The New York-New England region is the most densely forested region in the U.S. It is also one of the most densely populated, and experiences air pollution levels among the highest in the country.
Over 30 million people in the region seek potable water, forest products, recreation and aesthetic renewal from a complex patchwork of forested lands with a rich and diverse history of human use
and current ownership. The health of these forest lands and the waterways they feed are a critical component of the economic well-being and quality of life in the region.
Forest health and stream water quality both depend on the integrity of biogeochemical cycles within forest ecosystems. It has been clearly established that both atmospheric deposition and forest
management can alter the biogeochemistry of forest ecosystems, inducing significant changes in forest production, species composition and stream water quality. Nutrient-cation imbalances have
been associated with soil and stream acidification, nutrient imbalances in forest vegetation, and in some cases with decreased forest production.
Monitoring the biogeochemical status of forest and stream ecosystems is a key component of assessing environmental quality in the northeastern U.S. Any monitoring system requiring spatially-
continuous capabilities will need to utilize some form of remote sensing technology. Forest canopies are the only portion of the system accessible to optical remote sensing instruments and so
offer the most likely target surface for monitoring forest health in this spatial mode. The usefulness of remote sensing of canopy chemistry depends on tight relationships between canopy chemistry
and the critical processes of forest production and element losses in drainage water.
Image Registration
Atmospheric Correction
View Angle Correction
Topographic Correction
Interpolated
Soil and Water
Chemistry Maps
Plant Available Nutrient Status
Soil ChemistryForest Health
Chlorophyll Fluorescence
Transparency
Fine Twig Dieback
Live Crown Ratio
Defoliation Class
Photosynthetic Performance
Index
Foliar Calcium and Nitrogen
Foliar Chemistry
Surface Water
Calcium and Nitrogen
Water Chemistry
^
^
^
^
^
^
^
Woodstock
Lexington
Phoenicia
Claryville
Fleishmanns
West Shokan
Tannersville
High Ca
Medium Ca
Low Ca
Interpolated Stream Water Ca Map
Pepacton
Esopus
Schoharie
Neversink
Rondout
Interpolated Stream and Soil Chemistry
Maps
Regional stream and soil sampling efforts
conducted by the USGS and USFS – FIA
program have resulted in the ability to create
interpolated maps for a wide range of soil and
stream water chemical characteristics. These
data layers will be used in conjunction with
the remote sensing data layers to create an
integrated spatially explicit model that utilizes
a wide variety of ecologically sensitive
indicators to create maps of areas that are
sensitive to acid deposition
Satellite Imagery
popo
po
po
po
po
popo
po
po
po po popopo
po
po
po
popo
po
popopopopopopopo
popopo po
popopopopopopo
popopo
popo
po
po
popopo popo
popo
popo popo po popo
po popo po
po
popopo po popo
po popo
po
po
po
popo
po
po
po
po
po
popo
po
popo
po
po
popopo
po po
po
po
popo
po
po
po
po
po
po
po
po
popo
po
popo
po
po
popo
po
popo
popopo popo
po
po
popo
popo
popopo
po
po
po
popo popo
popo
po
popo po
popopo
popo
popo
po
po
popo
po
po
po
po
popo
popopo
popo
po
po
po
po
popopopo
po
po
po
popo
po
po
po
po
popo
popo
po
po
po
po popo
popo
po
po
popo po
po
po
popo
po
po
popo
po
po
po
po
po
popopopo
popopopo
popopopo po
po popopo
popopopopo
po
po
po
po
po
popo
popo
popo
po
po po
popopo
po
popopo
po
po
popo
popo
popo
popo
po popopo
po
popo
po popo
popo
popo po
po
popo po
popopo
popo
popo
popo
po
popo
popo
po
po
po popo po
popo
po
p
popopo
popo
po
popo
po
popopo
po
po
po
popo
po
po po popo
popo
popopo
po
popopopopo
popopo
po
popo
po
popopo
popopo
popo
po
po popo
po
po
po
po
po
po
po
po
po
po
po
po
"T"T"T"T
"T"T"T"T
"T
"T
"T"T"T
"T "T"T
"T "T"T"T
"T"T"T"T"T
"T
"T
"T
"T
"T
"T"T
"T"T
"T"T
"T
"T "T
"T"T"T
"T
"T"T"T
"T
"T
"T"T
"T"T
"T"T
"T"T
"T "T"T"T
"T
"T"T
"T "T"T
"T"T
"T"T "T
"T
"T"T "T
"T"T"T
"T"T
"T"T
"T"T
"T
"T"T
"T"T
"T
"T
"T "T"T "T
"T"T
"T "TT"T
"T"T
"T
"T"T
"T
T"T"T
"T
"T
"T
"T"T
"T
"T "T "T"T
"T"T
"T"T"T
"T
"T"T"T"T"T
"T"T"T
"T
"T"T
"T
"T"T"T
"T"T"T
"T"T
"T
"T "T"T
"T
"T
"T
"T
"T
"T
"T
"T
"T
"T
"T
"T
o
o ooo
ooo
ooo
oo
oo
oo
ooo
oo
o
o
o
oo
ooooo
o
o
o
oo
oo
oo
o
oo
o
^
^
^
^
^
^
^
Woodstock
Lexington
Phoenicia
Claryville
Fleishmanns
West Shokan
Tannersville
o Health Plots "T Soil Pitspo Stream Samples
Study area
687,000 acres = 278,000 hectares = 1,100 square miles
Calibration
Validation
Signature Analysis
MPLS Full Spectrum Regression
Simple Linear Regression
Foliar Chemistry
Predictive Equation
Forest Health
Predictive Equation
34567
Low Stress
High Stress
A three term linear
regression using a
unique combination of
Landsat TM bands was
able to predict a 10 class
decline rating with 57%
accuracy and to within 1
class with 98% accuracy.
The resulting coverage of
forest health was limited
to the range in plot level
health observed in field
measurements. While
this coverage is not
stress specific, it is likely
that these upper
elevational “hot spots” coincide with areas of
extreme forest tent
caterpillar defoliation in
2006.
2006 Landsat Predicted Decline
Actual
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Pre
dic
ted
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0N = 46Terms = 3R-square = 0.44RMSE = 0.54Avg Jacknifed Residual: 0.55Class Accuracy = 57%Within 1 Class Accuracy = 100%
Decline = 2.3 + (0.003 * B6) + (26.5 * B3/B4) - (1.8 * B5*B3 / B42 )
51-60%
61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%61-70%70%061-700%61-7070%%61-70%61-70%61-70%61-70%61-70%61-70%6161-70%661-70%61-70-70-70-70%7061-70%-70%00%70%70%1-70%61 70%61-70%61-70%61-70%61 70%1 70%61-70%61-70%661-7061-7061-70%70%0-701-70%0-70%%61-70%661-70%61-70%-70-70%1-70%%61-70%61-70%1 70%-7061-70%70%61-70%1-7061-70%7-70%0%%61-70-7070%61 70-70%-70%61 71-70%70%0%61 70-70%-70%-70%0%-700770%0%-70%%0%0%%7070%7000%70%77070%7 %
71-80%71-80%71-80%71-80%71-80%71-80%1-8071-80%71-80%71-80%71-80%71-80%71-8071-80%71-80%71-80%71-8071-80%81-8071-80%71-880%80%71-80%871-80%71-8071-80%871-80%71-80%71-80%81-80%71-80%%71 80%71-80%71-80%71-80%71 80%71 80%71-80%1 80%71-80%71-80%88871 80%-80%71-80%%%71-80%71-80%71-80%71-80%71-80%71-80%71-80%71-80%1 80%71-80%71-80%71-80%71-80%71-80%%%71-80%71-80%71-80%71-80%71-80%71-80%71-80%71-80%%%71-80%71-81 80%8-80%71-80%0080%71-80%80%%%71-80%71-80%1-871-80%%0%0%071-80%71 80%71-80%71-80%71 80%71 80%71-80%%%%71-80%71-80%71-80%71-80%7 8 %%71-80%80%71 80%7771 80%71 80%71 80%777 88871 80%77 %80%888Hemlock
<VALUE>
<20%
20-30%
31-40%
41-50%
51-60%
61-70%
71-80%
>80%
Basal Area
Basal Area
Landscape-scale coverages
shown in this poster are
currently being incorporated
into a spatially-explicit index of
forest sensitivity to biotic and
abiotic disturbance.
These data products will have
immediate utility for developing
forest harvest strategies in the
watersheds feeding the NYC
water supply.
Data Synthesis
Species Mapping
Final Processed
Image Reflectance
Minimum Noise Transform
Pixel Purity Index
Mixture Tuned Match Filtering
subbasin_bdry_
Sugar Maple
<VALUE>
<20%
20-30%
31-40%
41-50%
51-60%
61-70%
>70
Aviris_2001_st
Basal Area
Red oak, sugar maple,
and hemlock maps
above represent
preliminary results of
spectral mixture
analysis of the
hyperspectral data
sets—a somewhat
novel approach to
species mapping.
Rather than detecting
discrete classes
spectral mixture
analysis allows for
detection of sub-pixel
species composition.
These products will be
augmented with FIA
plot data to improve
fractional species
composition maps for
the Catskills region.
Foliar
Nitrogen
Low
High
Medium
Foliar nitrogen maps are
under development from
a number of data
sources. In 2001, the
entire region was imaged
by NASA's Airborne
Visible/InfraRed Imaging
Spectrometer (AVIRIS;
nitrogen image to the
left). In 2006-7, data
were acquired from
Hyperion (hyperspectral),
the Advanced Land
Imager (ALI; broadband)
and Landsat. 2006-7 field
data is being used with
these recent data to
update estimates of foliar
nitrogen for selected
areas.