Download - Dahl Winters December 6, 2006
Using a Geoprocessing Script to Aid in the Exploration of Linkages between Southern Pine
Beetle Outbreaks and Forest Density
Dahl Winters
December 6, 2006
Introduction
Southern pine beetle (Dendroctonus frontalis) is the most destructive insect to pine trees.
Vast numbers of individuals can become infested and die, with large ecological and economic consequences.
Images from http://www.na.fs.fed.us/spfo/pubs/fidls/so_pine_beetle/so_pine.htm andhttp://www.srs.fs.usda.gov/econ/data/spb/
Questions
How might SPB infestations correlate withthe density of various forest types?
Questions that might be answered by answering the above:
• Are SPB infestations more severe in pine plantations?
• How might smaller, more mixed stands of pine-hardwood trees contribute to outbreak severity?
Data and Outline of Approach
1. Land Cover Data: This data will be taken from the 2001 National Land Cover Dataset (NLCD 2001) – free download from the USGS Seamless Data Server.
2. SPB Infestation Data by County from 1960-2004
-Goal is to explore the environmental correlates (i.e. different quantities and patterns of forest cover) of SPB outbreaks.
-Will focus attention on one region in one year
-1) identify and isolate SPB habitat and non-habitat from the NLCD 2001 raster dataset
-2) Identify and isolate SPB infestation data for all counties in NC and SC in 2001
Methods
NLCD 2001 Land Cover
1960-2004 SPB Infestation
Data
Reclassified LULC data into 3 classes
Extract by Mask to get LULC for just NC and SC
Mosaic to join the two LULC rasters
Wrote Python code to extract LULC rasters
for each county
Copy pixel counts for the 3 classes into an Excel spreadsheet
Spatial Join to Infested Counties layer using
FIPS code to get desired graphs later
Select by Attributes to get just NC and SC
data for 2001
Normalization: use Editor to calculate
values for pixel counts (one pixel = 30 sq m)
Graph infestation severity vs. percent
cover types
Fit linear regressions to
graphs to quantify trends
Generate choropleth maps of each cover
type to show distribution
NLCD 2001 Reclassification
1. Definitely Habitat for SPB:
Evergreen Forest (42)
2. Potential Habitat for SPB:
Developed, Low Intensity (22) – SPB often attack stressed and injured trees in older-aged dense stands – some neighborhoods might have these
Deciduous Forest (41) – might be small isolated populations of pine trees here that serve as stepping stones for SPB dispersal
Mixed Forest (43) – likely to have more pine stands than deciduous forest
3. Non-Habitat for SPB:Open Water (11)Developed, Open Space (21)Developed, Medium Intensity (23)Developed, High Intensity (24)Barren Land (Rock/Sand/Clay) (31)Shrub/Scrub (52) – since SPB attacks
only mature trees, and the trees in this category would be young, early successional ones
Grassland/Herbaceous (71)Pasture/Hay (81)Cultivated Crops (82)Woody Wetlands (90)Emergent Herbaceous Wetlands (95)
Land Cover and Counties
Python Geoprocessing Script
Virtual memory problems…
Maps and Graphs
Calculated the following 4 percentages:
• percent pine forest = pine pixels/total pixels * 100• percent potential habitat = potential pixels/ total pixels * 100• percent forested = pine+potential habitat/total pixels * 100 • percent non-habitat = non-habitat pixels/total pixels * 100 (just the
inverse of percent forested)
Generated several maps using each of these percentages to visually estimate if there might be any correlations between infestation severity and the amount of different habitat types.
Percent Pine Forest
Percent Potential Habitat
Percent Forested (Pine + Potential)
Results - Percent Forest Coverage
Percent Pine Forest vs. Infestation Severity y = -0.2822x + 8.6383
R2 = 0.0418
0
10
20
30
40
50
0 10 20 30 40 50 60
Percent Pine Forest
Infe
stat
ion
Sev
erit
y (s
po
ts p
er t
ho
usa
nd
ac
res)
Percent Potential Habitat vs. Infestation Severity y = 0.2433x - 3.5999
R2 = 0.1616
0
10
20
30
40
50
0 20 40 60 80 100
Percent Potential Habitat
Infe
stat
ion
Sev
erit
y (s
po
ts p
er t
ho
usa
nd
ac
res)
The values on the X-axis have been normalized to represent (area of forest type / total forest area) * 100.
• As the percent of pine forest increases, interestingly the infestation severity decreases. This is opposite of what was expected.
• However, as the percent of smaller and more mixed pine-hardwood stands increases, infestation severity increases – 0.24 more infestations per 1000 acres for every % increase in potential habitat.
Results - Percent Forest Coverage
Percent Forest vs. Infestation Severity y = 0.2619x - 8.5805
R2 = 0.1353
0
10
20
30
40
50
0 20 40 60 80 100
Percent Forest
Infe
stat
ion
Sev
erit
y (s
po
ts p
er t
ho
usa
nd
ac
res)
• As the percent of both pine forest and smaller, mixed pine-hardwood forest increases, infestation severity increases more strongly from 0.24 to 0.26 infestations/1000 acres/% increase of forest area.
• Another way to look at this: the more non-habitat, the lower the infestation severity, which is expected since there are fewer places for SPB to exist.
Conclusion
• Prevention of SPB outbreaks should not focus on just pine plantations.
• Smaller, more mixed stands of pine in both natural and residential areas are important to SPB outbreaks—there are more of them and are more widely spread
• The connectivity of these smaller mixed stands is likely more important than we think it is, and merits further research into how it might affect SPB spread
• Further research directions: use more of the 1960-2004 infestation data to see how land cover correlates with SPB outbreaks over time