Advanced

Mapping and early-warning of insect defoliation in Scots pine with multi-temporal MODIS data

Thomas, Johansson and Eklundh, Lars LU (2009) In Lund electronic reports in physical geography 6.
Abstract
Methods were developed for post-detection and early-warning of defoliation in Scots pine [Pinus silvestris] forests in south-eastern Norway caused by the pine sawfly [Neodiprion sertifer] with the use of multi-temporal MODIS NDVI 16-day composite data. The post-detection method utilizes summer mean values and seasonal angle (showing whether values have increased or decreased during the season) to identify changed pixels. Damage detection was done by comparing 2005 summer mean and seasonal angle to normal values based on the years 2000 to 2002. In addition to 16-day NDVI the new index Wide Dynamic Range Vegetation Index (WDRVI) was tested. Classification results were evaluated with laser scanned LAI data. The damage classifications with... (More)
Methods were developed for post-detection and early-warning of defoliation in Scots pine [Pinus silvestris] forests in south-eastern Norway caused by the pine sawfly [Neodiprion sertifer] with the use of multi-temporal MODIS NDVI 16-day composite data. The post-detection method utilizes summer mean values and seasonal angle (showing whether values have increased or decreased during the season) to identify changed pixels. Damage detection was done by comparing 2005 summer mean and seasonal angle to normal values based on the years 2000 to 2002. In addition to 16-day NDVI the new index Wide Dynamic Range Vegetation Index (WDRVI) was tested. Classification results were evaluated with laser scanned LAI data. The damage classifications with 16-day NDVI had kappa coefficients between 0.48 and 0.63, and detected 71% to 82% of the damaged pixels. Although damage classification with WDRVI gave similar results, NDVI was retained for reasons of comparison with other work, and because the behaviour of WDRVI in forest is not yet well known. The developed early-warning method uses calculated differences in NDVI and a seasonal angle between the damage year and a normal year for every 16-day MODIS scene during the growing season. Calculated differences in NDVI and seasonal angle were tested with the Wilcoxon signed rank test for significant changes, and combined into seasonal damage maps. The seasonal damage maps display a consistent pattern through time, indicating the core damage areas with a fair accuracy, when comparing with evaluation data generated by laser scanning. In conclusion, time series of MODIS NDVI can be used for detecting defoliation due to pine sawfly in Norwegian forests, and for early-warning. The damage areas can be coarsely located with fair accuracy. Control of detected damage areas using high resolution remote sensing data or fieldwork is recommended for accurate delineation of the damages. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Book/Report
publication status
published
subject
in
Lund electronic reports in physical geography
volume
6
pages
23 pages
publisher
Department of Physical Geography and Ecosystem Science, Lund University
ISSN
1402-9006
ISBN
978-91-85793-11-2
language
English
LU publication?
yes
id
cc2c4b67-08c7-4b01-b5ec-3b9479e44444 (old id 4173816)
date added to LUP
2013-11-21 14:59:47
date last changed
2016-04-15 21:06:43
@techreport{cc2c4b67-08c7-4b01-b5ec-3b9479e44444,
  abstract     = {Methods were developed for post-detection and early-warning of defoliation in Scots pine [Pinus silvestris] forests in south-eastern Norway caused by the pine sawfly [Neodiprion sertifer] with the use of multi-temporal MODIS NDVI 16-day composite data. The post-detection method utilizes summer mean values and seasonal angle (showing whether values have increased or decreased during the season) to identify changed pixels. Damage detection was done by comparing 2005 summer mean and seasonal angle to normal values based on the years 2000 to 2002. In addition to 16-day NDVI the new index Wide Dynamic Range Vegetation Index (WDRVI) was tested. Classification results were evaluated with laser scanned LAI data. The damage classifications with 16-day NDVI had kappa coefficients between 0.48 and 0.63, and detected 71% to 82% of the damaged pixels. Although damage classification with WDRVI gave similar results, NDVI was retained for reasons of comparison with other work, and because the behaviour of WDRVI in forest is not yet well known. The developed early-warning method uses calculated differences in NDVI and a seasonal angle between the damage year and a normal year for every 16-day MODIS scene during the growing season. Calculated differences in NDVI and seasonal angle were tested with the Wilcoxon signed rank test for significant changes, and combined into seasonal damage maps. The seasonal damage maps display a consistent pattern through time, indicating the core damage areas with a fair accuracy, when comparing with evaluation data generated by laser scanning. In conclusion, time series of MODIS NDVI can be used for detecting defoliation due to pine sawfly in Norwegian forests, and for early-warning. The damage areas can be coarsely located with fair accuracy. Control of detected damage areas using high resolution remote sensing data or fieldwork is recommended for accurate delineation of the damages.},
  author       = {Thomas, Johansson and Eklundh, Lars},
  institution  = {Department of Physical Geography and Ecosystem Science, Lund University},
  isbn         = {978-91-85793-11-2},
  issn         = {1402-9006},
  language     = {eng},
  pages        = {23},
  series       = {Lund electronic reports in physical geography},
  title        = {Mapping and early-warning of insect defoliation in Scots pine with multi-temporal MODIS data},
  volume       = {6},
  year         = {2009},
}