Evaluation of IRS-1c LISS-3 satellite data for defoliation assessment on Norway spruce and Scots pine

Abstract

This study evaluates the performance of IRS-1c Linear Imaging Self-Scanning Sensor 3 (LISS-3) data for assessment of slight and moderate defoliation, induced by the different stress factors affecting the boreal forests of Västernorrland, Sweden, in stands dominated by Norway spruce and Scots pine. It shows that accuracies as good as with Landsat TM can be produced with IRS-1c LISS-3 data, and that defoliation in pine could be assessed with only a slightly lower accuracy than in spruce. As in previous studies of spruce defoliation, the near-infrared (NIR) band was best correlated with spruce defoliation. Among the single bands, the NIR band was also best correlated with pine defoliation, although the relationship was even stronger with the ratios between NIR and red, and NDVI. Compared to spruce, assessment of pine defoliation appeared to be more sensitive to the presence of species other than pine. The strength of the relationship between LISS-3 data and defoliation increased significantly (from r=−.73 to r=−.91 for NIR/red, note however the reduced sample size), when only plots with more than 80% pine, rather than 70%, were included in the analysis. An attempt to normalize satellite data for species composition did not raise correlation coefficients for spruce data, and only slightly for pine data. Topographic normalization on the other hand, strongly increased the correlation coefficients between NIR radiance and spruce defoliation. A large part of the pine plots were located on near-horizontal ground; thus, topographical normalization only slightly increased correlation coefficients. This study verifies that high-resolution satellite data provide a means to assess the overall needle loss situation on a forest stand level provided that: the stands are fairly homogenous, not too open, dominated by spruce (>60%) or pine (>70%), located in level or moderately undulating terrain (<15° slope), and express a needle loss range of at least 20%. The utility of IRS LISS data in addition to earlier evaluated Landsat TM data strongly improves the possibility to perform operational satellite-based monitoring in areas with frequent cloud cover.