Method analysis for collecting and processing hyperspectral needle reflectance data for monitoring spruce vitality

Abstract

Bark beetle infestations cause severe damage to forestry. In the context of climate change, inducing substantially warmer and possibly drier summers, this threat is expected to increase. Remote sensing has the capacity to detect large scale changes based on quantitative observations of spectral changes. State of the art very high spatial resolution satellite data offer a great potential for assessing tree vitality. However, a better understanding of the correlation between vitality decrease and its impact on the spectral behaviour of needles is needed. For improving the interpretation of remote sensing data detailed analyses of needles are therefore necessary. Despite, it is still unclear which method for needle spectra collection is most suitable. The scope of this study is to investigate two methods for spectral reflectance measurements of Norway spruce needles (Picea abies, L.). The spectra were retrieved by using portable spectroradiometers applying both a classical fore optics and a (so-called) contact probe. Ne edle samples were acquired in summer and fall 2013 in Bavaria (Germany) from two forest stands. Tree limbers collected the branches from the top of the crowns. For our study, needles of the four last age groups, from 2010 to 2013, were measured. The measurements were performed in-situ as well as in a laboratory and repeated three times during the growing season. We discuss methods for sample storage, preparation and the influence of sample freshness on the reflectance behaviour. Furthermore, we evaluate the spectral data collection in-situ and in a laboratory environment with different measurement settings and in terms of its practicality. After preprocessing (removal of artificial spikes in the data, outlier detection and smoothing) the data were analyzed statistically using different similarity measure techniques. This research serves as a preliminary study for field campaigns scheduled in 2014 to monitor tree vitality. Our results will be implemented to improve monitoring of tree damage with very high resolution WorldView-2 data as well as airborne HySpex data.