Relationship between remote sensing-based biomass activity and tree-ring index for Picea abies in Saxony Michael
1 Körner ,
Jakob
NDVI map (MOD13Q1) 18-02-2000 and sampling sites in Saxony (Germany)
2 Wernicke
Background
Results
Our aim in this work is to empirically determine the relationship between remote sensing data estimating biomass activity and time series of tree-ring index for Norway spruce (Picea abies (L.) H.KARST.). A high correlation between these data types could be used to improve forest growth simulators and provide predictions for annual growth.
The results indicate no clear relationships between the NDVI metrics and tree-ring index information (Fig. 4). The maximum increasing NDVI value (MaxIV) shows a positive correlation for a major part of stands and MaxP indicate the highest positive (0.58) and negative (-0.69) correlations for single plots. Fig. 5 presents the distribution for the maximum NDVI value (MaxV). Here we find a similar behavior for short intervals like 2000-2004 and 2011-2013, but no similarity over the whole period.
Data and Method Data were collected by Wolfgang Beck in 2015 along three altitudinal transects in the German federal state of Saxony. In 17 Norway spruce stands two borer cores were extracted from 20 trees per stand at breast height. For each stand a tree-ring index time series (Fig. 1) was calculated by using the standardization method from Beck (2007) and a cubic smoothing spline. To ensure data independence in the time series we used autoregressive first-order modeling (AR(1)) and tested the results with a Breusch-Godfrey test.
Fig. 3: Derivation of phenological metrics from smoothed NDVI time series. MaxIP=Maximum Increasing Point; MaxDP=Maximum Decreasing Point; DurP=Duration of greeness; MaxP=Maximum Point; MaxDV=Maximum Decreasing NDVI Value; MaxIV=Maximum Increasing NDVI Value; MaxV=Maximum NDVI Value; TIN=Time-Integrated NDVI.
The launch of the satellite Terra (Fig. 2) limited the investigation to the period from 2000 to 2015. The quantification of the annual course for the NDVI was accomplished by phenological metrics according to Reed et al. (1994). To estimate daily information a cubic smoothing spline was fitted to the MOD13Q1 time series and with its first derivative single points were determined (Fig. 3).
Fig. 5: Distribution of maximum NDVI (MaxV) and tree-ring index (TRI) for the investigation period from 2000 to 2015
However, for single stands a positive correlation could be established. These findings back the initial hypothesis that remote-sensing NDVI data may be used as representatives of annual variation in tree and forest growth.
Fig. 1: Time series of tree-ring index (TRI) with significant first-order autocorrelation (left) and without (right)
Conclusion
Satellite data from the sensor MODIS were used to describe biomass activity:
No generalizable relationship between NDVI metrics and TRI over all stands
Freely available product MOD13Q1 with the Normalized difference vegetation index (NDVI)
MaxIV in general positively correlated to TRI Further studies should include ground vegetation into biomass productivity assessment
16-day composite – 23 time steps per year Spatial resolution: 250 m
Additional filters and smoothing techniques should be tested
Fig. 2: Satellite Terra with sensor MODIS (source: www.nasa.gov)
Fig. 4: Pearson correlation coefficients between tree-ring index and phenological metrics for 17 individual plots in Saxony
1Public
Cited references: Beck, W. 2007. Finding best regression approach for description of climate-growth relationships by floating time spans of varying width. In: Haneca, K. et al. (eds.): TRACE, Vol. 5: Proceedings of the Dendrosymposium 2006 in Tervuren, Belgium: 44-53. FZJ Zentralbibliothek, Jülich Reed, B. C.; Brown, J. F.; Vanderzee, D.; Loveland, T. R.; Merchant, J. W.; Ohlen, D. O. 1994. Measuring phenological variability from satellite imagery. Journal of vegetation science, 5(5), 703-714.
Enterprise Sachsenforst (Germany) (c/o
[email protected]) 2Forest Research and Competence Centre ThüringenForst – AöR (Germany) (c/o
[email protected])