Influence of woody tissue and leaf clumping on vertically resolved leaf area index and angular gap probability estimates

Leaf area index L is a key vegetation parameter that can be used in soil–vegetation–atmosphere exchange modeling. To represent the structure of ecosystems in vertically distributed modeling, vertically resolved L distributions as well as vertical and angular gap probability Pgap distributions are needed, but they are rarely available. On the experimental side, studies often neglect woody plant components when using indirect methods for L or Pgap observations. This can lead to significantly biased results, particularly in semi-arid savannah-type ecosystems with low L values.The objective of this study is to compare three non-destructive leaf area index measurement techniques in a sparse savannah-type cork oak canopy in central Portugal in order to derive vertically resolved L as well as vertically and angularly resolved Pgap. We used the established LAI-2000 device as well as fast digital cover photography (DCP), which was vertically and angularly distributed. We applied object-based image analysis to DCP to exclude woody plant components. We compared the results with vertically distributed LAI-2000 measurements and with vertical estimates based on easily measurable crown parameters.Height and angularly distributed DCP was successfully applied here for the first time. It delivers gap probability Pgap and effective leaf area index Le measurements that are comparable to the established LAI-2000. The height and angularly dependent leaf clumping index Ω could be determined with DCP, which led to a 30% higher total leaf area index L for DCP compared to LAI-2000. The exclusion of woody tissue from DCP yields on average a 6.9% lower leaf area index L. Including Ω and excluding woody tissue, the L of DCP matched precisely with direct measurements using litter traps. However, the set-up and site-specific adjustment of the image analysis algorithm remains challenging. We propose a special filter for LAI-2000 to enhance data quality when used in open canopies. Finally, if height-dependent observations are not feasible, ground-based observations of crown parameters can be used to derive very reasonable L height distributions from a single, ground-based L observation.