Elsevier

Remote Sensing of Environment

Volume 112, Issue 10, 15 October 2008, Pages 3947-3957
Remote Sensing of Environment

Discrete return lidar-based prediction of leaf area index in two conifer forests

https://doi.org/10.1016/j.rse.2008.07.001Get rights and content

Abstract

Leaf area index (LAI) is a key forest structural characteristic that serves as a primary control for exchanges of mass and energy within a vegetated ecosystem. Most previous attempts to estimate LAI from remotely sensed data have relied on empirical relationships between field-measured observations and various spectral vegetation indices (SVIs) derived from optical imagery or the inversion of canopy radiative transfer models. However, as biomass within an ecosystem increases, accurate LAI estimates are difficult to quantify. Here we use lidar data in conjunction with SPOT5-derived spectral vegetation indices (SVIs) to examine the extent to which integration of both lidar and spectral datasets can estimate specific LAI quantities over a broad range of conifer forest stands in the northern Rocky Mountains. Our results show that SPOT5-derived SVIs performed poorly across our study areas, explaining less than 50% of variation in observed LAI, while lidar-only models account for a significant amount of variation across the two study areas located in northern Idaho; the St. Joe Woodlands (R2 = 0.86; RMSE = 0.76) and the Nez Perce Reservation (R2 = 0.69; RMSE = 0.61). Further, we found that LAI models derived from lidar metrics were only incrementally improved with the inclusion of SPOT 5-derived SVIs; increases in R2 ranged from 0.02–0.04, though model RMSE values decreased for most models (0–11.76% decrease). Significant lidar-only models tended to utilize a common set of predictor variables such as canopy percentile heights and percentile height differences, percent canopy cover metrics, and covariates that described lidar height distributional parameters. All integrated lidar-SPOT 5 models included textural measures of the visible wavelengths (e.g. green and red reflectance). Due to the limited amount of LAI model improvement when adding SPOT 5 metrics to lidar data, we conclude that lidar data alone can provide superior estimates of LAI for our study areas.

Introduction

The foliage component of a forest canopy is the primary surface that controls mass, energy, and gas exchange between photosynthetically active vegetation and the atmosphere (Fournier et al., 2003). A thorough characterization of leaf area index (LAI; the ratio of half of the total needle surface area per unit ground area) can therefore provide valuable information about nutrient cycling, hydrologic forecasting, and biogeochemical processes in a forested ecosystem. As a key vegetation structural characteristic that drives many vegetation functions, LAI is a primary parameter used in ecophysiological and biogeochemical models to describe plant canopies (Chen et al., 1997). For example, process-based models such as BIOMASS (McMurtrie & Landsberg, 1992), FOREST-BGC (Running & Coughlan, 1988) and RHESSys (Band et al., 1991) use LAI as a primary or intermediate variable for forest growth and productivity. Additionally, LAI is often employed as a critical calibration variable for remote sensing datasets to differentiate vegetation characteristics over a wide range of biomes (Coops et al., 2004). LAI has also been used to characterize forest radiation regimes and the amount of light available to the understory in tropical (e.g. Rich et al., 1993, Vierling and Wessman, 2000) and temperate conifer (e.g. Law et al., 2001a) and deciduous forests (e.g. Ellsworth & Reich, 1993). Given the role of LAI in determining many forest ecosystem processes, several techniques have been developed for rapid LAI estimation.

The most commonly employed methods for estimating LAI across landscapes rely on the relationships between LAI and various manipulations of spectral information from aircraft or satellite-based imagery. A significant amount of research has been dedicated to quantifying the connections between spectral vegetation indices (SVIs) that associate foliar composition in the visible red waveband, which is absorbed by chlorophyll a and b, and the near-infrared waveband, which is scattered by plant cellular structures. The normalized difference vegetation index (NDVI) (Rouse et al., 1974) and the simple ratio (SR) (Birth & McVey, 1968) are the most frequently used SVIs to estimate LAI for a variety of ecosystem types including coniferous forests (Chen et al., 1997, Curran et al., 1992), grasslands (Friedl et al., 1994) and deciduous forests (Coops et al., 2004). Recent studies have incorporated more complex vegetation indices by including spectral response from additional wavelengths in an effort to minimize the influences of atmospheric disparities and canopy background noise. For example, a mid-infrared correction proposed by Nemani et al. (1993) to NDVI and SR have been found by White et al. (1997) and Pocewicz et al. (2004) to improve LAI estimates in montane and temperate coniferous forests. Lymburner et al. (2000) developed the specific leaf area vegetation index (SLAVI) to account for mid-infrared sensitivity to varying canopy structure for heterogeneous forest/woodland compositions. Chen et al. (2004) examined the use of the enhanced vegetation index (EVI; Huete et al., 1997) to improve LAI and vegetation cover estimates in a ponderosa pine forest. The reduced simple ratio (RSR) has demonstrated success for estimating LAI in pine and spruce stands (Stenberg et al., 2004) and for a post-fire chronosequence in Siberia (Chen et al., 2005b).

Overall, commonly used SVIs serve as suitable surrogates to approximate LAI for canopies with relatively low LAI (e.g. LAI = 3–5) (Chen and Cihlar, 1996, Turner et al., 1999). However, for values above this LAI threshold, many SVIs tend to saturate such that LAI estimates for high biomass forests may be grossly underestimated. For most temperate coniferous forests, the ability to discriminate higher LAI values from optical remote sensing data has been a major challenge.

Lidar data provide an alternative approach for estimating LAI across the landscape. Throughout the past decade, many researchers have reported the utility of lidar data to estimate a suite of forest biophysical characteristics such as canopy height, basal area, crown closure, wood volume, stem density, and biomass (Maclean and Krabill, 1986, Means et al., 2000, Naesset and Bjerknes, 2001, Nelson et al., 1988, Popescu et al., 2003) over a range of forest structural types and regional (Lefsky et al., 2005a) to sub-regional scales (Jensen et al., 2006). More recently, researchers have attempted to relate the three-dimensional structural information captured with lidar data to both direct and indirect estimates of LAI based on various analytical methods. For instance, Magnussen and Boudewyn (1998) found that the proportion of lidar returns corresponding to calculated canopy heights was correlated with the fractional leaf area above canopy-specific height thresholds. Lefsky et al. (1999) explored a three-dimensional (volumetric) analysis of waveform lidar data to estimate leaf area index within a multiple regression framework. Chen et al. (2004) investigated the relationships between trees identified with lidar data tree cover response obtained by a discrete-return system to spectrally-derived vegetation indices and LAI. Riano et al. (2004) and Morsdorf et al. (2006) assessed the capacity of lidar and variable-radius plots to estimate LAI. Lefsky et al. (2005a) developed robust empirical estimates based on waveform lidar and regional LAI measurements for the U.S. Pacific Northwest and Koetz et al. (2006) inverted both actual and simulated 3-D lidar waveform models to estimate LAI and other biophysical parameters within a radiative transfer model.

LAI can be estimated from a variety of remote sensing datasets, warranting the exploration of lidar and multispectral data integration. Lidar/multispectral data integration (also referred to as data fusion or synergy) has been explored for retrieval of other forest characteristics such as canopy height (Hudak et al., 2002, Popescu and Wynne, 2004, Wulder and Seemann, 2003), volume and biomass (Hudak et al., 2006, Popescu et al., 2004), stand density (McCombs et al., 2003), forest productivity (Lefsky et al., 2005b), canopy change detection (Wulder et al., 2007) and characterization of foliage pigments (Blackburn, 2002). However, the potential for spatial and spectral data integration remains significantly unaddressed in terms of quantifying and mapping LAI in moderate to high biomass coniferous forests.

Previous studies of LAI in northern Idaho conifer forests have reported LAI ranging from 0 to 13, with the majority of observations exceeding LAI = 4 (Duursma et al., 2003, Pocewicz et al., 2004). In terms of geographic significance, the northern Idaho mountain ranges may represent the region of highest carbon uptake in the Rocky Mountain range, and thus the most substantial carbon sink between the Cascade Mountains and the Midwestern U.S. (Schimel et al., 2002). Therefore, accurate and reliable estimates of LAI are vital to adequately characterize ecosystem processes and monitor trajectories of change. Currently, operational LAI products from the MODIS sensor and SPOT VEGETATION provide repeat spatial and temporal coverage of biophysical variables used to describe vegetation structure (Baret et al., 2007, Yang et al., 2006), but at a much coarser spatial resolution such that heterogeneity of fine-to-medium scale landscape features is lost.

The specific objectives of our research are to determine 1) the capability of lidar-derived covariates to estimate measured and corrected LAI quantities, 2) the extent to which SPOT 5 spectral data may improve lidar-based LAI estimates, and 3) the applicability of a regional model to quantify LAI in northern Rocky Mountain forests.

Section snippets

Study areas

Forested regions of northern Idaho exhibit a wide range of stand characteristics representative of conifer forests in the Northern Rocky mountains, and more generally, the western United States. A diverse range of topographic and climatic conditions combined with forest management practices serve to determine species composition and land-use patterns in the Intermountain West. To meet our research objectives, two distinct forested areas were selected to represent the broader range of forest

Results

Exploratory data analysis indicated that LAI quantities were not normally distributed; thus, response data were transformed to satisfy the normality assumption for linear regression. A natural log transformation was used for the SJW and a square root transformation for the NPR. The combined dataset used the square root and natural log transformation for LAIe and corrected LAI quantities, respectively. Different transformations were required for the combined dataset because a single

Regression analysis

Lidar-derived covariates explained the largest proportion of variation in LAI and corrected quantities among the three datasets used in this analysis. Although existing methods to estimate LAI often rely on a single optically-derived SVI, the relationships are often asymptotic and can result in unreliable estimates for moderate to high biomass forests. The number of lidar covariates selected for each model was a balance between parsimony and relevance, or the explanatory value of individual

Conclusion

The two selected study areas represent a diverse assemblage of ecoregional characteristics, climatic conditions, and anthropogenic influences including management ideology and implementation. Such factors control the type, density, and location of vegetation both within an individual stand and the region as a whole. Despite this matrix of variable forest conditions, lidar data were able to account for a significant amount of variation in measured LAI for both individual study areas and when

Acknowledgements

Funding for this research supported by NSF Idaho EPSCoR program grant EPS-447689; by NASA Idaho Space Grant Consortium grant NGG-05GG29H, a NASA Earth Sciences Enterprise Application Division grant BAA-01-OES-01, and NASA EPSCoR grant NCC5-588. We would like to particularly thank the Potlatch Corporation and the Nez Perce Tribe for allowing this research to be conducted on their lands. The authors acknowledge Drs. Jing Chen and Sylvain Leblanc for their communications regarding LAI measurements

References (83)

  • CurranP.J. et al.

    Seasonal LAI in slash pine estimated with Landsat TM

    Remote Sensing of Environment

    (1992)
  • DuursmaR.A. et al.

    Leaf area index inferred from solar beam transmission in mixed conifer forests on complex terrain

    Agricultural and Forest Meteorology

    (2003)
  • FalkowskiM.J. et al.

    Characterizing and mapping forest fire fuels using ASTER imagery and gradient modeling

    Forest Ecology and Management

    (2005)
  • FernandesR. et al.

    Parametric (modified least squares) and non-parametric (Theil–Sen) linear regressions for predicting biophysical parameters in the presence of measurement errors

    Remote Sensing of Environment

    (2005)
  • GitelsonA.A. et al.

    Remote estimation of leaf area index and green leaf biomass in maize canopies

    Journal of Plant Physiology

    (2004)
  • GowerS.T. et al.

    Direct and indirect estimation of Leaf Area Index, fAPAR, and net primary production of terrestrial ecosystems

    Remote Sensing of Environment

    (1999)
  • HudakA.T. et al.

    Integration of lidar and Landsat ETM plus data for estimating and mapping forest canopy height

    Remote Sensing of Environment

    (2002)
  • JacquemoudS. et al.

    Extraction of vegetation biophysical parameters by inversion of the PROSPECT + SAIL models on sugar beet canopy reflectance data. Application to TM and AVIRIS sensors

    Remote Sensing of Environment

    (1995)
  • LawB.E. et al.

    Estimation of leaf area index in open-canopy ponderosa pine forests at different successional stages and management regimes in Oregon

    Agricultural and Forest Meteorology

    (2001)
  • LefskyM.A. et al.

    Lidar remote sensing of the canopy structure and biophysical properties of Douglas-Fir Western Hemlock Forests

    Remote Sensing of Environment

    (1999)
  • LefskyM.A. et al.

    Geographic variability in lidar predictions of forest stand structure in the Pacific Northwest

    Remote Sensing of Environment

    (2005)
  • LefskyM.A. et al.

    Combining lidar estimates of aboveground biomass and Landsat estimates of stand age for spatially extensive validation of modeled forest productivity

    Remote Sensing of Environment

    (2005)
  • MorsdorfF. et al.

    Estimation of LAI and fractional cover from small footprint airborne laser scanning data based on gap fraction

    Remote Sensing of Environment

    (2006)
  • NaessetE. et al.

    Estimating tree heights and number of stems in young forest stands using airborne laser scanner data

    Remote Sensing of Environment

    (2001)
  • NelsonR. et al.

    Estimating forest biomass and volume using airborne laser data

    Remote Sensing of Environment

    (1988)
  • PocewiczA. et al.

    View angle effects on relationships between MISR vegetation indices and leaf area index in a recently burned ponderosa pine forest

    Remote Sensing of Environment

    (2007)
  • RianoD. et al.

    Estimation of leaf area index and covered ground from airborne laser scanner (Lidar) in two contrasting forests

    Agricultural and Forest Meteorology

    (2004)
  • RichP.M. et al.

    Long-term study of solar radiation regimes in a tropical wet forest using quantum sensors and hemispherical photography

    Agricultural and Forest Meteorology

    (1993)
  • RunningS.W. et al.

    A general model of forest ecosystem processes for regional applications I. Hydrologic balance, canopy gas exchange and primary production processes

    Ecological Modeling

    (1988)
  • TuckerC.J.

    Red and photographic infrared linear combinations for monitoring vegetation

    Remote Sensing of Environment

    (1979)
  • TurnerD.P. et al.

    Relationships between leaf area index and Landsat spectral vegetation indices across three temperate zone sites

    Remote Sensing of Environment

    (1999)
  • VierlingL.A. et al.

    Photosynthetically active radiation heterogeneity within a monodominant Congolese rain forest canopy

    Agricultural and Forest Meteorology

    (2000)
  • WalthallC. et al.

    A comparison of empirical and neural network approaches for estimating corn and soybean leaf area index from Landsat ETM+ imagery

    Remote Sensing of Environment

    (2004)
  • WulderM.A. et al.

    Integrating profiling LIDAR with Landsat data for regional boreal forest canopy attribute estimation and change characterization

    Remote Sensing of Environment

    (2007)
  • WulderM.A. et al.

    Aerial image texture information in the estimation of Northern Deciduous and Mixed Wood Forest Leaf Area Index (LAI)

    Remote Sensing of Environment

    (1998)
  • Allen, D.M. (1971). The prediction sum of squares as a criterion for selecting predictor variables. Univ. of Ky. Dept....
  • BirthG.S. et al.

    Measuring the color of growing turf with a reflectance spectroradiometer

    Agronomy Journal

    (1968)
  • ChenJ.M. et al.

    Quantifying the effect of canopy architecture on optical measurements of leaf area index using two gap size analysis methods

    IEEE Transactions on Geoscience and Remote Sensing

    (1995)
  • ChenX.X. et al.

    Monitoring boreal forest leaf area index across a Siberian burn chronosequence: A MODIS validation study

    International Journal of Remote Sensing

    (2005)
  • ChenJ.M. et al.

    Leaf area index of boreal forests: Theory, techniques and measurements

    Journal of Geophysical Research

    (1997)
  • ClawgesR. et al.

    Use of a ground-based scanning lidar for estimation of biophysical properties of western larch (Larix occidentalis)

    International Journal of Remote Sensing

    (2007)
  • Cited by (134)

    View all citing articles on Scopus
    View full text