Application of Geostatistics to Characterize Leaf Area Index (LAI) from Flux Tower to Landscape Scales Using a Cyclic Sampling Design

AbstractAccurate characterization of leaf area index (LAI) is required to quantify the exchange of energy, water, and carbon between terrestrial ecosystems and the atmosphere. The objective of this study was to use a cyclic sampling design to compare the spatial patterns of LAI of the dominant terrestrial ecosystems that comprised the area around the 447-m WLEF television tower, equipped with an eddy flux system, near Park Falls, Wisconsin, USA. A second objective was to compare the efficiency of cyclic, random, and uniform sampling designs in terms of the precision of spatial information derived per unit sampling effort. The vegetation surrounding the tower was comprised (more than 80%) of four major forest cover types: forested wetlands, upland aspen forests, upland northern hardwood forests, and upland pine forests, and a fifth, nonforested cover type, grass (open meadow). LAI differed significantly among the five cover types and averaged 3.45, 3.57, 3.82, 3.99, and 1.14 for northern hardwoods, aspen, forested wetlands, upland conifers, and grass, respectively. The cyclic sampling design maximized information about the variance of vegetation characteristics of the heterogeneous landscape and decreased by 60% the number of plots needed to obtain the same confidence interval width using a random sampling design. The range of spatial autocorrelation for LAI was 147 m, but it was decreased to 117 m when vegetation cover was included as a covariate. The cyclic sampling design has several important advantages over other sampling designs. The cyclic sampling design increased the sampling efficiency by optimizing the placement of plots so they were distributed more efficiently for geostatistical analyses such as semi-variograms, correlograms, and spatial regression and can incorporate covariates (for example, vegetation cover, soil properties, and so on) that may explain the sources of spatial patterns. The cyclic sampling design was used to derive a spatial map of LAI and the average LAI for the 3 × 2 km area centered on the flux tower was 3.51 ± 0.89 (with a minimum of 0 and a maximum of 6.35). Airborne and satellite reflectance data have also been used to characterize LAI, but in this region, and many other forests of the world, remotely sensed vegetation indexes saturate in forests with an LAI greater than 3–5. The cyclic sampling design also provides a general ecological sampling approach that can be used at multiple scales.

[1]  M. MacKenzie,et al.  Effects of sensor spatial resolution on landscape structure parameters , 1995, Landscape Ecology.

[2]  M. Clayton,et al.  Cradle‐Knoll Patterns and Characteristics on Drumlins in Northeastern Wisconsin , 1997 .

[3]  W. G. Müller,et al.  Optimal designs for variogram estimation , 1999 .

[4]  Ronaldo Iachan,et al.  Systematic Sampling: A Critical Review , 1982 .

[5]  Noel A Cressie,et al.  Statistics for Spatial Data. , 1992 .

[6]  Marie-Josée Fortin,et al.  Spatial autocorrelation and sampling design in plant ecology , 1989, Vegetatio.

[7]  S. T. Gower,et al.  Leaf area index of boreal forests: theory, techniques, and measurements , 1997 .

[8]  P. Legendre.  Spatial Autocorrelation: Trouble or New Paradigm? , 1993 .

[9]  Douglas E. Ahl,et al.  Effects of aggregated classifications of forest composition on estimates of evapotranspiration in a northern Wisconsin forest , 2002 .

[10]  D. Bates,et al.  Mixed-Effects Models in S and S-PLUS , 2001 .

[11]  T. M. Lillesand,et al.  Estimating the leaf area index of North Central Wisconsin forests using the landsat thematic mapper , 1997 .

[12]  M. Clayton,et al.  OLD-GROWTH NORTHERN HARDWOOD FORESTS: SPATIAL AUTOCORRELATION AND PATTERNS OF UNDERSTORY VEGETATION , 2002 .

[13]  Karin S. Fassnacht,et al.  Comparison of the litterfall and forest floor organic matter and nitrogen dynamics of upland forest ecosystems in north central Wisconsin , 1999 .

[14]  Anthony N. Pettitt,et al.  Sampling Designs for Estimating Spatial Variance Components , 1993 .

[15]  M. Clayton,et al.  Confidence Intervals for Autocorrelations Based on Cyclic Samples , 1995 .

[16]  Mike Rees,et al.  5. Statistics for Spatial Data , 1993 .

[17]  S. Gower,et al.  Interrelationships among the edaphic and stand characteristics, leaf area index, and aboveground net primary production of upland forest ecosystems in north central Wisconsin , 1997 .

[18]  K. Hibbard,et al.  A Global Terrestrial Monitoring Network Integrating Tower Fluxes, Flask Sampling, Ecosystem Modeling and EOS Satellite Data , 1999 .

[19]  D. Russo,et al.  Optimal spatial sampling design for the estimation of the variogram based on a least squares approach , 1999 .

[20]  William D. Clinger,et al.  On Unequally Spaced Time Points in Time Series , 1976 .

[21]  C. Justice,et al.  Validating MODIS Terrestrial Ecology Products: Linking In Situ and Satellite Measurements , 1999 .

[22]  Philippe Aubry,et al.  GEOSTATISTICAL ESTIMATION VARIANCE FOR THE SPATIAL MEAN IN TWO-DIMENSIONAL SYSTEMATIC SAMPLING , 2000 .

[23]  Stephen P. Kaluzny,et al.  S+SpatialStats: User’s Manual for Windows® and UNIX® , 1998 .

[24]  Frank van Diggelen,et al.  GPS and GPS+GLONASS RTK , 1997 .

[25]  Pierre Dutilleul,et al.  Spatial Heterogeneity and the Design of Ecological Field Experiments , 1993 .

[26]  S. T. Gower,et al.  Direct and Indirect Estimation of Leaf Area Index, fAPAR, and Net Primary Production of Terrestrial Ecosystems , 1999 .

[27]  Pieter P. Tans,et al.  Measurements of carbon dioxide on very tall towers: results of the NOAA/CMDL program , 1998 .

[28]  S. T. Gower,et al.  A comparison of optical and direct methods for estimating foliage surface area index in forests , 1994 .

[29]  P. Crill,et al.  CO2 and CH4 flux between a boreal beaver pond and the atmosphere , 1997 .

[30]  S. T. Gower,et al.  Rapid Estimation of Leaf Area Index in Conifer and Broad-Leaf Plantations , 1991 .

[31]  Hong S. He,et al.  Periodicity in spatial data and geostatistical models: autocorrelation between patches , 2000 .