Stith T. Gower

Direct and Indirect Estimation of Leaf Area Index, fAPAR, and Net Primary Production of Terrestrial Ecosystems

A primary objective of the Earth Observing System (EOS) is to develop and validate algorithms to estimate leaf area index (L), fraction of absorbed photosynthetically active radiation (fAPAR), and net primary production (NPP) from remotely sensed products. These three products are important because they relate to or are components of the metabolism of the biosphere and can be determined for terrestrial ecosystems from satellite-borne sensors. The importance of these products in the EOS program necessitates the need to use standard methods to obtain accurate ground truth estimates of L, fAPAR, and NPP that are correlated to satellite-derived estimates. The objective of this article is to review direct and indirect methods used to estimate L, fAPAR, and NPP in terrestrial ecosystems. Direct estimates of L, biomass, and NPP can be obtained by harvesting individual plants, developing allometric equations, and applying these equations to all individuals in the stand. Using non-site-specific allometric equations to estimate L and foliage production can cause large errors because carbon allocation to foliage is influenced by numerous environmental and ecological factors. All of the optical instruments that indirectly estimate L actually estimate “effective” leaf area index (LE) and underestimate L when foliage in the canopy is nonrandomly distributed (i.e., clumped). We discuss several methods, ranging from simple to complex in terms of data needs, that can be used to correct estimates of L when foliage is clumped. Direct estimates of above-ground and below-ground net primary production (NPPA and NPPB, respectively) are laborious, expensive and can only be carried out for small plots, yet there is a great need to obtain global estimates of NPP. Process models, driven by remotely sensed input parameters, are useful tools to examine the influence of global change on the metabolism of terrestrial ecosystems, but an incomplete understanding of carbon allocation continues to hamper development of more accurate NPP models. We summarize carbon allocation patterns for major terrestrial biomes and discuss emerging allocation patterns that can be incorporated into global NPP models. One common process model, light use efficiency or epsilon model, uses remotely sensed fAPAR, light use efficiency (LUE) and carbon allocation coefficients, and other meteorological data to estimates NPP. Such models require reliable estimates of LUE. We summarize the literature and provide LUE coefficients for the major biomes, being careful to correct for inconsistencies in radiation, dry matter and carbon allocation units.

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

Leaf area index (LAI) is a key structural characteristic of forest ecosystems because of the role of green leaves in controlling many biological and physical processes in plant canopies. Accurate LAI estimates are required in studies of ecophysiology, atmosphere-ecosystem interactions, and global change. The objective of this paper is to evaluate LAI values obtained by several research teams using different methods for a broad spectrum of boreal forest types in support of the international Boreal Ecosystem-Atmosphere Study (BOREAS). These methods include destructive sampling and optical instruments: the tracing radiation and architecture of canopies (TRAC), the LAI-2000 plant canopy analyzer, hemispherical photography, and the Sunfleck Ceptometer. The latter three calculate LAI from measured radiation transmittance (gap fraction) using inversion models that assume a random spatial distribution of leaves. It is shown that these instruments underestimate LAI of boreal forest stands where the foliage is clumped. The TRAC quantifies the clumping effect by measuring the canopy gap size distribution. For deciduous stands the clumping index measured from TRAC includes the clumping effect at all scales, but for conifer stands it only resolves the clumping effect at scales larger than the shoot (the basic collection of needles). To determine foliage clumping within conifer shoots, a video camera and rotational light table system was used. The major difficulties in determining the surface area of small conifer needles have been largely overcome by the use of an accurate volume displacement method. Hemispherical photography has the advantage that it also provides a permanent image record of the canopies. Typically, LAI falls in the range from 1 to 4 for jack pine and aspen forests and from 1 to 6 for black spruce. Our comparative studies provide the most comprehensive set of LAI estimates available for boreal forests and demonstrate that optical techniques, combined with limited direct foliage sampling, can be used to obtain quick and accurate LAI measurements.

Testing the performance of a dynamic global ecosystem model: Water balance, carbon balance, and vegetation structure

While a new class of Dynamic Global Ecosystem Models (DGEMs) has emerged in the past few years as an important tool for describing global biogeochemical cycles and atmosphere-biosphere interactions, these models are still largely untested. Here we analyze the behavior of a new DGEM and compare the results to global-scale observations of water balance, carbon balance, and vegetation structure. In this study, we use version 2 of the Integrated Biosphere Simulator (IBIS), which includes several major improvements and additions to the prototype model developed by Foley et al. [1996]. IBIS is designed to be a comprehensive model of the terrestrial biosphere; the model represents a wide range of processes, including land surface physics, canopy physiology, plant phenology, vegetation dynamics and competition, and carbon and nutrient cycling. The model generates global simulations of the surface water balance (e.g., runoff), the terrestrial carbon balance (e.g., net primary production, net ecosystem exchange, soil carbon, aboveground and belowground litter, and soil CO2 fluxes), and vegetation structure (e.g., biomass, leaf area index, and vegetation composition). In order to test the performance of the model, we have assembled a wide range of continental and global-scale data, including measurements of river discharge, net primary production, vegetation structure, root biomass, soil carbon, litter carbon, and soil CO2 flux. Using these field data and model results for the contemporary biosphere (1965–1994), our evaluation shows that simulated patterns of runoff, NPP, biomass, leaf area index, soil carbon, and total soil CO2 flux agree reasonably well with measurements that have been compiled from numerous ecosystems. These results also compare favorably to other global model results.

A global terrestrial monitoring network integrating tower fluxes, flask sampling, ecosystem modeling and EOS satellite data

Abstract Accurate monitoring of global scale changes in the terrestrial biosphere has become acutely important as the scope of human impacts on biological systems and atmospheric chemistry grows. For example, the Kyoto Protocol of 1997 signals some of the dramatic socioeconomic and political decisions that may lie ahead concerning CO2 emissions and global carbon cycle impacts. These decisions will rely heavily on accurate measures of global biospheric changes Schimel 1998 , IGBP TCWG 1998 . An array of national and international programs have inaugurated global satellite observations, critical field measurements of carbon and water fluxes, and global model development for the purposes of beginning to monitor the biosphere. The detection by these programs of interannual variability of ecosystem fluxes and of longer term trends will permit early indication of fundamental biospheric changes which might otherwise go undetected until major biome conversion begins. This article describes a blueprint for more comprehensive coordination of the various flux measurement and modeling activities into a global terrestrial monitoring network that will have direct relevance to the political decision making of global change.

BOREAS in 1997: Experiment overview, scientific results, and future directions

The goal of the Boreal Ecosystem-Atmosphere Study (BOREAS) is to improve our understanding of the interactions between the boreal forest biome and the atmosphere in order to clarify their roles in global change. This overview paper describes the science background and motivations for BOREAS and the experimental design and operations of the BOREAS 1994 and BOREAS 1996 field years. The findings of the 83 papers in this journal special issue are reviewed. In section 7, important scientific results of the project to date are summarized and future research directions are identified.

Carbon distribution and aboveground net primary production in aspen, jack pine, and black spruce stands in Saskatchewan and Manitoba, Canada

The objectives of this study are to (1) characterize the carbon (C) content, leaf area index, and aboveground net primary production (ANPP) for mature aspen, black spruce, and young and mature jack pine stands at the southern and northern Boreal Ecosystem-Atmosphere Study (BOREAS) areas and (2) compare net primary production and carbon allocation coefficients for the major boreal forest types of the world. Direct estimates of leaf area index, defined as one half of the total leaf surface area, range from a minimum of 1.8 for jack pine forests to a maximum of 5.6 for black spruce forests; stems comprise 5 to 15% of the total overstory plant area. In the BOREAS study, total ecosystem (vegetation plus detritus plus soil) carbon content is greatest in the black spruce forests (445,760–479,380 kg C ha−1), with 87 to 88% of the C in the soil, and is lowest in the jack pine stands (68,370–68,980 kg C ha−1) with a similar distribution of carbon in the vegetation and soil. Forest floor carbon content and mean residence time (MRT) also vary more among forest types in a study area than between study areas for a forest type; forest floor MRT range from 16 to 19 years for aspen stands to 28 to 39 years for jack pine stands. ANPP differs significantly among the mature forests at each of the BOREAS study areas, ranging from a maximum of 3490 to 3520 kg C ha−1 yr−1 for aspen stands to 1170 to 1220 kg C ha−1 yr−1 for jack pine stands. Both net primary production (NPP) and carbon allocation differ between boreal evergreen and deciduous forests in the world, suggesting global primary production models should distinguish between these two forest types. On average, 56% of NPP for boreal forests occurs as detritus and illustrates the need to better understand factors controlling aboveground and below-ground detritus production in boreal forests.

NITROGEN MINERALIZATION AND PRODUCTIVITY IN 50 HARDWOOD AND CONIFER STANDS ON DIVERSE SOILS

The generality of relationships between soil net nitrogen (N) mineralization, aboveground N cycling, and aboveground net primary production (ANPP) for temperate forest ecosystems is unclear. It is also not known whether these variables and their rela- tionships differ between evergreen and deciduous forests, or across soil types. To address these questions we compiled data on annual rates of in situ net N mineralization and ANPP for 16 conifer and 34 hardwood forests, including plantations and natural stands on a range of soils at six locations in Wisconsin and Minnesota, USA. For 31 natural stands, 48 stands with native species (including plantations), and all data, ANPP increased linearly with annual net N mineralization rates. Native evergreen conifer and two deciduous hardwood types (oaks and mesic hardwoods) followed similar patterns in this regression, indicating common functional relationships at the ecosystem level. The relationship of N mineraliza- tion and ANPP differed between finer textured Alfisol soils and sandier Entisols, with higher ANPP at any given N mineralization level in Alfisols. A multiple regression of N mineralization on soil texture (percentage silt plus clay), litterfall N, and mean annual temperature explained 81% of the variance in annual N mineralization for natural stands, and a multiple regression of ANPP on soil texture and annual N mineralization rate explained 83% of the variance in ANPP. Naturally regenerated forest types differed in mean annual net N mineralization, litterfall N, and ANPP, and all were greater in oaks than in mesic hardwoods or conifers, respectively. However, differences among the 50 stands and six locations were largely a result of dif- ferences in soils and stand origin. For all natural hardwood stands, ANPP and N miner- alization were greater on fine-textured Alfisols than on sandy Entisols. For evergreen co- nifers, ANPP and N mineralization were greater in plantations on Alfisols than in natural stands on Histosols, Entisols, or Spodosols. Hardwood and evergreen conifer stands did not differ significantly in ANPP or N mineralization on comparable soils and stand origin: they differed neither as plantations on Alfisols nor as natural stands on Entisols. This suggests that observed average differences among natural forest types in ANPP and N mineralization resulted largely from variation in their distribution on differing soils, and not from feedback effects on N mineralization or differing productivity per available N. These data suggest that, at a regional scale, at least half of the variation in ANPP can be attributed to variation in annual N mineralization. Both ANPP and N mineralization differ more strongly with soil type/parent material than with forest type; ANPP at any given level of N mineralization is higher on silty/loamy Alfisols than on sandy Entisols, Histosols, or Spodosols, but not different for coniferous and broad-leaved deciduous species. There is no indication of N saturation of ANPP within the range of natural N availability in these

Aboveground net primary production decline with stand age: potential causes

Aboveground net primary production (ANPP) commonly reaches a maximum in young forest stands and decreases by 0-76% as stands mature. However, the mechanism(s) responsible for the decline are not well understood. Current hypotheses for declining ANPP with stand age include: (1) an altered balance between photosynthetic and respiring tissues, (2) decreasing soil nutrient availability, and (3) increasing stomatal limitation leading to reduced photosynthetic rates. Recent empirical and modeling studies reveal that mechanisms (2) and (3) are largely responsible for age-related decline in ANPP for forests in cold environments. Increasing respiratory costs appear to be relatively unimportant in explaining declining productivity in ageing stands.

NET PRIMARY PRODUCTION AND CARBON ALLOCATION PATTERNS OF BOREAL FOREST ECOSYSTEMS

The three objectives of this paper were: to summarize net primary production (NPP) and carbon allocation patterns for boreal forests, to examine relationships between climatic and biological variables and NPP, and to examine carbon allocation coefficients for all boreal forests or types of boreal forests that can be used to estimate NPP from easily measured components of NPP. Twenty-four Class I stands (complete NPP budgets) and 45 Class II boreal forest stands (aboveground NPP [NPPA] and budget only) were identified. The geographic distribution of the Class I stands was not uniform; 46% of the stands were from two studies in North America, and only one stand was from the important larch forests of Eurasia. Total (above- and belowground) net primary production (NPPT) ranged from 52 to 868 g C·m−2·yr−1 and averaged 424 g C·m−2·yr−1. NPPA was consistently larger for deciduous than for evergreen boreal forests in each of the major boreal regions, especially for boreal forests in Alaska. Belowground net prima...

An improved strategy for regression of biophysical variables and Landsat ETM+ data

Empirical models are important tools for relating field-measured biophysical variables to remote sensing data. Regression analysis has been a popular empirical method of linking these two types of data to provide continuous estimates for variables such as biomass, percent woody canopy cover, and leaf area index (LAI). Traditional methods of regression are not sufficient when resulting biophysical surfaces derived from remote sensing are subsequently used to drive ecosystem process models. Most regression analyses in remote sensing rely on a single spectral vegetation index (SVI) based on red and near-infrared reflectance from a single date of imagery. There are compelling reasons for utilizing greater spectral dimensionality, and for including SVIs from multiple dates in a regression analysis. Moreover, when including multiple SVIs and/or dates, it is useful to integrate these into a single index for regression modeling. Selection of an appropriate regression model, use of multiple SVIs from multiple dates of imagery as predictor variables, and employment of canonical correlation analysis (CCA) to integrate these multiple indices into a single index represent a significant strategic improvement over existing uses of regression analysis in remote sensing. To demonstrate this improved strategy, we compared three different types of regression models to predict LAI for an agro-ecosystem and live tree canopy cover for a needleleaf evergreen boreal forest: traditional (Yon X) ordinary least squares (OLS) regression, inverse (X on Y) OLS regression, and an orthogonal regression method called reduced major axis (RMA). Each model incorporated multiple SVIs from multiple dates and CCA was used to integrate these. For a given dataset, the three regression-modeling approaches produced identical coefficients of determination and intercepts, but different slopes, giving rise to divergent predictive characteristics. The traditional approach yielded the lowest root mean square error (RMSE), but the variance in the predictions was lower than the variance in the observed dataset. The inverse method had the highest RMSE and the variance was inflated relative to the variance of the observed dataset. RMA provided an intermediate set of predictions in terms of the RMSE, and the variance in the observations was preserved in the predictions. These results are predictable from regression theory, but that theory has been essentially ignored within the discipline of remote sensing. D 2002 Elsevier Science Inc. All rights reserved.