Printer Friendly

Vegetation biomass data in the Amazon: how good is historical data for remote sensing ground truth?

Abstract

One of the most important remaining forests_in the world is in th Amazon region. A great part of the world's carbon is stored in the Amazon's vegetation biomass, but its exact amount, and the assessment of the region as either a sink or a source of carbon is still open to scientific discussion. The first step in understanding biomass behavior is its accurate measurement. In this paper, I analyze how biomass field measurements are made and possible problems with the data. I discuss its suitability as remote sensing ground truth and make the case for remote sensing as a tool to scale punctual and/or site-specific data.

Introduction

With the rise in the atmosphere's carbon dioxide and the resulting climate change, there has been an increase in the number of studies that attempt to measure forest biomass. The objective of these studies is to determine if forests are sinks or sources of carbon and their subsequent role in the global climate. An often complicated process, forest biomass must be measured at different time intervals to determine if carbon is increasing or decreasing. Although a series of studies have been conducted in the Amazon in the last decade, the biomass numbers are very different, even when performed in the same region.

In this paper, I review a limited number of studies on the different methods and inferences used to assess vegetation biomass (quantification and partitioning). I also calculate means and coefficients of variation for the results presented. I discuss how to measure vegetation biomass in one moment of a time series and I comment on the methods and uncertainty of the sources. Certain authors have recently questioned methods of acquiring vegetation biomass values (e.g. Clark et al. 2001; Clark 2002; Keller et al. 2001), thus I discuss the importance of those alleged problems when using this data as remote sensing ground truth. Finally, I make the case for remote sensing as a tool to scale punctual and/or site-specific data.

Measuring Biomass

Different approaches can be used to assess vegetation biomass. The simplest method that is less prone to errors is to remove all vegetation biomass in a certain area and then to dry and weigh it. This method, called the destructive method, is usually used when measuring small areas only since all the vegetation is destroyed.

The most common approach when measuring forest biomass is to use the destructive method for a small area, and, based on the results, to devise allometric equations. These equations can measure the vegetation biomass of larger areas indirectly by using non-destructive measurements such as diameter at breast height (DBH) or basal area as inputs. Trees above a certain DBH are measured and their species are identified, an important detail that determines the appropriate allometric equation. Depending on the tree species, the density will have a different weight in the equation or may not be included (e.g. allometric equations for hollow species such as Cecropia spp, do not include density).

Partitioning

Aboveground live biomass (AGLB) is the most common vegetation biomass measurement. Because this method was developed by the timber industry concerned primarily with economically valuable wood in natural or agricultural plots, belowground live biomass (BGLB) is not considered. However, AGLB does take into account leaves and tree branches, as well as smaller plants.

A more representative quantity in carbon budget studies is total biomass (TB) which is the sum of total aboveground biomass (TAGB) plus standing and fallen dead material and BGLB. Belowground biomass can be measured directly, but it is more commonly inferred from AGLB using root/ shoot ratios obtained from the literature (Houghton et al. 2001) since it is difficult to destructively measure big tree roots.

Allometric Equations

Honzak et al. (1996) provide a good example of how allometric regression equations can be used. They tested the TAGB calculation using different allometric equations in an area north of Manaus, Brazil that contains trees smaller than 35 cm in diameter. The inputs were DBH (D, cm), tree height (H, m), and for some of the equations, wood density (S, Mg m-3). The equations are:

TAGB = exp (-2.327 + 0.937 ln(D2HS)), with wood density or:

TAGB = exp (-3.068 + 0.957 ln(D2H)), without wood density.

Belowground biomass was assumed to be 19% of the TAGB based on studies conducted in the Amazon forests. In these studies, the reported BGLB values for tropical forest range from eight percent to 25%.

Although height can be measured directly with an inclinometer using trigonometry formulas, sometimes the top of trees cannot be viewed in a dense tropical forest. In these situations, height estimation equations correlating height to diameter are used. Since each species has a characteristic structure, the equations can be species specific. In more uniform forests, the same equation may be used. In their study, Honzak et al. (1996) used the following equation:

H = exp(1.387+0.539 ln D)

The series of biomass studies done in the Amazon forest demonstrates that species composition greatly influences the amount of biomass calculated using different allometric equations. Nelson et al.'s (1999) study on allometric regressions for improved estimate of secondary forest biomass in the central Amazon illustrates how to adapt original equations to different species compositions. The authors developed species-specific and mixed-species regressions for estimating AGLB using eight abundant forest tree species in central Amazon secondary forests. Neslon et al. (1999) obtained better results when factoring in wood density in the mixed-species equations than in the species-specific equations. They also showed a slight reduction in error when they used tree height. In addition, Nelson et al. tested other equations developed for "primary" forests. Their tests showed that biomass was overestimated by as much as 60%. This overestimation was even bigger when Cecropia DBH was included in the calculations.

Sampling method is another important component in measuring biomass. Using simulation, Keller et al. (2001) tested four different equations and sources of sampling error. Based on their results, a sample of 21 randomly selected 0.25-ha plots in a 392-ha patch of moist tropical forest will produce 20% sampling error with a 95% confidence.

Uncertainty Sources

Hollow Trees, Wood Density, Species

Composition

Fearnside (1992) outlines uncertainty sources when measuring biomass in his rebuttal to Brown and Lugo's (1992) study of AGLB estimates for tropical moist forests in the Brazilian Amazon. Hollow trees are one source of uncertainty. Besides the always structurally hollow Cecropia, other hardwoods acquire this characteristic with maturity. Fernside (1992) cites studies that found that nearly 27% of trees with DBH greater than 40 cm in Manaus are hollow. This means that 30% of the tree stem volume is either air or light material such as debris from termites. Another important factor in calculating biomass is wood density, but the values for wood density are unavailable for most of the Amazonian species. As mentioned above, figures are available only for trees with commercial value (trees with low density) resulting in an artificially low average density. Another problem with determining density is that bark is measured with DBH although it is less dense than wood, and the relation between bark and wood changes as trees grow.

Recently, more fine-tuned allometric equations have been proposed. Nelson et al. (1999) proposed methods to improve forest biomass estimation. By applying species-specific and mixed-species allometric equations that use DBH and specific density of the wood as inputs, the authors reduced error from 10-60% overestimation to 10-15%.

Randomness of Plot Selection

Another important source of bias when measuring biomass is plot selection. In many cases, accessibility, transportation and other logistic issues prevent forest survey plots from being selected at random. In their discussion of sampling and allometric uncertainties, Keller et al. (2001) highlight that most of the forest biomass estimates in the Amazon depend mainly on a limited database of forest plots sampled over three decades. The location of these plots was not randomly selected or distributed over any spatial or vegetation classification scheme, making the assessment of possible bias in site selection difficult to determine. In her analysis of long-term plot data, Clark (2002) also comments on site selection subjectivity and experiment unreplicability. To alleviate this problem, remote sensing can be used to extrapolate results from a specific area to a whole region Data collected by remote sensing can be applied only if the sampling plots exact position is available and plot size and sensor resolution are equivalent.

Methodological Artifacts

Clark (2002) also expressed concern about methodological artifacts in allometric equation inputs such as the use of diameter measurements at breast height (~1.4 m). Most tropical trees have buttresses and other protuberances on their stem. These irregularities have a disproportional rapid radial increase. Thus Clark proposed that measurements should be made above the buttresses Phillips et al. (2002) examined this suggestion by evaluating potential biases related to changes in growth of tropical forests. They concluded that even if the errors identified by Clark occur, they are on the order of 10 percent for basal area, which will not impede the use of most tropical-forest plot data.

Large Scale Biomass Figures

The numbers proposed for the Amazon forest biomass fluctuate greatly due both to the natural variability in a forest over such a large region and to the different methodological approaches in the quantification. Tables 1, 2, 3, and 4 illustrate the discrepancy of some Amazon forest biomass figures.

Table 1 displays the values for total biomass for the whole Brazilian Amazon as found in Fearnside (1992) and Brown and Lugo (1992). Brown and Lugo's data (1992) is separated into four categories: a) studies conducted during the 1950's; b) small sampling areas; c) RADAMBRASIL study plots; and d) results corrected by the authors. The average value was 288 Mg ha-1, with a standard deviation of 69 Mg ha-1 and a coefficient of variation (standard deviation as a percent of the mean) of 24%.

Total biomass values for selected areas are shown in Table 2 for the Tapajos area (Keller et al. 2001) and areas north of Manaus (Fearnside 1992; Lucas et al. 1996). The mean was 323 Mg ha-1, with a standard deviation of 62 Mg ha-1 and a coefficient of variation of 19%.

Table 3 shows the values for the AGLB for selected areas in Tapajos (Keller et al 2001; Luckman et al. 1997) and in areas north of Manaus (Laurance et al. 1999; Carvalho et al. 1995). The mean was 330 Mg ha-1, with a standard deviation of 50 Mg ha-1 and a coefficient of variation of 15%.

Houghton et al. (2001) grouped 34 sites in the Amazon, including areas outside Brazil, according to three different plot sizes (Table 4) and calculated their mean AGLB. The mean of these three classifications was 290 Mg ha-1, with a standard deviation of 40 Mg ha-1 and a coefficient of variation of 14%

Discussion

In this article, I reviewed a limited but representative sample of studies that quantify Amazon forest biomass, a necessary calculation in determining the role of the Amazon forest in the global carbon budget. Assessing whether the forest is capturing carbon dioxide, releasing carbon in the atmosphere, having no net primary productivity is vital for global climate change research Forest biomass is also crucial in determining the amount of carbon release by the anthropogenic action of deforestation and conversion of forest to agriculture or pasture. In addition, biomass measurements make it possible to determine if changes in forest growth are a result of an increase in the atmospheric carbon dioxide.

Biomass estimates are geographically and temporally variable. However, a comparison of the values from Fearnside (1992) and Brown and Lugo' (1992) corrected value in Table 1, show that they are very close, and the value, for Manaus and Tapajos in Table 3 are almost the same.

Biomass values vary only slightly when calculated for large regions using data collected from many sites. This is evident in Table 4, which shows a coefficient of variation of 14% when analyzing plots of different sizes. The mean AGLB computed using Houghton et al.'s (2001) results is very close to the values calculated by Fearnside (1992) and the corrected values calculated by Brown and Lugo (1992) (Table 1). In contrast, the values for total biomass given by Fearnside (1992) and Lucas et al. (1996) in Table 2 differ greatly. This difference is expected considering Fearnside's (1992) value is an average for all the area north of Manaus, while Lucas et al.'s (1996) is an average of only two plots in the area.

Honzak et al. (1996) recommended using data collected from remote sensing to estimate biomass. They emphasized the importance of basing predictive relationships on accurate ground data. These data can be obtained from historical plots or from traditional field campaigns. Errors from historical plots may be smaller since measurements are taken in the same site over time, which allow the use of area specific allometric equations.

Radar remote sensing data offers one of the best approaches to extrapolating biomass to large areas (Luckman et al. 1997). The radar saturates its signal only in very high biomass areas, contrary to optical vegetation indexes. Using precise biomass measurements to calibrate the backscatter models makes it possible to estimate the biomass across large regions, distinguishing among areas with varied biomass density.

Datos de Biomasa Vegetativa en el Amazonas: Que tan Correctos Son los Datos Historicos para la Verificacion en el Terreno de Informacion Obtenida con Sensores Remotos?

Resumen

La region amazonica es uno de los remanentes boscosos mas importantes en el mundo. Una gran parte del carbon mundial se encuentra almacenada en la biomasa vegetal del Amazonas, pero las cantidades exactas de carbon y la evaluacion de la region como una zona de absorcion o produccion de carbon es un asunto aun sin resolver. El primer paso en, entendimiento del comportamiento de la biomasa es lograr una medicion exacta de la misma. En este articulo analizo la forma en que la obtencion en campo de datos de carbon es Ilevada a cabo, asi como los posibles problemas con este tipo de datos. Ademas, discuto la posibilidad de usar informacion obtenida con sensores remotos como una herramienta de evaluacion de biomasa.

Biomasse de Vegetation Dans l'Amazone: Sont les Donnees Historiques Pour la Verification Sur le Terrain de Teledetection Bonnes?

Resume

La region de l'Amazone a un des forets restantes les plus importantes dans le monde. Il a une grande partie du carbone du monde stocke dan la biomasse de vegetation, mais les quantites exactes de carbone et l'evaluation de la region comme puits ou source de carbone est raison pour une discussion scientifique. La premiere etape dans comprehension du comportement de biomasse est d'avoir une mesure precise de la biomasse. Dans cet article, j'analyse comment les releves de biomasse sur le terrain sont faites et discute des problemes possibles avec ce genre de donnees. Je discute aussi sa convenance en tant que verification sur le terrain de teledetection et fais le point de droit pour la teledetection comme outil pour l'elargissement des donnees opportunes et/ou specifiques aux emplacements.

FOCUS ON NATURE[TM] by Rochelle Mason Insight into the lives of animals

Walking along the muddy bank in search of a meal, an Iriomote mountain cat (Mayailurus iriomotensis) spots movement up ahead. He slinks forward eyeing a mudskipper on a mangrove root. As he pounces the fish drops into the shallow water. Undaunted, he follows the wiggling body and is rewarded for his small effort. The size of a housecat, this crepuscular feline has short legs and a short, bushy tail and is an excellent swimmer and tree-climber. Most of the prey species (mammals, birds, reptiles and fish) here on Iriomotejima (Japanese island off Taiwan) cannot escape the stealth and agility of this cunning carnivore. Following another urge, he howls for a female in estrus. She must be somewhere within his territory that ranges from the beach into the subtropical forest and farmland. A soft reply is heard.

Artwork and text by Rochelle Mason

Copyright 2000-2003

(808) 985-7311

rmason@rmasonfinearts.com
Table 1. Total biomass for
the Brazilian Amazon

 TB (Mg ha-1)

Fearnside (1992) 272
Brown and Lugo (1992) 1950's 227
Brown and Lugo (1992) Small Samples 414
Brown and Lugo (1992) RADAMBRASIL 227
Brown and Lugo (1992) corrected 300
 Mean 288
 Standard deviation 69

Table 2. Total biomass for
selected areas in Tapajos
and Manaus

 TB (Mg ha-1)

Keller et al. (2001) Tapajos 372
Fearnside (1992) Manaus 235
Lucas et al. (1996) Manaus 362
 Mean 323
 Standard deviation 62

Table 3. Above ground live
biomass for selected areas

 AGLB (Mg ha-1)

Keller et al. (2001) Tapajos 282
Luckman et al. (1997) Tapajos 284
Laurance et al. (1999) Manaus 356
Carvalho et al. (1995) Manaus 399.3
 Mean 330
 Standard deviation 50

Table 4. Mean AGLB for 34 sites
in Amazonia grouped according
to plot size

 AGLB (Mg ha-1)

Houghton et al. (2001) plots > 5 ha 241
Houghton et al. (2001) 0.5 < plots < 5 ha 290
Houghton et al. (2001) plots < 0.5 ha 339
 Mean 290
 Standard deviation 40


Literature Cited

Brown S., Lugo A.E. 1992. Aboveground biomass estimates for tropical moist forests of the Brazilian Amazon. Interciencia 17:8-18.

Brown S., Lugo A.E. 1992b. Biomass of Brazilian Amazon forests: the need for good science. Interciencia 17:201-203.

Carvalho Jr. J.A., Santos J.M., Santos J.C., Leitao M.M. 1995. A tropical rainforest clearing experiment by biomass burning in the Manaus region. Atmospheric Environment 29(17): 2301-2309.

Clark D.A., Brown S., Kicklighter D.W., Chambers J.Q., Thomlinson J.R., Ni J., Holland E.A. 2001. Net primary production in tropical forests: an evaluation and synthesis of existing field data. Ecological Applications 11(2):371-384.

Clark D.A. 2002. Are tropical forests an important carbon sink? Reanalysis of the long-term plot data. Ecological Applications 12(1):3-7.

Fearnside P.M. 1992. Forest biomass in Brazilian Amazonia: comments on the estimate by Brown and Lugo. Interciencia 17:19-27.

Honzak M., Lucas R.M., Amaral I., Curran P.J., Foody G.M., Amaral S. 1996. Estimation of the leaf area index and total biomass of tropical regenerating forests: comparison of methodologies. Pp 365-381 in Gash J., ed. Amazonian deforestation and climate. John Wiley & Sons, Chichester.

Houghton R.A., Lawrence K.T., Hackler J.L., Brown S. 2001. The spatial distribution of forest biomass in the Brazilian Amazon: a comparison of estimates. Global Change Biology 7:731-746.

Keller M., Palace M., Hurtt G. 2001. Biomass estimation in the Tapajos National Forest, Brazil: examination of sampling and allometric uncertainties. Forest Ecology and Management 154:371-382.

Laurance F.L., Fearnside P.M., Laurance S.G., Delamonica P., Lovejoy T.E., Rankin de Merona J.M., Chambers J.Q., Gascon C. 1999. Relationship between soils and Amazon forest biomass: a landscape-scale study. Forest Ecology and Management 118:127-138.

Lucas R.M., Curran P.J., Honzak M., Foody G.M., Amaral I., Amaral S. 1996. Disturbance and recovery of tropical forests: balancing the carbon account. Pp 383-398 in Gash J., ed. Amazonian deforestation and climate. John Wiley & Sons, Chichester.

Luckman A., Baker J., Kuplich T.M., Yanasse C.C., Frery A.C. 1997. A study of the relationship between radar backscatter and regenerating tropical forest biomass for spaceborn SAR instruments. Remote Sensing of the Environment 60:1-13.

Nelson B.W., Mesquita R., Pereira J.L., Souza S.G., Batista G.T., Couto L.B. 1999. Allometric regressions for improved estimate of secondary forest biomass in the central Amazon. Forest Ecology and Management 117:149-167.

Phillips O.L., Malhi Y., Higuchi N., Laurance W.F., Nunez Vargas P., Vasquez Martinez R., Laurance S., Ferreira L.V., Stern M., Brown S., Grace J. 1998. Changes in the carbon balance of tropical forests: evidence from long-term plots. Science 282:439-442.

Phillips O.L., Malhi Y., Vinceti B., Baker T., Lewis S., Higuchi N., Laurance W.F., Nunez Vargas P., Vasquez Martinez R., Laurance S., Ferreira L.V., Stem M., Brown S., Grace J. 2002. Changes in growth of tropical forests: evaluating potential biases. Ecological Applications 12(2):576-587.

Oton Barros

Instituto Nacional de

Pesquisas Espaciais

School of Natural Resources and the Environment

University of Michigan

430 E. University

Ann Arbor. MI 48109-1115

obarros@umich.edu

Oton Barros earned a degree in Agronomics Engineering from the Universidade Federal do Amazonas in 1987. In 1992 he received a Master in Science in Remote Sensing from the Instituto Nacional de Pesquisas Espaciais, Brazil. He is a researcher at the Brazilian Space Agency (INPE) where he is studying agriculture and forestry using remote sensing techniques. Since 1999 he has been working on a INPE/University of Michigan joint project called "Radar Remote Sensing of Land Cover and Biomass in the Amazon." He is currently a PhD student at the School of Natural Resources and Environment at the University of Michigan.
COPYRIGHT 2003 University of Michigan, School of Natural Resources
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Barros, Oton
Publication:Endangered Species Update
Date:Nov 1, 2003
Words:3475
Previous Article:The Whaling Season: an Inside Account Of The Struggle to Stop Commercial Whaling: Kieran Mulvaney Island Press 2003.
Next Article:Making Sense of Intractable Environmental Conflicts: Concepts and Cases.
Topics:


Related Articles
A chronicle of change: globe-watching encyclopedia pools diverse world views.
Biomass burning ignites concern.
Remote Sensing and Human Health: New Sensors and New Opportunities.
Accuracy assessment of vegetation and background fraction analysis through NDVI and image degradation. (South Carolina Academy of Sciences Abstracts).
Field methods in remote sensing.
The Effects of Land Use Changes on Forest Fires in Major Islands of Indonesia.
From sky sensors to forest floors.
Gastropod abundance and biomass relationships with salt marsh vegetation within ocean-dominated South Carolina, USA estuaries.
Bulletin of the Forestry and Forest Products Research Institute: Patterns of Slash-and-Burn Land Use and Their Effects on Forest...

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters