Professional Documents
Culture Documents
09527
09527
09527
attached copy is provided by Elsevier for the author’s benefit and for the benefit of the author’s institution, for
non-commercial research and educational use including without limitation use in instruction at your institution,
sending it to specific colleagues who you know, and providing a copy to your institution’s administrator.
All other uses, reproduction and distribution, including without limitation commercial reprints, selling or licensing
copies or access, or posting on open internet sites, your personal or institution’s website or repository, are
prohibited. For exceptions, permission may be sought for such use through Elsevier’s permissions site at:
http://www.elsevier.com/locate/permissionusematerial
Davis C., and Petersen R , Tools for Monitoring Global Deforestation, Reference Module in Earth Systems and
Environmental Sciences, Elsevier, 2016. 21-Jan-2016 doi: 10.1016/B978-0-12-409548-9.09527-0.
Background 1
Needs for Global-Scale Forest Monitoring 1
The Challenge of Global-Scale Forest Monitoring 2
Defining Forest – What Are We Monitoring? 2
Current Tools for Monitoring Changes in Forest Area 2
Ground-Based 2
Satellite-Based 3
Airborne 3
Global-Scale Forest Monitoring Initiatives 4
The Global Forest Resources Assessment 4
Global Forest Watch 4
Limitations 6
Policy Applications for Global Forest Monitoring Data 6
Climate Change and REDD + 6
Private Sector Sustainability Commitments 6
Combating Illegal Logging and Associated Trade 6
Global Policy Commitments 7
The Next Frontier of Forest Monitoring – Closing Knowledge Gaps 7
Building Monitoring Capacity in the Tropics 7
Improving the Accuracy and Precision of Remote Sensing Approaches 7
Uniting Ground-Based and Satellite-Based Approaches 7
Conclusions 8
References 8
Background
Forests provide critical services to the globe. They contribute to the livelihoods of around 1.6 billion rural poor worldwide and
provide home to no fewer than 300 million people, but the products and services that forests provide – cleaning our air and water,
helping to prevent soil erosion, and regulating the climate – are essential to every aspect of life (World Bank, 2004). Globally,
Costanza et al. (1997) estimate the total annual value of forest ecosystem goods and services at $4.7 trillion.
Forests are critical for many global and transboundary environmental factors. For example, deforestation and forest degradation
account for 12–15% of current anthropogenic emissions – the second-largest source of CO2 after fossil fuel combustion (Van der
Werf et al., 2009). Recent science has also linked deforestation to drought, as forests are intricately connected with the movement of
moisture in the atmosphere and levels of rainfall (Nobre, 2014; Sheil and Murdiyarso, 2009).
Currently, there are three main types of tools used for forest monitoring: satellite-based, ground-based, and airborne.
As mentioned, in order to monitor forest change at a global level, these tools need to enable methods that are precise, accurate,
repeatable, scalable, and affordable. They should also be consistent, allowing for comparison of results from different geographies.
Below, we review each method and its relative advantages and disadvantages for monitoring forest changes.
Ground-Based
The oldest way humans have monitored forests is by documenting their characteristics from the field. This remains one of the most
detailed and accurate ways to capture changes occurring in forests. For example, many governments conduct structured forest
inventories, where ecologists and researchers venture into forests to document detailed information about their structure (canopy
density, tree height, diameter at breast height, or DBH), tree species, biodiversity, and other qualities such as the soil carbon content
or seedling density. Field inventories are the only way to currently capture such accurate, detailed information about the physical
qualities and biodiversity values of individual forests.
The advent of geographic positioning systems technology has enabled ground patrols and forest inventories to be much more
spatially precise, as patrollers, forest managers, and ecologists can record more detailed coordinate points of the observations they
collect and upload those data into a geographic information system for systematic visualization, analysis, and modeling.
However, ground-based tools pose several challenges for global-scale forest monitoring. Firstly, they do not scale geographi-
cally; these efforts are very time- and resource-intensive, and are often conducted at the scale of individual forest management units,
protected areas, or only in sampled points across a country. Secondly, forest inventories are often conducted only periodically –
such as every 5 years – and it is difficult to guarantee consistency across time in these methods. Finally, the remote nature of many
forests, particularly in the tropics, makes it difficult to cover some of the most important remaining forests of the world. While field
inventories have decent coverage of the temperate and boreal zones, much of the tropics remains distant from roads and is difficult
to access for forest inventories.
While these challenges render ground-based tools unsuitable for wholescale global forest monitoring, they remain important
inputs for calibrating and validating (ground-truthing) remotely sensed data on forests, as discussed below, as well as for
conducting detailed studies of the ecological values of specific forests, such as for environmental impact assessments.
Satellite-Based
One of the most accurate ways to detect deforestation is to see it with your own eyes. However, some of the world’s most
endangered forests are hundreds of kilometers from the nearest road or city, making in-person monitoring difficult or impossible.
This is particularly true as new climatic regimes and natural phenomena such as fire bring change to remote forests around the
world. To overcome this challenge, satellite imagery is a practical necessity (Achard et al., 2010; Baker et al., 2010).
Satellite remote sensing – or the scanning of objects from space by satellites – provides the most promise for global, consistent,
efficient monitoring of the world’s forests (Chambers et al., 2007). These instruments circle the globe capturing information about
the physical properties of earth, and have enabled large-scale monitoring of the earth’s natural resources for the first time.
Earth observation satellites host sensors, which record information about the earth. Sensors are divided into two fundamental
types: passive and active. Passive sensors receive energy reflected off the earth’s surface, light that is visible in the electromagnetic
spectrum. The sensors do not emit energy themselves, which is why they are called ‘passive.’ ‘Active’ sensors emit radiation, which is
reflected off the earth, which is then detected and measured by the sensor. While active sensors currently populate a smaller portion
of the world’s satellite fleet, their major advantage is that many can pierce through cloud cover, which is often an impediment to
getting a clear image in tropical regions.
These sensors each have a unique temporal (how often) and spatial (how large) resolution. Remote sensing involves an inherent
tradeoff between the frequency with which images are captured, the size of the image footprint, and the resolution of the image. For
example, the latest Landsat Satellite (Landsat 8), the workhorse of the NASA fleet, circles the earth every 16 days, capturing images
at a 30 m by 30-m resolution. In contrast, the two MODIS sensors, on the NASA Aqua and Terra satellites, capture images much
more frequently (twice a day), but only at a 250-m resolution. Both of these satellite systems have been used for mapping changes
in forest cover (see e.g., Hansen et al., 2008; Jia et al., 2014).
There are two fundamental ways to use remote sensing data to assess forest area change. The first is the ‘subtract the map’
approach. In this method, maps of forest cover are made for each time interval (e.g., once per year), and then deforestation is
calculated by subtracting the current year’s forest from the previous year’s forest. This is how most governments (such as the Space
Agency in Brazil) conduct forest inventories using remote sensing data, but can amplify errors inherent in the original data
(INPE, 2013).
The second method focuses on changes within individual pixels. In this method, the individual pixel is the focus of analysis, and
all available data for all dates is collected for each pixel. These data are then assembled into a time series for each pixel, and this time
series is analyzed to detect abrupt and persistent drops in vegetation cover that indicate deforestation (e.g., Hansen et al., 2013).
One of the first global-scale efforts to map forest changes was the TREEs project of the Joint Research Centre of the European
Commission. The TREEs project began in the early 1990s with a goal of developing a forest cover assessment throughout the tropics
that could eventually become operational (Achard et al., 2001). The system utilized Advanced Very High Resolution Radiometer
data, carried by NOAA’s Polar Orbiting Environmental Satellites, whose frequency of coverage at the time made it useful for forest
monitoring applications.
However, several years previously in 1972, NASA launched Landsat, the first in a series of earth observation satellites that
captured 30-m data that would become critical for global-scale monitoring (Fagan and Defries, 2009). In decades since, the most
radical shift in this field has been economic – not technological – in nature. It occurred when in 2009, USGS opened all Landsat
imagery from 1972 to 2009 freely to the public, enabling global-scale, historical analysis of the earth for the first time. It was not
until 2013 that scientists harnessed Landsat data to create the first-ever maps of annual tree cover loss and gain at a global scale
(Hansen et al., 2013).
The application of remote sensing detection of forest changes has increased immensely, thanks to the advent of advanced
computing power to process millions of pixels, and the availability of diverse set of high-resolution imagery for validation and
calibration, such as from DigitalGlobe, SPOT, and RapidEye. However, large scale remote sensing of forests continues to face
several key challenges. For example, it is difficult to detect the degradation of forests at scale. While it is easy to accurately classify
forest/non-forest, it remains challenging to correctly identify the types of forest (primary, secondary, plantation) globally with
precision.
Airborne
Remote sensing started as cameras mounted on planes taking aerial photographs. This technique of making maps has been used
since the first flights, when photogrammetry and photointerpretation was used to carry out forest mapping and monitoring. Aerial
methods are still utilized today when the purpose of monitoring requires much more detailed information than we can see from
space. In this case, airborne instruments – sensors attached to planes, or cameras mounted on drones – are used to capture detailed
information about a specific area of forests at high resolutions. Their application ranges from private companies counting the
number of trees harvested in a plantation, to a government conducting a detailed carbon inventory of their forests.
For example, airborne light detection and ranging (LiDAR) sensors can capture detailed information about the physical
structure of forests (Asner, 2009). Airborne LiDAR can collect information at a resolution of 1 m to up to 10 cm, and is detailed
enough to see individual tree crowns and delineate tree species (Baldeck et al., 2015). This method is very well suited for calculating
detailed aboveground biomass measurements, useful in forest carbon inventories.
Small, recreational unmanned area vehicles are also increasingly popular technologies to quickly capture images of an area of
forest. These low-cost drones, with small cameras attached, can be quickly flown over areas to visually confirm deforestation and
even illegal activities (see e.g., Paneque-Gálvez et al., 2014). However, using drone imagery to monitor forests can be challenging;
the equipment is difficult to land in rough terrain or has a short battery life. The biggest barrier to utilizing drones in many countries
is receiving the necessary permits or permissions to fly these vehicles in regulated airspace.
Overall, airborne tools are very well suited for detailed mapping of small areas of interest. Their limited coverage and high costs
render them inefficient tools for systematic monitoring of forest change at scale.
Few initiatives exist to monitor forests worldwide, presumably due to the cost and logistical and political challenges associated with
global scale monitoring. Efforts to map the world’s forests are much more common, and sometimes global maps may be updated to
depict changes over time. However, monitoring implies continuous and systematic measurement that is not readily achieved by
repeated mapping. Two general approaches to global forest monitoring currently exist: those that aggregate national scale
monitoring activities and others that apply consistent global methods. These approaches are exemplified by the two major global
forest monitoring initiatives that exist today.
Table 1 Change in natural and planted forests according to the FAO forest resources assessment (2015)
2010-2015 (million ha) Global Southern Africa Africa Asia Europe North and Central America
cover change data can be overlaid with other geospatial datasets showing land cover and land use to draw additional insights into
the drivers and impacts of forest change.
These data indicate that the world is losing tree cover at a rate of 18 million hectares per year since 2001. In addition, the rate of
tree cover loss is increasing across the tropics. While Brazil and Indonesia accounted for 40% of all tropical tree cover loss in 2014,
the fastest increases in tree cover loss between 2001 and 2014 occurred in countries like Cambodia, Paraguay, and the Democratic
Republic of Congo. Tree cover loss in the temperate and boreal forest biomes is roughly equivalent in magnitude to tropical tree
cover loss, although the dynamics of loss are quite different. Whereas tree cover loss in the tropics is largely driven by conversion of
natural forests for alternative land uses such as agriculture, tree cover loss in the temperate and boreal zones generally reflects
dynamics such as natural fire and logging in the context of forest management (De Groot et al., 2013).
Limitations
At this time, neither the FRA nor GFW provides an entirely satisfactory solution for monitoring global deforestation. While the FRA
and GFW apply very different approaches, they face common challenges in two areas: scope and accuracy.
Scope pertains to what is being monitored, which also relates to the definition of key terms. The FAO’s definition of
deforestation is often criticized for omitting dynamics such as industrial scale clear-cutting of natural forests, which is considered
temporary unstocking rather than permanent land use change. Furthermore, the FAO acknowledges that the response rate from
countries regarding deforestation statistics is low and data quality is typically poor.
Meanwhile, GFW is currently monitoring tree cover loss, which is broader in scope than deforestation. Current global satellite-
based monitoring systems are not yet capable of distinguishing loss in different types of tree cover, such as natural forests, planted
forests, and tree cover associated with agricultural land use. Subsequently, tree cover loss can include vastly different change
dynamics, from deforestation of a primary forest to clearing of an oil palm plantation (Tropek et al., 2014). However, ancillary
data – such as detailed primary forest or plantations maps – can help overcome this challenge and monitor loss within different
types of forests (Margono et al., 2014).
Accuracy is a source of criticism for both initiatives (Grainger, 2008). The FRA exhibits vast differences in data quality across
countries, which arise from varying capacities and political incentives to accurately report forest change dynamics at the national
level. This results in general inconsistency in reporting of forest metrics, as countries use different timelines and methodologies, or
sometimes just fail to report, making systematic cross-country comparisons difficult.
For the satellite-based systems on GFW, global accuracy assessments mask potentially significant variability in accuracy across
different countries and forest biomes; 30-m change detection is not precise enough to detect very small-scale deforestation and
forest degradation (Tyukavina et al., 2015).
The growing demand for global forest monitoring is a direct response to increased international attention on the issue of
deforestation. A number of emerging policy frameworks and voluntary commitments require accurate, robust monitoring systems
in order to ensure compliance moving forward.
Near-real-time deforestation alerts from satellite-based monitoring systems have proven effective in strengthening law enforce-
ment activities in Brazil to reduce illegal deforestation (Assunção et al., 2013). Brazil is well known for pioneering such systems,
which include PRODES (annual, 30-m data based on Landsat) and DETER (monthly, 250-m data based on MODIS). Through
GFW, similar systems are now available with pantropical and global coverage, providing an opportunity for other countries to learn
from Brazil’s success (Wheeler et al., 2014). However, current satellite-based monitoring systems are unable to detect illegal
activities associated with selective logging due to the small scale of canopy disturbance (Asner et al., 2002).
Our ability to consistently monitor forest area changes at scale has improved significantly in the last few decades. Here are some
further advancements we can expect in the future.
Conclusions
Technological and scientific advancements of the last few decades have allowed us to understand – with greater accuracy than ever
before – where the world’s forests exist and how they are changing. However, significant challenges remain for creating operational
global monitoring systems at scale.
All global forest monitoring systems will have tradeoffs in terms of spatial accuracy, temporal frequency. However with the
advancement of computer power and increasingly detailed satellite imagery, it might be a question of time before global forest
monitoring can provide global-scale but locally relevant data on precisely what types of forests exist where, and how they are changing.
In the end, how accurate a forest monitoring system needs to be depends upon its application. While it is now technologically
feasible to monitor tree cover loss at a global scale, for example, these data may not be precise enough for local-level REDD
monitoring, reporting and verification. Ultimately, consumers of global monitoring data must weigh the relative strengths and
weaknesses of monitoring methodologies to find the right tool for the job.
Finally, the lack of universal definitions and a universal monitoring system elevate the need for transparent dialog around the
methods and results of the many fragmented approaches to ensure engagement and accountability around these issues.
References
Achard F, Eva H, and Mayaux P (2001) Tropical forest mapping from coarse spatial resolution satellite data: Production and accuracy assessment issues. International Journal of
Remote Sensing 22: 2741–2762.
Achard F, Stibig H-J, Eva HD, Lindquist EJ, Bouvet A, Arino O, and Mayaux P (2010) Estimating tropical deforestation from earth observation data. Carbon Management 1: 271–287.
Asner GP (2009) Tropical forest carbon assessment: Integrating satellite and airborne mapping approaches. Environmental Research Letters 4: 034009.
Asner GP, Keller M, Pereira R Jr, and Zweede JC (2002) Remote sensing of selective logging in Amazonia: Assessing limitations based on detailed field observations, Landsat ETM +,
and textural analysis. Remote Sensing of Environment 80: 483–496.
Assunção, J., Gandour, C., and Rocha, R. (2013) DETERring deforestation in the Brazilian Amazon: Environmental monitoring and law enforcement. CPI Rio, Working Paper 3.
Baccini A, Goetz SJ, Walker WS, Laporte NT, Sun M, Sulla-Menashe D, Hackler J, Beck PSA, Dubayah R, Friedl MA, Samanta S, and Houghton RA (2012) Estimated carbon dioxide
emissions from tropical deforestation improved by carbon-density maps. Nature Climate Change 2: 182–185.
Baker DJ, Richards G, Grainger A, Gonzalez P, Brown S, DeFries R, Held A, Kellndorfer J, Ndunda P, Ojima D, Skrovseth P-E, Souza C, and Stolle F (2010) Achieving forest carbon
information with higher certainty: A five-part plan. Environmental Science & Policy 13: 249–260.
Baldeck CA, Asner GP, Martin RE, Anderson CB, Knapp DE, Kellner JR, and Wright SJ (2015) Operational tree species mapping in a diverse tropical forest with airborne imaging
spectroscopy. PLoS One 10: e0118403.
Chambers JQ, Asner GP, Morton DC, Anderson LO, Saatchi SS, Espı́rito-Santo FDB, Palace M, and Souza C (2007) Regional ecosystem structure and function: Ecological insights
from remote sensing of tropical forests. Trends in Ecology & Evolution 22: 414–423.
Costanza R, d’Arge R, de Groot R, Farber S, Grasso M, Hannon B, Limburg K, Naeem S, O’Neill RV, Paruelo J, Raskin RG, Sutton P, and van den Belt M (1997) The value of the world’s
ecosystem services and natural capital. Nature 387: 253–260.
De Groot WJ, Cantin AS, Flannigan MD, Soja AJ, Gowman LM, and Newbery A (2013) A comparison of Canadian and Russian boreal forest fire regimes. Forest Ecology and
Management 294: 23–34.
Fagan M and DeFries R (2009) Measurement and monitoring of the world’s forests: A review and summary of remote sensing technical capability, 2009–2015. Washington, DC:
Resources for the Future.
FAO (2015) Global forest resources assessment 2015. Rome: United Nations Food and Agriculture Program.
Goetz SJ, Baccini A, Laporte NT, Johns T, Walker W, Kellndorfer J, Houghton RA, and Sun M (2009) Mapping and monitoring carbon stocks with satellite observations: A comparison
of methods. Carbon Balance and Management 4: 2.
Grainger A (2008) Difficulties in tracking the long-term global trend in tropical forest area. Proceedings of the National Academy of Sciences 105: 818–823.
Hammer D, Kraft R, and Wheeler D (2014) Alerts of forest disturbance from MODIS imagery. International Journal of Applied Earth Observation and Geoinformation 33: 1–9.
Hansen MC, Roy DP, Lindquist E, Adusei B, Justice CO, and Altstatt A (2008) A method for integrating MODIS and Landsat data for systematic monitoring of forest cover and change
in the Congo Basin. Remote Sensing of Environment 112: 2495–2513.
Hansen MC, Potapov PV, Moore R, Hancher M, Turubanova SA, Tyukavina A, Thau D, Stehman SV, Goetz SJ, Loveland TR, Kommareddy A, Egorov A, Chini L, Justice CO, and
Townshend JRG (2013) High-resolution global maps of 21st-century forest cover change. Science 342: 850–853.
Instituto Nacional de Pesquisa Espaciais (INPE) (2013) Metodologia para o Cálculo da Taxa Anual de Desmatamento na Amazônia Legal. São José dos Campos: INPE. Available at:
http://www.obt.inpe.br/prodes/.
Intergovernmental Panel on Climate Change (IPCC) (2002) Good practice guidance and uncertainty management in national greenhouse gas inventories” available at http://www.ipcc-
nggip.iges.or.jp/public/gp/english/index.html.
Jia K, Shunlin L, Lei Z, Xiangqin W, Yunjun Y, and Xianhong X (2014) Forest cover classification using landsat ETM+ data and time series MODIS NDVI data. International Journal of
Applied Earth Observation and Geoinformation 33(December): 32–38.
Kho LK and Jepsen MR (2015) Carbon stock of oil palm plantations and tropical forests in Malaysia: A review. Singapore Journal of Tropical Geography 36: 249–266.
Margono BA, Potapov PV, Turubanova S, Stolle F, and Hansen MC (2014) Primary forest cover loss in Indonesia over 2000–2012. Nature Climate Change 4: 730–735.
Nobre, A. D. (2014). The future climate of Amazonia, Scientific assessment report. Sponsored by CCST-INPE, INPA and ARA. São José dos Campos, Brazil, 42 p.
Paneque-Gálvez J, McCall M, Napoletano B, Wich S, and Koh L (2014) Small drones for community-based forest monitoring: An Assessment of their feasibility and potential in
tropical areas. Forests 5: 1481–1507.
Savilaakso S, et al. (2014) Systematic review of effects on biodiversity from oil palm production. Environmental Evidence 3(1): 1–21.
Sheil D and Murdiyarso D (2009) How forests attract rain: An examination of a new hypothesis. BioScience 59: 341–347.
Souza, C. M., Hayashi, S. and Verı́ssimo, A. (2009). Near real-time deforestation detection for enforcement of forest reserves in Mato Grosso. Land Governance in Support of the
Millennium Development Goals: Responding to New Challenges, Washington DC, USA, World Bank.
Tropek R, Sedláček O, Beck J, Keil P, Musilová Z, Sı́mová I, and Storch D (2014) Comment on “High-resolution global maps of 21st-century forest cover change” Science 344: 981.
Tyukavina A, Baccini A, Hansen MC, Potapov PV, Stehman SV, Houghton RA, Krylov AM, Turubanova S, and Goetz SJ (2015) Aboveground carbon loss in natural and managed
tropical forests from 2000 to 2012. Environmental Research Letters 10: 074002.
UNFCCC. (2009) Methodological guidance for activities relating to reducing emissions from deforestation and forest degradation and the role of conservation, sustainable
management of forests and enhancement of forest carbon stocks in developing countries. Decision 4/CP.15 United Framework Convention on Climate Change FCCC/CP/2009/11/
Add.1.
UNFCCC. (2013) Modalities for national forest monitoring systems. Warsaw framework for REDD-plus Decision 11/CP.19 United Framework Convention on Climate Change FCCC/
CP/ 2013/10/Add.1.
Union of Concerned Scientists. Nuclear Weapons & Global Security: UCS Satellite Database [website]. Cambridge, MA: Union of Concerned Scientists (updated 1 September 2015).
Available: http://goo.gl/wSaVJI
Van der Werf GR, Morton DC, DeFries RS, Olivier JG, Kasibhatla PS, Jackson RB, Collatz GJ, and Randerson JT (2009) CO2 emissions from forest loss. Nature Geoscience
2: 737–738.
Wheeler D, Hammer D, Kraft R, and Steele A (2014) Satellite-based forest clearing detection in the Brazilian Amazon: FORMA, DETER, and PRODES. Washington DC: World Resources
Institute.
World Bank (ed.) (2004) Sustaining forests: A development strategy. Washington, DC: The World Bank.
Further Reading
Achard F and Hansen MC (eds.) (2013) Global forest monitoring from earth observation. Boca Raton, FL: CRC Press.
Goetz, S., Hansen, M., Houghton, R. A., Walker, W., Laporte, N. T. and Busch, J. (2015). Measurement and monitoring for REDD +: The needs, current technological capabilities, and
future potential. Center for Global Development Working Paper.