Biomass estimation of mixed forest landscape using a Fourier transform texture-based approach on very-high-resolution optical satellite imagery
The estimation and monitoring of forest biomass from satellite data is of vital importance for efforts to preserve forests using their carbon value, but remains a technical challenge. We have a paper out (led by former MPhil student Jojo Singh) in the International Journal of Remote Sensing that explores if the texture of a high resolution satellite image can yield information about biomass, by capturing the dimensions of the crowns of the dominant trees. We apply this to the SAFE field site in Sabah, Malaysia, in a heterogeneous landscape of old growth forests, logged forests of varying intensity, and oil palm plantations. The paper shows that such Fourier transform approaches do a decent job at estimating and mapping biomass.
Assessment of forest structure parameters via remote-sensing data offers the opportu- nity to examine stand parameters and to detect degradation and forest dynamics, such as above-ground biomass (AGB), at the landscape scale. While much attention has focused on spectrum-based and radar backscatter approaches for assessing forest biomass, texture-based approaches show strong promise. This work makes use of the novel Fourier transform textural ordination (FOTO) method, which involves the combination of 2D fast Fourier transform (FFT) and ordination through principal component analysis (PCA) for characterizing the structural and textural properties of vegetation. This technique presents the potential of Fourier transform approaches in estimating the different forest types, their stand structure, and biomass dynamics in the context of an oil palm–tropical forest landscape in Sabah, Malaysian Borneo. The method was applied to the recordings of very-high-resolution (VHR) Satellite Pour l’Observation de la Terre (SPOT) imagery of the study area. The technique proved useful in distinguishing between the forest types and developing individual biomass estimate models for various forest types. Results show that the FOTO method is able correctly to resolve high AGB values of various forest types. These findings are in agreement with the results based on ground measurements.
Singh M., Malhi Y. Bhagwat S. (2014) Biomass estimation of mixed forest landscape using a Fourier transform texture-based approach on very-high-resolution optical satellite imagery, International Journal of Remote Sensing, 35, 9
We have a new paper in Global Ecology and Biogeography that explores how satellite maps of the biomass of Amazonian forests compare to our extensive ground inventory network RAINFOR.
Mitchard E.T.A. et al. (2014) Markedly divergent estimates of Amazon forest carbon density from ground plots and satellites. Global Ecology and Biogeography, DOI: 10.1111/geb.12168
The accurate mapping of forest carbon stocks is essential for understanding the global carbon cycle, for assessing emissions from deforestation, and for rational land-use planning. Satellite remote sensing maps are currently a key tool for this purpose, but remote sensing does not estimate vegetation biomass directly, and thus may miss significant spatial variations in forest structure. We test the stated accuracy of pantropical carbon maps using a large independent field dataset, the RAINFOR forest inventory dataset.
Two recent pantropical remote sesnsing maps of vegetation carbon (from Saatchi et al. and Baccini et al.) were compared to the RAINFOR ground-plot dataset, involving tree measurements in 413 inventory plots located in nine countries. The remote-sening maps were compared directly to field plots, and kriging of the field data was used to allow area-based comparisons.
The study finds that the two remote sensing carbon maps fail to capture the main gradient in Amazon forest biomass, from the densely wooded tall forests of the north-east, to the light-wooded, shorter forests of the south-west. The differences between plots and RS maps far exceed the uncertainties given in these studies, with whole regions over- or under-estimated by > 25%, whereas regional uncertainties for the maps were reported to be < 5%.
Pantropical biomass maps are widely used by governments and by projects aiming to reduce deforestation using carbon offsets, but we conclude that they may have significant regional biases. Carbon-mapping techniques must be revised to account for the known ecological variation in tree wood density and allometry to create maps suitable for carbon accounting. The use of single relationships between tree canopy height and above-ground biomass inevitably yields large, spatially correlated errors. This presents a significant challenge to both the forest conservation and remote sensing communities, because neither wood density nor species assemblages can be reliably mapped from space.
The permanent archive of the field plot data can be accessed at: http://dx.doi.org/10.5521/FORESTPLOTS.NET/ 2014_1
Last week I visited the second of our study sites in the ECOLIMITS project, funded by the ESPA (Ecosystem Services and Poverty Alleviation) programme of NERC and DfID. The project is looking at high different intensities of small-holder land use (cocoa farms in Ghana and coffee farms in Ethiopia) affect biodiversity, ecosystem services and poverty alleviation. Are there benefits to farming near forests or high tree cover, e.g. through increased dew fall in the dry season, or pollination or pest control? Are there ecosystem limits, where farming too intensively causes ecosystem feedbacks that enhance poverty rather than diminish it?
Cocoa originated in Amazonia, but Ghana now produces almost all of the chocolate eaten in Britain, and is the world's second largest producer after neighbouring Côte d'Ivoire. It contributes substantially to the income of rural Ghana, and cocoa production is enmeshed in a complex quilt of traditional land ownership, tenant farmers, migrants and national cocoa policies. A well-managed, low shade cocoa farm can be highly productive, but poor in biodiversity. A shadier, less intensively managed cocoa farm may be less productive in the short term, but more sustainable in the longer term and richer in biodiversity.
Our work is focussed in the communities, cocoa farms and forests in and around Kakum National Park. We spent a very enjoyable few days there, selecting study sites and introducing our project to local communities. Everywhere we went the communities were welcoming, gracious and co-operative. Rural Ghana is such a welcoming landscape to wander around in.
How biogeographical history determines what termites are found where: first comparison of quantitative estimates of termite biomass and abundance reveals strong intercontinental differences
We have a nice new paper led by DPhil student Cecilia Dahlsjö, looking at the abundance, biomass and phylogeny of termites across the tropics, comparing three sites in Peru, Cameroon and Malaysia. Termites are social insects (essentially evolved and modified cockroaches) completely unrelated to other social insects such as ants.
The paper shows that there are large differences in termite biomass, function and diversity between the three ytopical continents driven by biogeographical history. Africa has much higher termite biomass, dominated by soil feeding termites, while the highest biomass of wood-feeding termites was in Asia. Wood feeding termites dominate in biomass in both Asia and South America. The unique fungus-growing "farming" termites are very important food wood descomposition in Africa and Asia, yet completely absent from South America ,where leaf-cutter ants have a similar fungus'growing "agriculture". The likely causes for these differences are biogeographical history.
The Termitidae probably evolved and dispersed from Africa with higher dispersal rates for wood- feeding termites compared with soil-feeding termites. Soil-feeding termites are probably more dependent on the buffered environment within tropical rain forest, which offer less fluctuation in temperature and moisture compared with non-forest habitats. Therefore, arid and semi-arid habitats such as woodland, savanna and desert would be more likely to be barriers to dispersal for those groups. Wood-feeding termites on the other hand (with the probable exception of fungus-
growing termites) are more likely to have been able to disperse across different types of barrier, including bodies of water, because of their rafting abilities. These activities are probably dependent on their harder and thicker exoskeletons, which support their foraging behaviour and therefore the likelihood of allowing dispersal across suboptimal habitat types. The biogeographical patterns of wood-feeding termites may therefore have evolved through a wider range of dispersing types than soil- feeding termites, which may have had relatively low dispersal rates. However, it seems likely that soil- feeding termites may have evolved independently from rafting wood-feeding termites on several occasions across the continents leading to the relatively depauperate humus- and soil-feeder assemblages of the non-African sites.
Macrotermitinae (fungus-growing termites) seem to have evolved in African tropical forests and although four dispersal events from Africa to South-East Asia appear to have taken place, fungus-growing termites have failed to colonize South America. Macrotermitinae are poor dispersers, as they depend on their close mutualistic relationship with Termitomyces, a fungus, which breaks down organic matter within the termite mound . Generally termites depend on the presence of the reproductive caste (king and queen), however, the species of Macrotermitinae also need spores of Termitomyces to be available in their new environment. The successful colonization of Asia but their failure to inhabit South America contributes to the diversity anomaly of functional groups across the equatorial regions.
Dahlsjö C.A.L, Catherine L. Parr, Yadvinder Malhi, Homathevi Rahman, Patrick Meir, David T. Jones and Paul Eggleton (2014). First comparison of quantitative estimates of termite biomass and abundance reveals strong intercontinental differences . Journal of Tropical Ecology, 30, pp 143-152 doi:10.1017/S0266467413000898
While Nova Xavantina, Mato Grosso, Brazil , last week I gave a couple of lectures, one on "the lost elephants of the Brazilian cerrado", freshly inspired by our recent megafauna conference. I pointed out how recently huge elephants roamed the landscapes of Brazil, and how different the vegetation and its ecology probably was then (elephants love destroying trees), and speculating whether the closed woody "cerrado" savannas of Brazil are a legacy of megafaunal extinction 10-15,000 years ago, and what "ghosts" of the elephants remain in the ecology of the region today. Somebody brought in a few bones that had been found in a farm in the region. The fact these were BONES, not lithified fossils, really brought it home how recent this was. We made a little science video summarising the talk, which can be seen here:
The lost elephants of Brazil (click here)
This week saw the start of the second chapter of our multi-year plan to understand the variation of plant biotic attributes (traits) along various transects in the tropics. Last year saw CHAMBASA in Peru (Challenging Attempt to Measure Biotic Attributes Along the Slope of the Andes). The next few months are dominated by BACABA (Biotic Attributes at the Cerrado-Amazonia Boundary. Bacaba is the name of a characteristic local palm (Oenocarpus bacaba) and also the name of the area where much of this work is taking place. The focal area is around the small town on Nova Xavantina in Mato Grosso, Brazil.
The work is being implemented by the lab team of Drs Ben Hur and Bia Marimon, and on the international sites by Dr Imma Oliveras (Wageningen/Oxford).
The global activity is centrally supported by my award from the European Research Council, GEM-TRAITS.
Yadvinder Malhi is an ecosytem ecologist and Professor of Ecosystem Science at Oxford University