Dmitry Schepaschenko1*, Linda See1, Steffen Fritz1, Ian McCallum1, Christian Schill2, Christoph Perger1,3, Alessandro Baccini4, Heinz Gallaun5, Georg Kindermann1, Florian Kraxner1, Sassan Saatchi6, Michael Obersteiner1, Maurizio Santoro7, Christiana Schmullius8, Anatoly Shvidenko1, Maria Schepaschenko9
1 ESM Program, International Institute for Applied Systems Analysis, Austria
2 Felis, University of Freiburg, Germany
4 The Woods Hole Research Center, Falmouth, MA, USA.
5 Joanneum Research, Graz, Austria.
6 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA.
7 GAMMA Remote Sensing Research and Consulting AG, GÌ_mligen, Switzerland.
8 Friedrich Schiller University, Jena, Germany.
9 Russian Institute of Continuous Education in Forestry, Pushkino, Russia
Terrestrial biomass is considered as an essential indicator for the monitoring of the Earth’s ecosystem and climate. In recent years, many regional biomass datasets have been produced. These were obtained using a wide range of methods — from pure remote sensing RS to the collection of field measurements. The Biomass Geo-Wiki is a new tool from the family of Geo-Wiki.org, which has been launched to bring together different biomass datasets so that they can be viewed and compared with high resolution imagery on Google Earth. The ultimate goal is to perform gap analysis, cross-product validation, harmonization and hybrid product development leading to improved global biomass datasets in the future.
Accurate estimates of terrestrial biomass and its dynamics are crucial for a wide variety of applications. For example, biomass is an important indicator of land use and land use change, and it provides one critical input to the development of climate change mitigation policies such as REDD+. Carbon budgeting is another important area that requires biomass monitoring. One more point of recent global interest has been biofuels and the food vs. fuel debate [e.g. 10]. These are just a few examples that illustrate the importance of biomass datasets.
Remote sensing is one of the most effective ways to monitor land resources in a timely manner, in particular the monitoring of vegetation productivity, carbon stocks and land use dynamics (e.g. deforestation). However, many of the spatial datasets produced show discrepancies when compared with one another, and they have been subject to limited validation, which remains the most serious problem that requires attention. More recently, new opportunities  have arisen to collect additional spatial information via the Internet and mobile devices, which could aid in the validation process.
Geo-wiki was originally a single crowd-sourcing system for land cover validation [6, 7] but has now evolved into a family of tools with different branches. Biomass is only one branch along with agriculture, urban, human impact, competition and many others. The core datasets in Geo-Wiki are registered in the geo-portal and it represents one of many ongoing GEO activities, which contribute to building a Global Earth Observation Systems of Systems (GEOSS) . Following the principles of GEOSS, Biomass Geo-Wiki provides a broad range of information on biomass to a wide variety of users in the form of global, regional and local biomass datasets. In addition to visualization, volunteers are encouraged to validate datasets using Google Earth as well as share their own data, measurements and opinions.
Biomass Geo-Wiki content and tools
The design of Geo-wiki follows that of a standards-based geospatial portal as outlined by the Open Geospatial Consortium, with portal, portrayal, data and catalog services. The architecture of Geo-Wiki is based on well-known, mainly open-source components and technologies (Figure 1).
Biomass Geo-Wiki presents a collection of global, regional and in-situ biomass datasets produced by a number of institutions, which are overlaid on the Google Earth platform (Figure 2).
The main Biomass Geo-Wiki interface contains a toolbar and a list of available datasets as shown on the left side of Figure 2. The main display shows a Google Earth image on which different thematic layers can be draped and displayed with a legend. All of the datasets have been standardized to common units, and a consistent color scheme has been applied to facilitate visual comparison.
The full list of datasets is presented in Table 1. The datasets are grouped into 3 blocks: above ground live biomass, forest woody biomass and in-situ forest biomass measurements. They span spatial scales from global to national to regional as well as down to the fine local scale as plot level measurements.
|Table 1. List of datasets currently available in Biomass Geo-Wiki|
|Region||Base year||Resolution||Provider||Object, units||Methods|
|Above Ground Live Biomass|
|Global||2005||30 arc min||IIASA||Forest, Mg dm/ha||FAO data downscaled by RS |
|Europe||2005||1 km||IIASA||Forest, Mg dm/ha||FAO data downscaled by RS and forest map (JRC, Corine, GLC2000, GlobCover) |
|Russia||2009||1 km||IIASA||Vegetation, Mg dm/ha||National forest & land statistics downscaled by RS and supplementary material |
|USA||2000||30 m||WHRC||Vegetation, Mg dm/ha||Landsat+ Forest Inventory|
|Tropics||2000||1 km||NASA||Forest, Mg dm/ha||LiDAR (GLAS) + spatial imagery (MODIS, SRTM, QSCAT) + Inventory plots |
|Tropics||2010||500 m||WHRC||Forest, Mg dm/ha||LiDAR (GLAS) + MODIS + Inventory plots |
|Forest Woody Biomass|
|Tropical Africa||2003||1 km||WHRC||Woody Biomass, Mg dm/ha||MODIS + field measurements |
|Europe||2003||1 km||Joanneum Research||Growing Stock, m3/ha||MODIS + forest inventory |
|Russia||2009||1 km||IIASA||Growing Stock, m3/ha||National forest statistics downscaled by RS and supplementary material |
|Sweden||2005||0.01å¡||Gamma Remote Sensing, Friedrich Schiller Univ.||Growing stock, m3/ha||ENVISAT ASAR, MODIS VCF |
|In Situ Forest Biomass Measurements|
|Norther Eurasia||1952-2007||Sample plots (3500)||IIASA design||Tree height, m; Above ground live biomass, Stem biomass, Coarse woody debris, t C/ha||Destructive sample tree measurements |
Figure 3 shows a database of in-situ biomass measurements that can be displayed by clicking on the points displayed on top of Google Earth.
The Geo-Wiki toolbar includes features which provide information that can be queried at the pixel level and for comparing datasets.
The info button (Figure 4) provides biomass information from the different datasets available on Biomass Geo-Wiki for the pixel shown together with the Google Earth image. It is clear from this example that a wide spectrum of biomass assessments (from 21 to 340 Mg/Ha) exists for this particular location in Sweden.
In addition to finding out which biomass dataset is closest to reality, users can load geo-tagged field photos (e.g. from Panoramio.com, Confluence.org, or from research projects such as the Global Monitoring for Food Security Project ÛÒ GMFS). These photos can be used as additional information to help determine what the landscape is in particular pixels. The photos are especially important when Google Earth provides lower resolution imagery in certain areas.
Another interesting tool built into Geo-Wiki is the Normalized Difference Vegetation Index (NDVI) profiler. For any pixel on the Earth’s surface, a five year average NDVI profile is displayed. NDVI can be used to examine vegetation content so a high NDVI indicates the presence of the green part of vegetation and helps to distinguish, for example, between deciduous/evergreen forests or grassland and arable land. Deciduous trees lose their leaves during the seasons so the NDVI profile will drop at certain times of the year corresponding to when the foliage is shed. Harvest and tillage will also show patterns of NDVI falling throughout the year.
While a great deal of work has gone into producing biomass datasets in recent years, there is a need to begin harmonizing these efforts. Existing products do not yet provide a consensus on the spatial distribution or the amount of biomass. This disagreement becomes very apparent when the different products are compared with one another. One product might show a high biomass value while another product suggests a low or no biomass pool. The availability of different datasets is confusing to the users of these products who have no idea which product is the best one to choose for their particular application nor is there much guidance available to aid their choice. Another issue is independent validation, which is a crucial task. Consequently, the main drivers behind the development of are on the one hand: the large amount of disagreement between the different biomass maps but also the lack of good validation data.
We overlaid two recent biomass dataset for the tropics and calculated where they disagree and by how much. The map in Figure 5 shows where more biomass appears in the NASA  product and where more is represented in the WHRC  one. There is a reasonable amount of disagreement on the map as shown by the dark red and dark blue colors.
Biomass Geo-Wiki provides an opportunity to compare these two datasets with imagery from Google Earth. We cannot say that one biomass dataset is better than another one. Both datasets have strengths and weaknesses. Figure 6 shows two examples of disagreement or false estimation.
From Figure 6.1, one can see that the WHRC dataset gives 120 t/ha biomass for a water dominant pixel and 160 for a settlement while the NASA dataset demonstrates a better recognition (30 and 90 t/ha respectively) in spite of the coarse resolution. Figure 5.2 demonstrates a spatial shift (of 2-7 km) for this area. This is one kind of error that can be picked up easily using crowd-sourcing. Volunteers can also directly validate the datasets on Biomass Geo-Wiki as explained below.
Case Study: Biomass Validation
With an ever increasing amount of very fine spatial resolution images available on Google Earth, it is becoming possible for every Internet user (including non-remote sensing experts) to distinguish land cover features with a high degree of reliability. Such an approach is inexpensive and allows Internet users from any region of the world to get involved in this global validation exercise.
Currently in Biomass Geo-Wiki.org, volunteers have the ability to view information from all biomass datasets available for any specific place (Figure 7) overlaid on Google Earth imagery. Volunteers are then asked to decide if the biomass products correctly capture what they see or know to exist on the ground. High resolution images can usually distinguish between forest/non forest vegetation, and canopy features. This information is often enough to judge the disagreement between datasets. In addition, it is possible to recommend a land cover class (i.e. select from a list of possible land cover types, and upload available photos to support the decision).
For the blue rectangle shown in Figure 7, we see a mature forest (wide crown with long shadow on the forest edge) with about 20% grassland. Different datasets (see Table 1 for details) suggest between 21 and 340 Mg biomass per ha. As a validator we see that the highest estimation is likely to be correct for the dense forest and the lowest values represent the non-forest vegetation. Observed land cover structure (80% covered by forest) leads us to the conclusion that the biomass dataset based on JRC forest map  (241-250 Mg dm/ha) is the most appropriate for the chosen pixel.
All information entered by volunteers is recorded in a publically available spatial database. This validation database contains a land cover type and biomass stock value, and can be used in the future to create an improved hybrid dataset.
The core Geo-Wiki, which is devoted to land cover validation, has collected more than 100K validation points. This information is also useful for biomass dataset validation and for producing a hybrid product.
The next major development in Biomass Geo-Wiki is a mobile application that can measure forest biomass. Modern smartphones have good enough gyro sensors to measure angles. Applications are already available in the marketplace for measuring tree height. Implementation of allometric functions can turn the phone into a biomass measuring instrument. These measurements would be cost effective, even for a non-experienced user. The results can be automatically uploaded to Biomass Geo-Wiki together with pictures and will be stored in the database.
A longer term goal is to perform various scientific tasks once further datasets are assembled. These include:
1. Gap analysis: To identify places where disagreement between products is highest;
2. Cross-product validation: To use in-situ data from forest inventory plots to determine which dataset is best at which location or data collected on the ground from Geo-Wiki-mobile for biomass or eyeball estimates based on Google earth imagery;
3. Possible harmonization and hybrid product development: Based on the best performing biomass maps when compared to validation points and in-situ information the best maps can be identified for certain locations and a global mosaic can be created. More sophisticated statistical approaches can be used to fuse the different biomass maps and the validation points collected.
Biomass Geo-Wiki will ultimately lead to the provision of better information on global biomass, which is a necessary component to the development of national and international level REDD+ policies. Biomass inputs are also critical to Monitoring, Reporting, and Verification (MRV) frameworks of carbon inventories and will therefore provide a valuable resource for national governments in the future.
Furthermore, this tool could potentially provide the necessary scientific platform to enhance collaboration in the area of global biomass monitoring including exchange of methodologies and datasets, working towards an open and fully transparent environment. Such a collaboration tool could also substantially contribute to different important ecological studies, e.g. to studying major biogeochemical cycles, biodiversity, deforestation, and others, and it could be used as a unique educational tool, which clearly demonstrates the discrepancies and difficulties that exist in global biomass monitoring while simultaneously providing input to improve the current situation.
1. Baccini, A., Laporte, N., Goetz, S.J., et al. (2008). A first map of tropical Africa’s above-ground biomass derived from satellite imagery. Environmental Research Letters. 3(4), 045011.
2. Baccini, A., Goetz, S.J., Walker, W.S., et al. (2012). Estimated carbon dioxide emissions from tropical deforestation improved by carbon-density maps. Nature Clim. Change. 2(3), 182-185.
3. BÌ¦tcher, H., Kindermann, G.E., Fritz, S., et al. (2012). European forest biomass maps – comparison and validation. In preparation.
4. Christian, E. (2005). Planning for the Global Earth Observation System of Systems (GEOSS). Space Policy. 21(2), 105-109.
5. Doan, A., Ramakrishnan, R., & Halevy, A.Y. (2011). Crowdsourcing systems on the World-Wide Web. Communications of the ACM. 54(4), 86-96.
6. Fritz, S., McCallum, I., Schill, C., et al. (2009). Geo-Wiki.Org: The Use of Crowdsourcing to Improve Global Land Cover. Remote Sensing. 1(3), 345-354.
7. Fritz, S., McCallum, I., Schill, C., et al. (2012). Geo-Wiki: An online platform for improving global land cover. Environmental Modelling & Software. 31, 110-123.
8. Gallaun, H., Zanchi, G., Nabuurs, G.J., et al. (2010). EU-wide maps of growing stock and above-ground biomass in forests based on remote sensing and field measurements. Forest Ecology and Management. 260(3), 252-261.
9. Kindermann, G.E., McCallum, I., Fritz, S., et al. (2008). A global forest growing stock, biomass and carbon map based on FAO statistics. Silva Fennica. 42(3), 387-396.
10. Pimentel, D., Marklein, A., Toth, M.A., et al. (2009). Food Versus Biofuels: Environmental and Economic Costs. Human Ecology. 37(1), 1-12.
11. Saatchi, S.S., Harris, N.L., Brown, S., et al. (2011). Benchmark map of forest carbon stocks in tropical regions across three continents. PNAS. 108(24), 9899-9904.
12. Santoro, M., Beer, C., Cartus, O., et al. (2011). Retrieval of growing stock volume in boreal forest using hyper-temporal series of Envisat ASAR ScanSAR backscatter measurements. Remote Sensing of Environment. 115(2), 490-507.
13. Schepaschenko, D., McCallum, I., Shvidenko, A., et al. (2011). A new hybrid land cover dataset for Russia: a methodology for integrating statistics, remote sensing and in-situ information. Journal of Land Use Science. 6(4), 245-259.
14. Shvidenko, A., Schepaschenko, D., McCallum, I., et al. (2007). Russian Forests and Forestry. [cited 01.02.2012]; Available from: http://www.iiasa.ac.at/Research/FOR/forest_cdrom.
Dmitry Schepaschenko works for Ecosystems Services and Management Program at the International Institute for Applied Systems Analysis (IIASA). He graduated in forest sciences from Moscow State Forest University (MSFU) and received a PhD degree in soil science from Dokuchaev Soil Science Institute in Moscow. He was awarded a university professor degree in ecology from MSFU. Schepaschenko is a member of the Academic Council of MSFU.