336 publications from this institution
The map of Russian forest for the year 2009 is a part of Integrated Land Information System (ILIS) for Russia (Gordov et al., 2013). It contains a land cover map (Schepaschenko 2011, 2015) and associated forest data that are based on the State Forest Register (SFR). Downscaling of the SFR regional data was accomplished using a multi-sensor remote sensing approach, geographically weighted regression and reference data obtained using visual interpretation of very high resolution imagery (Schepaschenko et al., 2019). The map contains the spatial distribution of the forest parameters including the major tree species, the average age, the relative stocking, site index, and biomass conversion factors (Schepaschenko et al., 2018). The data base has spatial resolution of 5 arc second (or ca 150 m) and contains following fields: ID_FOR - unique ID of forest<br> SPEC_MOD - tree species <br> AGE - average forest age<br> SI - site index<br> RS - relative stocking (tree density)<br> GS_ha - growing stock, m3/ha<br> bef_mn - biomass expansion factor (BEF)<br> bef_sd - BEF standard deviation<br> bcef_mn - biomass conversion and expansion factor (BCEF)<br> bcef_sd - BCEF standard deviation<br> RtS_mn - root-to-shoot ratio (R:S)<br> RtS_sd - R:S standard deviation How to cite: Schepaschenko et al. (2011) https://doi.org/10.1080/1747423X.2010.511681 Schepaschenko et al (2015) https://doi.org/10.1134/S1995425515070136 Schepaschenko et al (2018). https://doi.org/10.3390/f9060312.
Global measurements of forest height, biomass are urgently needed as essential climate and ecosystem variables, but can benefit from greater co-operation between remote sensing (RS) and forest ecological communities. The Forest Observation System - FOS (https://forest-observation-system.net/ [https://forest-observation- system.net/]) is an international cooperation to establish a global in-situ forest biomass database to support earth observation and to encourage investment in relevant field-based observations and science. FOS aims to link the RS community with ecologists who measure forest biomass and estimating biodiversity in the field. The FOS aims to overcome data sharing issues and introduce a standard biomass data flow from tree-level measurement to the plot-level aggregation served in the most suitable form for the RS. Ecologists benefit from the FOS with improved access to global biomass information, data standards, gap identification and potentially improved funding opportunities to address the known gaps and deficiencies in the data. FOS closely collaborate with the CTFS-ForestGEO, the ForestPlots.net (incl. RAfNFOR, AfriTRON and T-FORCES), AusCover, TmFO and the llASA network. FOS is an open initiative with other networks and teams most welcome to join. The online database provides open access for forest plot location, canopy height and above-ground biomass. Plot size is 0.25ha or larger. Comparison of plot biomass data with available global and regional maps (incl. Kindermann et al., 2013; Thurner et al., 2013; Saatchi et al., 2011; Baccini et al., 2012; Avitabile et al., 2016; Hu et al., 2016; Santoro et al., 2018) shows wide range of uncertainties associated with biomass estimation.
No abstract is provided for this article.
Net Ecosystem Carbon Balance (NECB) of Russian forest for 2007-2009 is presented based on consistent application of applied systems analysis and modern information technologies. Use of landscape-ecosystem approach resulted in the NECB at 546+/-120 Tg C year ^1, or 66+/-15 g C ^-2 year ^-1. There is a substantial difference between the NECB of European and Asian parts, as well as the clear zonal gradients within these geographical regions. While the total carbon sink is high, large forest areas, particularly on permafrost, serve as a carbon source. The ratio between net primary production and soil heterotrophic respiration, together with natural and human-induced disturbances are major drivers of the magnitude and spatial distribution of the NECB of forest ecosystems. Using the Bayesian approach, mutual constraints of results that are obtained by independent methods enable to decrease uncertainties of the final result.
No abstract is provided for this article.
We provide four data records: 1.The reference data set as a comma-separated file ("reference_data_set.csv") with the following attributes: “ID” is a unique location identifier “Latitude, Longitude” are centroid coordinates of a 100m x 100m pixel. “Land_use_ID “is a land use class: 11 - Naturally regenerating forest without any signs of human activities, e.g., primary forests. 20 - Naturally regenerating forest with signs of human activities, e.g., logging, clear cuts etc. 31 - Planted forest. 32 - Short rotation plantations for timber. 40 - Oil palm plantations. 53 - Agroforestry. “Flag” identifies a data origin: 1- the crowdsourced locations, 2- the control data set, 0 – the additional experts' classifications following the opportunistic approach. 2. The 100 m forest management map in a geoTiff format with the classes presented - "FML_v3.2.tif ". 3. The predicted class probability from the Random Forest classification in a geoTiff format - "ProbaV_LC100_epoch2015_global_v2.0.3_forest-management--layer-proba_EPSG-4326.tif" 4. Validation data set as a comma-separated file ("validation_data_set.csv) with the following attributes: “ID” is a unique location identifier “pixel_center_x” , “pixel_center_y ” are centroid coordinates of a 100m x 100m pixel in lat/lon projection “first_landuse_class “is a land use class, as in (1). “second_landuse_class “is a second possible land use class, as in (1), identified in case it was difficult to assign one class with high confidence. 5. Original crowdsourced data set as a .csv table. 6. Compiled FAO FRA forest statistics and mapped classes by countries into one table (.csv format).
No abstract is provided for this article.
Indicators of biological productivity of forests (live and dead biomass, net primary production, net and grossgrowth) are crucial for both assessment of the impacts of terrestrial ecosystems on major biogeochemical cyclesand practice of sustainable forest management. However, different information and the diversity of methods usedin the assessments of forests productivity cause substantial variation in reported estimates. The paper containsa systems analysis of the existing methods, their uncertainties, and a description of available information. Withrespect to Northern Eurasian forests, the major reasons for uncertainties could be categorized as following: (1)significant biases that are inherent in a number of important sources of available information (e.g., forest inventorydata, results of measurements of some indicators in situ); (2) inadequacy and oversimplification of models ofdifferent types (empirical aggregations, process-based models); (3) lack of data for some regions; and (4) upscalingprocedure of “point” observations.Based on as comprehensive as possible adherence to the principles of systems analysis, we made an attemptto provide a reanalysis of indicators of forests productivity of Russia aiming at obtaining the results for whichuncertainties could be estimated in a reliable and transparent way. Within a landscape-ecosystem approach it hasrequired (1) development of an expert system for refinement of initial data including elimination of recognizedbiases; (2) delineation of ecological regions based on gradients of major indicators of productivity; (3) transitionto multidimensional models (e.g., for calculation of spatially distributed biomass expansion factors); (4) use ofprocess-based elements in empirical models; and (5) development of some approaches which presumably do nothave recognized biases. However, taking into account the fuzzy character of the problem, the above approach (aswell as any other individually used method) is not able to recognize structural uncertainties. In order to assess those,a special statistical procedure for harmonizing the multiple constraints of the estimates obtained by independentmethods (landscape-ecosystem approach; flux measurements; process-based vegetation models; inverse modeling)was used to estimate uncertainty of the final results. Application of the above methodology resulted in a reliableassessment of major indicators of productivity. For instance, live biomass at the country’s level is estimated withuncertainty of 4-6%, net primary production – 7-10% (confidential interval 0.9). It was recognized the tendencyof increasing productivity of Russian forests during the last four decades at level of 0.5 0.2% year
Net Ecosystem Carbon Balance (NECB) of Russian forest for 2007-2009 is presented based on consistent application of applied systems analysis and modern information technologies. Use of landscape-ecosystem approach resulted in the NECB at 546+/-120 Tg C year ^1, or 66+/-15 g C ^-2 year ^-1. There is a substantial difference between the NECB of European and Asian parts, as well as the clear zonal gradients within these geographical regions. While the total carbon sink is high, large forest areas, particularly on permafrost, serve as a carbon source. The ratio between net primary production and soil heterotrophic respiration, together with natural and human-induced disturbances are major drivers of the magnitude and spatial distribution of the NECB of forest ecosystems. Using the Bayesian approach, mutual constraints of results that are obtained by independent methods enable to decrease uncertainties of the final result.
Carbon cycling of terrestrial ecosystems is a fuzzy (underspecified) system that imposes substantial constrains on possibility to get unbiased estimates of basic intermediate components (e.g., Net Primary Production, Heterotrophic Respiration) and final results (e.g., Net Ecosystem Carbon Budget) of the account within strictly defined confidential intervals based on any individually used carbon cycling method or model. We present a methodology attempting at minimizing possible biases and restricting the multivariate uncertainty.s space. The methodology follows the principles of applied systems analysis and is based on integration of major independent methods of carbon cycling study (landscape-ecosystem approach, process-based models, eddy covariance and inverse modelling) with following harmonizing and mutual constraints of the results. Based on a case study for Russia's forests, we discuss strengths and limitations of the outlined methodology.
Many global and regional forest cover products have recently become available. The most advanced and comprehensive of these include the global land cover datasets (GLC2000, MODIS, GLOBCOVER), MODIS Vegetation Continuous Fields (VCF), LANDSAT based (e.g. Sexton et al., 2013) and radar based (e.g. Saatchi et al., 2010; Baccini et al., 2012; Santoro et al., 2012) products. However, they often contradict each other and are typically inconsistent with forest statistics. In particular, global land cover datasets contradict each other in many areas, have limited information about forest density and are not consistent with forest statistics. VCF most likely provides the most comprehensive information about forest density with a spatial resolution of 230m during 2000-2010. However when observing VCF dynamics for individual pixels, one can see variation that cannot be explained by forest cover dynamics, but instead by unstable pixel geometry and clouds. Landsat based products also suffer from cloud cover and cannot recognize sparse forest with canopy closure of 30% or less. Space-based radar is free from cloud, but still cannot reliably delineate areas as forest/non forest (Santoro, 2012). We compare all of the above mentioned remote sensing products with a sample of high resolution imagery provided by Google Earth. We have applied the crowd sourcing platform Geo-Wiki (Fritz et al., 2010, 2012) to collect 22K training points where the percentage of forest cover was estimated for a 1km pixel size. We applied the method of geographically weighted regression to calculate the map of probability of forest cover and the map of forest share. This involved the use of the Geo-Wiki training points in combination with the land cover products, MODIS VCF and LANDSAT. The synergy of remote sensing, statistics and crowd sourcing approaches was investigated to better understand the spatial distribution of forests. Both calibrated (using FAO FRA statistics) and non-calibrated (.best guess.) forest cover datasets were obtained based on this method. We compared the Geo-Wiki training points with initial datasets and the final hybrid forest cover (Table). 8505 (out of 22K) Geo-Wiki points were classified as forest. The hybrid product shows a very close approximation in terms of the amount of forest pixels while other datasets vary from -28% to +62%. The hybrid dataset shows the highest overall agreement (when 22K points are compared) -89%. Globcover and VCF have the highest forest agreement (8505 forest points compared) because of overestimates of forest area. The highest correlation (R2=0.76) is obtained with the hybrid dataset when comparing the forest share per pixel. The hybrid forest cover is currently under validation and available for visualization at http://biomass.geo-wiki.org.
We consider practicality of a verified account of Net Ecosystem Carbon Budget for forest ecosystems (CA) that supposes reliable assessment of uncertainties, i.e. understanding uncertainty of The FCA is a fuzzy (underspecified) system, of which membership function is inherently stochastic. Thus, any individually used of FCA is not able to estimate structural and usually reported method are inevitably partial. Attempting an estimation of full uncertainties of the studied system we follow the requirements of applied systems analysis integrating the major methods of terrestrial ecosystems carbon account, assessing the method for intermediate and final indicators of CA with their following mutual constraints. Landscape-ecosystem approach (LEA) 1) serves for strict systems designing the account, 2)contains all relevant spatially distributed empirical and semi-empirical data and models, and 3) is presented in form of an Integrated Land Information System (ILIS). By-pixel parametrization of forest cover is provided by utilizing multi-sensor remote sensing data (1 RS products used) within GEO-wiki platform and other relevant information based on special optimization algorithms. Major carbn fluxes within the LEA (NPP, HR, disturbances etc.) are estimated based on fusion of empirical data with process-based elements by sets of regionally distributed models. Uncertainties within LEA are assessed for each module and at each step of the account. Withi method results and (including LEA, process-based models, eddy covariance, and inverse modelling) are harmoized based on the Bayesian approach. The above methodology have been applied to carbon account of Russian forests for 2000-2010;uncertainties of the FCA for individual years were estimated in limits of 25%. We discussed strengths and weaknesses of the paproach, system requirements to different methods of FCA, information and research needs, obtained and potential levels of uncertainties.
In accordance with the concept that only full accounting of major greenhouse gases corresponds to the goals of the United Nations Framework Convention on C
A number of global and regional maps of forest extent are available, but when compared spatially, there are large areas of disagreement. Moreover, there was no global forest map that is consistent with forest statistics from FAO (Food and Agriculture Organization of the United Nations). By combining these diverse data sources into a single forest cover product, it is possible to produce a global forest map that is more accurate than the individual input layers and to produce a map that is consistent with FAO statistics. In this paper we applied geographically weighted regression (GWR) to integrate eight different forest products into three global hybrid forest cover maps at a 1 km resolution for the reference year 2000. Input products included global land cover and forest maps at varying resolutions from 30 m to 1 km, mosaics of regional land use/land cover products where available, and the MODIS Vegetation Continuous Fields product. The GWR was trained using crowdsourced data collected via the Geo-Wiki platform and the hybrid maps were then validated using an independent dataset collected via the same system. Three different hybrid maps were produced: two consistent with FAO statistics, one at the country and one at the regional level, and a “best guess” forest cover map that is independent of FAO. Independent validation showed that the “best guess” hybrid product had the best overall accuracy of 93% when compared with the individual input datasets. The global hybrid forest cover maps are available at http://biomass.geo-wiki.org. More details can be found in the paper: Schepaschenko D., See L., Lesiv M., McCallum I., Fritz S., et al. (2015). Development of a global hybrid forest mask through the synergy of remote sensing, crowdsourcing and FAO statistics. <em>Remote Sensing of Environment </em>162 208-220. https://doi.org/10.1016/j.rse.2015.02.011. The data set consists of following files: 1. for2000_bg.zip - Global forest mask "best guess" - percentage forest cover at a 1 km spatial resolution for the year 2000;<br> 2. for2000_ca_cou.zip - Global forest mask calibrated to the FAO FRA statistics at national scale;<br> 3. for2000_ca_reg.zip - Global forest mask calibrated to the FAO FRA statistics at continental scale;<br> 4. training_pc.csv - training data, which contains visual interpretation of very high resolution imagery at 20159 locations;<br> 5. validation.csv - validation data, which contains visual interpretation of very high resolution imagery at 1816 locations.
