Reducing Emissions from Deforestation and Forest Degradation (REDD) is an international effort to create a financial value for the carbon stored in forests, offering incentives for developing countries to reduce emissions from forest and invest in low-carbon paths to sustainable development. Of critical importance is the estimation of carbon stocks in forests and their dynamics. The paper presents a scientifically sound and technically feasible approach for carbon stock assessment over large areas by applying a multi-stage inventory concept. This concept combines in-situ data and wall-to-wall forest monitoring using medium to very high resolution remote sensing data from both, optical and SAR satellite systems. This approach assures that not only forest cover is assessed but that estimates on forest degradation and forest dynamics can be monitored with high accuracy.

Self-organizing maps (SOMs) are an advanced neural networks application. SOMs were applied for the spatially explicit estimation of forest carbon stocks for a test region in Thuringia (Germany) The approach utilizes in situ national forest inventory data and satellite remote sensing data (Landsat 7 ETM+) and provides maps showing a high-resolution spatial distribution of forest carbon stocks The generated maps are compared to alternative estimates obtained by the k-nearest neighbour (kNN) method a remote sensing based carbon assessment Beside maps the SOM- and kNN-approaches were utilized to calculate statistical estimates of carbon stock and growing stock. The statistical estimates were validated by calculating bias and mean square errors with reference to in situ assessments.SOM- and kNN-approaches have been tested in a forested region in Central Germany The results show that SOMs are an approach that has the ability to reproduce the spatial pattern of forest carbon stocks. SOMs are with some restrictions comparable to spatially explicit estimates generated by the kNN-method. (C) 2010 Elsevier B V All rights reserved

BACKGROUND:Forests occur across diverse biomes, each of which shows a specific composition of plant communities associated with the particular climate regimes. Predicted future climate change will have impacts on the vulnerability and productivity of forests; in some regions higher temperatures will extend the growing season and thus improve forest productivity, while changed annual precipitation patterns may show disadvantageous effects in areas, where water availability is restricted. While adaptation of forests to predicted future climate scenarios has been intensively studied, less attention was paid to mitigation strategies such as the introduction of tree species well adapted to changing environmental conditions.RESULTS:We simulated the development of managed forest ecosystems in Germany for the time period between 2000 and 2100 under different forest management regimes and climate change scenarios. The management regimes reflect different rotation periods, harvesting intensities and species selection for reforestations. The climate change scenarios were taken from the IPCC's Special Report on Emission Scenarios (SRES). We used the scenarios A1B (rapid and successful economic development) and B1 (high level of environmental and social consciousness combined with a globally coherent approach to a more sustainable development). Our results indicate that the effects of different climate change scenarios on the future productivity and species composition of German forests are minor compared to the effects of forest management.CONCLUSIONS:The inherent natural adaptive capacity of forest ecosystems to changing environmental conditions is limited by the long life time of trees. Planting of adapted species and forest management will reduce the impact of predicted future climate change on forests.

The study describes the potential of various habitat suitability indices (HSI) using remotely sensed data (Landsat 5 and 7) and other mapped information in digital format for the characterization and monitoring of rare species habitats at the landscape level over time. The main focus was to develop a flexible and open system for habitat monitoring which allows a pragmatic overview of habitat development without field assessments, or with very limited field assessments. The potential HSI models for the exemplary key species Red Kite (Milvus milvus) and Black Stork (Ciconia nigra) are analysed with regard to their sensitivity to changing environmental conditions, and also with regard to the influences of individual attributes used as input for the models. The Moritzburg area located close to the city of Dresden, Germany was selected as the study site. It is characterised by a pronounced heterogeneity of landscape elements such as forests, meadows and lakes. The remote sensing data for the year 2000 were combined with ground data collected in the field campaign of the EU research project “MNTFR”. In addition, the database “Datenspeicher Wald” provided forest information for the year 1989 based on the forest inventories at the company level. Attributes, based on Natura 2000, such as food supply or nesting resources, were utilised as input for HSI models. The in situ data were combined with satellite data using a spatial statistic approach called kNN method for extending in situ attributes to the entire area of interest. Habitat suitability maps for both occasions (1989 and 2000) were compared for the individual key species. The methods described underlay the three HSI models tested in this study: a) The Habitat Suitability Index (HSI) with binary attribute maps, b) The Enhanced Habitat Suitability Index (EHSI) applying binary attribute maps enhanced with fuzzy sets and c) The Habitat Suitability Index with Home Rage Aspect (HR-HSI) applying recalculated attribute maps with an activity radius of 200 m for each pixel. Each of these HSI models includes two levels of consideration: the attribute level and the life requisite level. The multiplicative approach with multiplicative combination of life requisites resulted in the original models HSI, the EHSI and HR-HSI and the summation approach with additive recombination of life requisites resulted in the models HSI+, EHSI+ and the HR-HSI+. While all six HSI models are able to detect habitat changes and to predict future habitat development, the EHSI model proved to be efficient to enhance purely binary data into discrete transition probabilities along suitable pixel with a decreasing probability within a distance of 150 m. The HR-HSI model proved to be useful in describing neighbourhood relations of habitat attributes. It offers a graduation of habitat potentials calculating continuous transition probabilities. The HR-HSI model is sensitive for areas of minimum 25 hectares indicating potential habitat loss or gain in the test site. The model approaches can support decision- and policy-making concerning landscape management, as well as enabling simulation of changing individual attributes. The main obstacle to a successive implementation of the HSI models is a comprehensive description of factors driving habitat suitability that have hardly been presented in quantitative terms. Therefore an interdisciplinary knowledge transfer is recommended to realise the implementation of quantitative information of species specific requirements.

The paper describes the potential of remotely sensed data (Landsat 5 and 7) for the characterization and monitoring of forest habitats at the landscape level over a period of I I years (1989 and 2000). The Moritzburg area located close to the city of Dresden, Germany was selected as the study site. It is characterised by a pronounced heterogeneity of landscape elements such as forests, meadows and lakes. A part of the Landsat scenes, which cover 2830.5 ha was used for the application of habitat suitability models for two selected key species: kite (Milvus milvus) and black stork (Ciconia nigra). The remote sensing data for the year 2000 were matched with ground data from a field campaign. In addition, the database "Datenspeicher Wald" provided field information for the year 1989 describing the past forest management activities, forest structure and inventory data. Attributes, based on Natura 2000, such as food or nesting resources, were used as input for habitat suitability models. The ground surveys were combined with the satellite data using the kNN-method for extending derived attributes to the entire area of interest. Habitat suitability maps for both occasions (1989 and 2000) were compared for the individual key species. Between 1989 and 2000 the area of potential habitats for kite has increased from 4.6% to 5.9%. Most of the suitable habitat areas have changed their location during this period of time. The potential habitat of the stork increased from 12.8% to 14.8% of the area, but nearly all the habitat locations changed, as well. It was found that the habitat model is a useful approach to qualify potential habitats for umbrella species at the landscape level. For improving the accuracy of maps showing potential habitats the utilisation of auxiliary terrestrial data sources proved to be essential.

A species-habitat approach is suggested for assessment and monitoring of biodiversity at landscape level and a model framework for constructing species-specific habitat models is presented. The use of umbrella species as indicators for monitoring biodiversity at a landscape level is a cost-effective and promising approach. Within the DMMD (Development of Methods and Tools for Monitoring Forest Diversity as a Contribution to Sustainable Forest Management in Europe, carried out under contract to JRC-Ispra) project the suggested framework has been applied for nine species distributed over four test sites. The framework recognizes important behaviors of the target species that are modeled using one or several parameters. The quality, abundance and distribution of these parameters are estimated with a method using satellite data in combination with field measurements. The resulting habitat models are spatially explicit habitat suitability index (HSI) models based on expert judgments. The presented approach could be used for monitoring changes in habitat suitability for various species at a Pan-European perspective, e.g. in Natura 2000 areas.