This file was created by scanning the printed publication. Errors identified by the software have been corrected; however, some errors may remain. Towards Harmonization for Monitoring Key Forest Variables in Europe Using Earth Observation Data 1 Sten Folving2 Pam Kenned y 2 Niall McCormick2 Abstract-The Member States ofthe European Union established the European Forest Information and Communication (EFICS) Program in 1989 with the aim to improve forest information in Europe and to facilitate the availability of the information. In 1994 the Space Applications Institute ofthe European Commission set up the Forest Information from Remote Sensing (FIRS) Project in order to support EFICS by developing methods for deriving forest information from earth qbservation data, principally mapped, geo-referenced information but also statistical data. Two studies carried out in the frame of these two activities clearly revealed the need for harmonization of the nomenclature and the defInitions and methods used for assessing the forest variables within the Pan-European area. As mapped forest information is lacking for most of Europe it has been considered practical to combine the provision of harmonized key variables with the development of methods for, eventually, providing the information in geo-referenced mapped format as derived from remotely sensed data. The use of earth observation data furthermore provides a continuous monitoring capability of some ofthe key variables which can be assessed by remote sensing at an acceptable degree of precision and accuracy. Therefore, several development studies on application of remote sensing for PanEuropean forest monitoring have been launched, e.g. forest area, species composition, structural diversity and change. These development studies are being supported by research projects under the so-called Framework Program of the European Commission. The member states of the European Union (EU) have not agreed upon a common forestry policy. Wood products were not among the commodities included in the original treaty, and forestry as such has not been included in the EU common agricultural policy (CAP). However, the European nongovernmental and governmental forest organizations have expressed their need for close cooperation, not least due to the fact thatEUROSTAT, (the statistical office of the EU), is obliged to collect reliable statistics on the EU forest sector. Thus, the Standing Forestry Committee (SFC), an advisory body to the European Commission (EC) was established in 1989. The SFC, assisted by the Working Party on Forestry Statistics under the Agricultural Statistics Committee of EUROSTAT, is the main player in EU forestry matters. Ipaper presented at the North American Science Symposium: Toward a Unified Framework for Inventorying and Monitoring Forest Ecosystem Resources, Guadalajara, Mexico, November 1-6,1998. 2Forest and Catchment Sector, Environmental Mapping and Modeling Unit, Space Applications Institute, Joint Research Center of the European Commission, 21020 Ispra (VA), Italy. e-mail: sten.folving@jrc.it USDA Forest Service Proceedings RMRS-P-12. 1999 Major decisions can only be taken by the Council, the assembly of EU ministers responsible for forestry in their respective countries. In 1989 the SFC had the so-called EFICS (European Forest Information and Communication System) regulation approved by the Council. The EFICS regulation requests the EC to provide a frame for collecting, coordinating and processing data concerning the forestry sector and its development. The EFICS is planned to take account of existing data, and in particular statistics compiled by EUROSTAT. It will make use of information available in the Member States, and in particular data contained in national forest inventories and any other relevant databases accessible at community and international level. But, still it is the SFC that has the mandate to decide on both the actual content and the implementation. The EFICS regulation has been extended until the year 2002. Concurrently with the preparation of the EFICS regulation the Joint Research Centre (JRC) of the EC initialized and implemented a European program (the European Collaborative Program for the use of high resolution Second Generation Earth Observation Satellites iIi the Management of the Less Favoured Areas) ECP. The ECP had the main objective to test the applicability of remotely sensed data in local management, e.g. land use planning and forestry, and to assist local users in the application of earth observation (EO) data for their own specific purposes (Folving and Megier, 1992). The idea was to create a link between practical local management and a centralized information system. The ECP clearly revealed wide interest from the forestry community in using remotely sensed data in daily management. Via the ECP it became evident that the main obstacles for using EO data were found in the cost of the data and the lack of cheap, user friendly software. Due to the principle of subsidiarity the program was stopped in 1992, and the development of methods for application of remotely sensed data for local forest management was suspended accordingly. In 1994, the FIRS (Forest Information from Remote Sensing) Project was launched in support to EFICS. The idea was to utilize the results from the ECP to implement the practical application of EO data in the monitoring of European forest areas (Kennedy et. aI., 1994). The main objective of the FIRS Project is to contribute to the development of a unified European forest information system (i.e., EFICS), by developing methods for providing both sectorial (i.e. production- related) and environmental (i.e. ecology-related) forest information in the form of both 371 statistical and mapped data - with emphasis on mapped data (Folving et. aI., 1995; Kennedy et. aI., 1995). The methods are based on the application of EO data and GIS techniques. Three basic inputs had to be created before the actual method development and their applications at a Pan-European scale could be fully embarked upon. Firstly, regionalization of the Pan-European area into major forest ecosystems had to be produced because it was foreseen that the methods had to be adjusted to "local" forest characteristics. This work was finished in 1995 and the results have been published (e.g., SAl, 1995) and are illustrated on the WWW (http://www.jrc.sai.egeo.firs). Secondly, a system of nomenclature had to be defined which could be applied to the Pan-European area using EO data. This work was finished in 1996 (Kohl and Paivinen, 1996) and is available at the same UTL address given above. Thirdly, a network of common test areas had to be selected, and so-called 'ground truth' for each site had to be provided. Due to a number of constraints, not least costs and the lack of harmonized information, such a network could not be established per se. It was therefore decided that test areas, and the data from these would be compiled in an ongoing manner through the various subprojects or application modules of the FIRS Project. Three major applications modules will be described in §3. Studies on nomenclature are dealt with in the next section which will also looks at new requirements in the ED for forest monitoring, in terms of the development of the Common Agricultural Policy (CAP). Section 4 presents the very preliminary work being done to address the challenges for the newest environmental trend in requiring sustainable development, preservation and enhancement of bio-diversity. Harmonization of Nomenclature and Information Priorities For many years organizations like EUROSTAT and UNFAO (United Nations - Food and Agriculture Organization) have compiled and published statistics on forests and forest products. This statistical information is based on national statistics. However, recent international conventions such as the Convention on Biological Diversity, the Ministerial Conference on the Protection of Forests in Europe require new types of information and new data to be collected. When the FIRS Project was launched, high priority was given to studies dealing with the definition of the European user needs concerning forestry and how the existing variations in assessment methods, nomenclature and definitions could be dealt with. The aim was to provide a proposal for a common nomenclature to be used for the EO data applications. The first study was very much directed towards spatial, geo-referenced information. In parallel, the EFICS Program launched a more general study on the same theme, but which was not restricted to the application of EO data. The study reviewed the majority of NFl systems currently in existence in Europe. The results are published in - European Commission, 1997. The FIRS Project's Nomenclature Study This study was carried out by a working group lead by the EFI (European Forest Institute) and the Swiss Federal 372 Institute for Forest, Snow and Landscape Research (Kohl and Paivinen, 1996). Three main "user" groups were asked which forest attributes were of most interest to them, and then to rank these attributes according to importance. The result is summarized in Table 1. AI though, as expected, most attributes are common to the three user groups, as expected, their priorities are somewhat complementary. 'Forest area' is ranked as a high priority by all groups, 'bio-diversity' and 'landscape-related' attributes are ranked very low by foresters engaged in forest productivity. Foresters mainly working with environmental protection and landscape management, however, put little emphasize on productivity information. Interestingly, several of the attributes requested are not included in many existing National Forests Inventories. This is especially true for many of the newly requested environmental attributes. Some of these parameters are, however, available from other sources or organiza tions, such as information on soil and water. The last column in Table 1 reveals that several attributes need to be harmonized because currently no common nomenclature or assessment method exists in Europe. This is somewhat worrying as it also clearly indicates that international statistics compiled from national statistics may not always produce comparable data. The feasibility of using EO data (from high, medium and low-resolution sensors), for assessing the attributes at various scales was also investigated. The outcome, summarized in Table 2, shows that high resolution EO data are potentially very important for assessing and mapping forest attributes but that data from instruments with medium to low spatial resolution are of little interest in European forestry. The feasibility of using EO data for forestry was also discussed in working groups set up as part of two international workshops initiated by the FIRS Project. One in J oensuu, Finland, (Kennedy et. al., 1995) and one in Vienna, Austria, (Kennedy, 1997). The groups discussed and evaluated the potential of using remote sensing for the assessment of key forest attributes. On the first occasion the group was also asked to evaluate the potential for using remote sensing operationally for mapping of the forest attributes at the European scale (Table 3). The interesting point is that foresters in academic research have rather low expectations ofthe application of EO data, whereas foresters dealing with practical inventory problems seem to have larger expectations of remote sensing technology (Tables 1 and 2). It is also interesting to note that there has been very Ii ttle change in expecta tions of academic foresters over a 2-year period. However, the launch of several SAR (Synthetic Aperture Radar) instruments seems to have created at least a small improvement in the expectation towards the use of microwave data in this field. The EFICS Information Needs Assessment In 1996, the European Commission decided to launch a detailed study on information needs, data acquisition methods etc., in Europe to support the EFICS. The European Forest Institute led the consortium carrying out the work. The Study is unique in providing a full overview, and comparison of forest surveys and inventory systems in 22 European countries (European Commission, 1997b). USDA Forest Service Proceedings RMRS-P-12. 1999 Again the difference in priorities between "production" foresters and "environment" foresters is evident. The ranking of attributes in Table 4 has been carried out using questionnaires. A total of 500 questionnaires were sent out, a little less than half, 43%, were returned. The organizations or persons contacted were asked to list and rank the 15 most important attributes. The rankings of 'important' and 'very important' are used for the total ranking in shown in Table 4. The study revealed slight differences in regional interests in Europe. Northern European countries seem to be a little more interested in forest volume and in costs, whilst in central Europe the protection function of the forests seem to be of greater concern, than in the other parts of Europe. Northern European countries are unique in listing timber quality, the Atlantic countries in afforestation and central Europe in woody biomass, recreation and non-wood goods. In general, forest area (83% ofthe replies) and tree-species composition (79%) are the most important attributes, followed by protection functions (77%) and nature conservation area, volume of annual increment and cut (76%) and, finally, the attribute biological richness (75%) and diversity (7%). Growing stock volume (71%) and Health (69% of the replies) are also considered of high interest. Putting the Nomenclature System Into Use The interest of having a common nomenclature and common assessment methods is not just due to the need for comparable statistical information. It is also linked to needs created by the Common Agricultural Policy of the ED, and to a strong political need for rural development, especially in the so-called less favoured areas. The less favoured areas of the ED are of course defined in economic terms, but, more often than not they correspond to the mountainous and hilly regions, and are frequently characterized by forested or other wooded land. Such areas are also the most important sources for fresh water. Table 1.-The ran kings of the forest attributes according to the FIRS Project's information needs assessment. Attributes Land cover (type, density etc.) Actual forest area Other wooded area Potential forest area 1 Stand structure (Species composition, layers, density.) Age Diameter Height Q~~ Health Defoliation Damage (fire, storm, ins., diseases, game, pollution) V~~e Assortments Timber value Woody biomass 1 Herb biomass 1 Growth/Increment Drain/removals Soil types 1 Site factors Vegetation types Topography Climate1 Productivity Regeneration Stand history Ownership Management objective Value of protected infrastructure 1 Water resources 1 Protection status 1 Naturalness 1 Threats to species diversity 1 Environmental impact1 Non-wood goods and services1 Scenic beauty1 Ranks for information groups Land use Production Environment 5 1 12 6 2 4 2 2 2 3 3 6 3 9 12 8 1 1 ~ 11 2 8 8 2 4 1 3 3 11 13 3 13 13 12 6 4 11 5 13 13 9 11 12 7 7 14 ~ 5 4 10 9 8 3 6 10 7 4 9 8 14 13 9 9 16 3 5 1 6 5 7 7 7 14 9 5 3 4 2 minor required existing required required required required 2 2 2 1 1 1 1 required 8 4 6 10 3 required required required 7 6 12 15 14 10 2 3 2 Need for harmonization minor required required 5 8 5 2 6 9 4 6 required required required ~ 1Attributes not assessed by most of the national forest resource assessments. 1 highest rank - 5 lowest rank. USDA Forest Service Proceedings RMRS-P-12. 1999 373 Rural development and a Common Agricultural Policy are synonymous with the payment of subsidies. Within the EU, this means subsidies for taking land out of agricultural production, subsidies for afforestation on these so-called set aside land, and subsidies to the farmer until the point in time when the newly forested areas provide an income. Harmoni- zation at the European level is therefore a necessity to aid the implementation of tools to control the expenditure of the subsidies which are allocated and implemented on the basis of 'local' or regional conditions. The most recent EC regulation on the Common Agricultural Policy and Rural Development is not yet available. Table 2-Feasibility of using Earth Observation data at various resolutions. Attribute Ha Forest area 0.5 1 10 100 0.5 1 10 100 0.5 1 10 100 0.5 1 10 100 0.5 1 10 100 Tree stand Tree stand 0.5 Other wooded land Land cover Stand structure Vegetation type Diameter Height Volume Woody biomass Drain/removals Damage Health Increment Topography 10 1QO 0.5 1 10 100 0.5 1 10 100 0.5 1 10 100 tree 0.5 1 10 100 0.5 1 Spatial patch arrangement 0.5 1 10 100 374 Nomenclature Definition High Forest cover 21 -100 % Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Open forest, forest cover 5 - 20 %, species able to grow> 7m Open forest, forest cover 5 - 20 %, shrubland 21 - 100 % Crown density in 5 % classes Crown density in 20 % classes Crown density in 40 % classes Crown density in 40 % classes Species spp. Stand mean diameter, 1.3 m diameter Height of single tree, Mean height in 5 m classes Above ground volume of standing trees 50 cubic meter pr ha classes Dry weight of woody plants 50 t pr ha classes Over bark volume of trees In 50 cubic meter pr ha classes Two classes: >50 cubic meter pr ha no damage Crown thinning characteristics like: shape and color Increment between two successive assessments 50 cubic meter pr ha pr 5 years No ? Yes Yes No ? Yes Yes No ? Yes Yes No ? Yes Yes ? ? Yes Yes No ? ? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes Elevation in 100 m classes, Aspects in 8 categories and slope in 10% classes, Relies in plane, convex and concave Dominance, contagion, fractal dimension etc. Feasibility Medium ? ? Yes Yes Yes Yes ? No No ? Yes No No ? Yes No ? ? Yes Yes ? ? Yes No No No ? Yes No ? No ? Yes Yes Low No No No ? No No No ? No No No ? No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No USDA Forest Service Proceedings RMRS-P-12. 1999 Table 3.-Potential of Remote Sensing for mapping forest attributes. Potential of using EO-data1 O~tical data Active microwave data Attribute Forest / non-forest Coniferous / broad-leaved Broad-leaved genus Coniferous genus Biomass and volume Closure Stage of forest development Spatial diversity - fragmentation Hydrological conditions Infrastructure . Topography Drastic & rapid changes Slow & gradual changes 1994 1996 1994 2 2 4 2 2 4 3 3 3 3 4 4 4 4 4 4 4 4 2 2 3 3 2 4 2 4 3 3 1 1 2 1 1 2 3 4 2 4 4 1996 Operatlonallty2 3 1 1 4 4 4 4 4 3 3 3 4 2 3 3 3 2 2 2 2 3 3 2 2 2 3 1Potential: 1 = very high; 2 = high; 3 = moderate; 4 = poor. 20perationality (at a European scale): 1 =operational; 2 = semi-operational; 3 = research. What is well known however, is that the conventions on biodiversity and sustainable development of forests in the context of rural landscape will be included. This means that there will be a need for much stronger links between the forestry sector and the environmental organizations on the requirements for maintenance of ecological values. These links will undoubtedly benefit from having a harmonized set of definitions when dealing with the forest ecosystems. As a consequence, in order to incorporate these shifts in policy, the FIRS Project was partly re-focussed at the beginning of 1997 (E.C., 1997). The aim is to accommodate, as far as possible, the new requirements, and to integrate the new needs emerging from EU regulations. The key attributes needed by environmentalists and ecologists had to be incorporated and more efforts are being placed on the mapping and monitoring forests rather than on statistics. Six attributes were selected as prime foci of the FIRS Project. These are, in order of present'priority: 1) forest area; 2) other wooded land; 3) structure and composition; 4) volume, biomass and fuel; 5) bio-diversity; and, 6) environmental indicators. Special emphasis should be given to change in these attributes. Some of the major on-going projects dealing with the provision of information on these attributes at the European level are described in §3. One result of adopting these new developments has been that new challenges come to the surface. There is an urgent need for example, to be able to assess the precision and accuracy of spatial data, and to study Europe, less as a continent of countless administrative divisions, as is done for the more traditional statistical reporting, but more as a geographic entity divided into facets governed both by natural environmental features and socio-economic activities. There is a shift therefore, to utilize natural watersheds in order to allow holistic approaches to studying and understanding entire forest ecosystems and their functions. On-Going Projects The research and development (R&D) activities being coordinated, monitored and undertaken within the FIRS USDA Forest Service Proceedings RMRS-P-12. 1999 Project are outlined in Table 5. Only a small part of the RID is carried out in-house as the policy of the EC is to involve the Member State institutions as much as possible. The contracted R&D is competitive and to a large extent being financed by other EC services which have specific interests in the results or methods developed. The FIRS Project is also engaged in third party work and participates in Share Cost Actions, whereby studies are selected and 50% funded by the EC. The FIRS Project also assists the Directorate General in the coordination of the forest R&D projects which include components utilizing remote sensing techniques. Two major projects, a "Pilot study in the field of monitoring forested areas" and the "Forest Monitoring in Europ~ using Remote Sensing (FMERS) Project" deal with assessing the forest attributes identified as priority variables in Europe. These are mapping forest and other wooded land the identification and monitoring of criteria and indicators of structural biodiversity and biomass estimation. A project entitled "An AVHRR-based probability map" aims at the provision of a tool for probabilistic mapping of forest areas in Europe using statistics from EUROSTAT. The project with Regione dell'Umbria is a support study to the Share Cost Action entitled MARIE-F (Monitoring and Assessment of Resources in Europe - Forest) in which the FIRS project is a partner. All these projects supply input for the in-house R&D (primarily software development and modelling), and assist in providing the necessary background information for the definition of new R&D actions. Changes in Forest Lands The objective of the project "Pilot study in the fields of monitoring forested areas" is to develop standardized methods for using remotely sensed data and GIS techniques for the provision of statistical and mapped data on European forest resources. The developments are based on existing experiences and will provide cost-effective tools, for long term monitoring of the structural bio-diversity and major changes of the forested areas of Europe. 375 Table 4.-The final ranking of the most important attributes C CI) "t:I e cu E e ten ~ .s; CI) L.. .E :::a 5i~ CI) t- > 0 Cl 222 49 2 "0 Decrease of forest land E eL.. .2 iii N ·ccu ij Cl 0 0 en ~ ~ 1;; CI) L.. .E S ~ 12 18 Z1 10 23 49 1 3 8 4 5 2 11 4 8 1 4 8 3 3 4 5 1 8 4 6 5 9 5 Protective function and natural conservation 5 Volume of annual cut 6 8 3 Volume of annual increment 7 10 3 Biological richness and diversity 8 6 3 Changes in growing stock volume 9 9 10 2 11 10 Growing stock! stem volume 11 8 Transfer of 'exploitable forest' to other use 13 13 Exploitable forest 14 13 17 18 Timber quality/ assortments 19 Forest damage (excluding fire) 19 Woody biomass 21 Productivity/ site quality 22 Recreation! nwgs 23 Wildlife habitat 24 Potential land for afforestation 25 Vegetation type 26 5 27 11 28 23 Changes in above-ground biomass 3J Total biomass 31 13 4 9 2 11 10 6 9 11 1 5 13 12 5 32 11 11 Stand structure (density, layers) 33 Accessibility 35 ~ 9 9 3 33 ~ 9 4 10 13 9 14 9 Landscape/ scenic beauty Soil 6 4 5 12 Recreation! forest area Forest damage by fire 5 8 Silvicultural treatment "Naturalness" 6 11 9 6 10 5 11 Ia a: 3 2 9 11 m 3 11 Ownership Volume of mortality (natural losses) 5 3 Health condition! vitality of standing trees ~ a. ::l Increase of forest land Protection function .c > .i: J! 1;; Tree species composition 15 cuo ~Cl)J! ene eCl) oE u e "t:I 5 16 ._ Cl _L.. ..E 4 Plantations c·c OCU en 2 Age class distribution .E CI) -; .2 iii N ~~ ::l.-> CUe ZCl) CI) Forest land Other wooded land 376 :::a e E::l e::: L..::l Cl)u >.oL.. Cl~ cu Attribute eCI) e e 4 8 11 USDA Forest Service Proceedings RMRS-P-12. 1999 Table 5.-The major studies of the FIRS project. Activities within the FIRS project Contracted R&D projects Financed by Consortium Pilot study in the field of monitoring forested areas Directorate General VI, Agriculture, fisheries and forestry GAF mbH, Germany, JOANNEUM Research, Austria, IVL, Sweden Agence M.T.D.A., France University of Freiburg, Germany FMER8-1 Forest mapping CEO Project VTT Automation, Finland CESBIO, France European Forest Institute, Finland SCOT Conseil, France University of Bologna, Italy (+ nine associates) FMER8-2 Bio-mass assessment CEO Project SSC, Sweden METLA, Finalnd SLU, Sweden ISA, Portugal AVHRR Forest Probability Map EGEO VTT, Finland MARIE-F- Regione dell'Umbria EGEO Regione dell'Umbria Forest Department, Italy Share cost projects and third party work MARIE-F DG XII-8hare cost action EARS, Holland FIRS, Joint Research Centre, University of Freiburg, Germany METLA, Finland University of Leicester, England CEMAGREF/ENGREF, France Evaluation of RS data for classifying and mapping the forests of Ireland Irish Department of Agriculture, Food and Forestry Coillte Teoranta, Ireland FIRS, Joint Research Centre NRSC, England Coordination and technical assistence in support to other services of the European Commission The study consists of two parts: 1) Forest monitoring: The objective here, is to develop and demonstrate a system to identify significant changes in forest cover. The major focus will be on the detection of changes that differ from the normal vegetative succession ofthe ecosystems. Such changes are for example, related to growth conditions, cuttings and damage. The system will be as independent from ground data as possible. The performance of the system will be based on optical satellite data, and facilitate both large and small-scale monitoring capabilities. 2) Structural diversity: The aim ofthis component is to define, develop and test a system for monitoring the structural diversity of forested areas. Pilot studies are being carried out in the main forest ecosystem regions of Europe. The system will be based on high and medium spatial resolution satellite data, and will be able to discriminate forest patches and classify them according to content and shape. The system is linked to the change detection system developed in the first part of the project. Forest Monitoring in Europe (FMERS-1) Study The FMERS project has two parts. Part 1 focuses on the applications of medium spatial resolution EO data for large area forest mapping. Part 2 is a research and development study to evaluate EO data for assessing the above-ground woody biomass of forest and other wooded land in Europe. The main objectives of the forest area mapping component of the FMERS project are: 1) to investigate the utility of medium spatial resolution satellite data for forest monitoring at a European scale; and 2) to complement the forest data already being collected by the national forest inventories using more traditional methods. USDA Forest Service Proceedings RMRS-P-12. 1999 The project will develop and implement methods for using both optical and microwave remotely sensed data for the provision of standardized geo-referenced information (i.e., location, size and composition of forested and other wooded land) and for the provision of statistical information (i.e., area estimates) describing the forests and other wooded lands in Europe. The first part of the study developed the methodology and tested the applicability of the developed method in test sites located in the major forest eco-regions of Europe (Hame et.al., 1998). The second part is focusing on two large study regions, one comprising the Atlantic and Mediterranean forest types, the other the boreal and central-western European temperate forest regions. The aim is to evaluate and quantify the differences both in technical and financial terms between using medium and high spatial resolution satellite data for mapping, and for deriving area estimates of the main forest categories and their composition (groups of main species) of both forest and other wooded land. It is also the aim to clearly identify the potential and limitations of these satellite data to complement the more traditional methods, and to investigate the portability of the methods to the entire continent of Europe for up-dating on a 3 to 5-year basis. Forest Monitoring in Europe (FMERS-2) Study The objective of this study is to evaluate remotely sensed data for assessing the above-ground woody biomass, (and volume where feasible), of forest and other wooded land of Europe. The study will develop models which can be used with remotely sensed data and calibrated using ground information, for deriving estimates of above-ground woody biomass for large regions. A boreal and an Atlantic-Mediter- 377 ranean test-area have been selected to test the approach. In both test areas the model and approach will be investigated at three different pixel resolutions. These are:- I) very high (less than 10 m pixel); II) high (less that 100 m pixel), and III) medium (between 100 and 300m pixel). The very high resolution data will be simulated from air photographs. The analyses of the method-performance in the test areas will be used to evaluate the possibilities of utilizing the approach at a European level. MARIE-F Study The FIRS project is a partner in the project's team, but, has at the same time contracted a supporting study to one of the regional forestry authorizations in Italy. The objectives of the study are:- 1) to investigate and develop an objective satellite-based methodology for inventory monitoring of forest timber and forest vitality in Europe using a Forest Light Interaction Model (FLIM) developed by the leading partner at EARS in Holland, and 2) to evaluate the potential of the model to provide uniform and comparable forest baseline data and statistics, for larger areas and at lover costs, in support to national and European forest strategies. The Regione dell'Umbria Forest Department assists the FIRS Project in assessing and evaluating the performance and utility of the outputs from the model once applied to forest land in Umbria. Evaluation of Remote Sensing Data for Ireland The in-house developed software package, SILVICS, (McCormick, 1998; McCormick and Folving, 1998) which has been developed under the FIRS project has been improved and adapted to large scale real application within the frame of a third party contract from the Irish Department on Agriculture, Food and Forestry. SILVICS is a user-friendly software package, which runs on most computer platforms, and which provides advanced techniques for the geometric and radiometric correction, structural and statistical analysis, and classification of multispectral satellite imagery. The software is being implemented as part of a national forest inventory in Ireland. Figure 1 shows the forest classification of a Landsat TM imagery of an Irish test site. The Landsat TM data was classified into eight forest classes, in three main steps: 1) Prior to classification, a standard clustering algorithm (Isodata) was applied to the imagery. Each of the resulting clusters was then assigned a land cover category, based on a subjective visual comparison with available ground information. 2) The "raw" spectral samples (signatures) for each forest class of interest were then "purified", by using the clusters from step 1 to filter out pixels belonging to nonforest categories or to other forest classes. 3) The purified signatures from step 2 were then divided into two independent data-sets. The first set of purified signatures was used to train the four SILVICS classification algorithms, and to classify the image. The second set of purified signatures was then used to assess classification accuracy. A comparison of the classification accuracies for the four SILVICS image classification algorithms is shown in Table 6. 378 As can be seen the mahaianobis distance and maximum likelihood algorithms gave the highest classification accuracies. It is important to note that the high classification accuracies shown in Table 6 are due to the subjective interpretation of image clusters, which were used to purify the raw signatures (as described in steps 1 and 2 above). Thus, the classification accuracies in Table 6 primarily illustrate the relative performances of the four classification algorithms. An assessment of the true classification accuracy should be determined by ground surveying. Summary And Outlook _ _ _ __ Other studies on the application of remote sensing in forestry have been partly, or in total, financed by the EC under the EUFourth Framework Programme. These Projects are more directed towards primary research than the ones under the FIRS Project, described above. However, results and methods which can be directly, or at least easily implemented and used for practical purposes will be included in the so-called 'toolbox' being developed for linking the application of EO data into the EFICS. Figure 2 shows this toolbox concept: EFICS will link users with National Forest Inventory data from the Member States, and with metadata bases, Global Forest Networks etc. Most of the data being supplied to the potential user in this way will be statistical and will be enumerated for various statistical units, regions, forest districts and so on. The projects under FIRS and the Share Cost Action projects will contribute to the toolbox assisting the users to retrieve, and to combine information from EO data with statistical information for their various needs. Publicly available software like the SILVICS package SILVICS ClASSIFICAnON OF lANDSAT TM IMAGE FOR IRISH NA1l0NAl FOREST INVENTORY 157000 158000 159000 160000 161000 208000 208000 207000 207000 206000 206000 205000 205000 204000 204000 157000 158000 159000 160000 Class Names • Spruce - mature ( 1) II Spruce - ma:ure (2) Spruce - young • 161000 Class_Names Spruce - open • Broadleai Pine • Mixed conifer Mixed pine I I broadleaf spruce Figure 1.-Map from one of the Irish test sites. USDA Forest Service Proceedings RMRS-P-12. 1999 Table S.-Relative accuracies of SILVICS classification methods. Forest class Spruce-mature (1) Spruce-mature (2) Spruce-young Spruce-open Pine Mixed pine/spruce Broadleaf Mixed conifer/broad leaf Overall accuracy (%) Classification accuracy for four selected algorithms (%) Neural Minimum Mahalanobls Maximum likelihood network distance distance 92.5 88.1 97.3 66.5 87.9 90.5 95.2 86.4 87.6 will constitute an important part ofthe toolbox, but the most important part will consist a of common nomenclature and definitions providing comparable information. The main requirements are presently to develop modules for assessing the diversity of the forest cover. The EU commitments towards international conventions, such as the Helsinki Process (Finnish Ministry of Agriculture and Forestry, 1993) has shifted the priority away from pure production related indicators towards environmental indicators. This, together with the increasing needs for mapped information for global, regional and national modeling and especially the needs from physical planning, have influenced the priorities of the work program for the next 4 year period. The forest diversity module follows the outline in figure 3 (McCormick and Folving, 1998). The compositional elements are being supplied through the FMERS Project and the structure and change elements are supplied from the "pilot study in the field of monitoring forested areas". The basic techniques consist of standard image processing procedures, but the processing methods and the nomenclature and, eventually, the verification will follow the findings in the mentioned supporting studies. As environmental issues are slowly becoming more and more important for the forestry sector, and as this new trend 95.4 94.7 89.8 84.7 94.7 85.7 93.6 92.5 91.3 93.9 94.7 91.3 83.0 89.1 93.9 94.8 89.3 91.0 97.6 90.5 87.4 76.5 91.5 90.2 95.2 91.8 89.8 Figure 3.-The forest diversity components of the toolbox being developed under the FIRS project in support to the European Forest Information and Communication System. is closely linked to environmental planning and modeling, the challenge for National Forest Inventories are twofold: Firstly, the new bio-diversity indicators and indicators on sustainable development have to be defined. Secondly, spatial data sets need to be prepared as it is not sufficient to know the magnitude of some forest variable or indicators, but, more - it is necessary to know exactly where to find the type, its locations, shape and extension. Forest areas are becoming important as environmental and socio-economic buffers. The FIRS project is accordingly being changed to meet these new challenges. The Project is being merged wi th activities dealing with watershed management. The next Framework period will thus be concentrated on the development of criteria and indicators for assessing and monitoring forest diversity at the landscape scale, and on the development of new forest-vegetation typologies and vegetation abundance descriptions suited for run-off and erosion modeling, and for general watershed management. References --------------------------------- Figure 2.-The "toolbox" concept of the FIRS project. The Project is aSSisting EFICS in providing the necessary methods for extracting relevant forest information from EO data USDA Forest Service Proceedings RMRS-P-12. 1999 European Commission. 1997. FIRS Status Report, October 1997. Space Applications Institute, Joint Research Centre of the European Commission. Special reports. S.P.1.97.94. European Commission. 1997. Study on European Forestry Information and Communication System. Reports on forestry inventory and survey systems. Luxembourg: Office for Official Publications of the European Communities. 379 Finnish Ministry of Agriculture and Forestry. 1993. The Resolutions and the General Declarations of the Second Minesterial Conference ofthe Protection of Forests in Europe. Conference on the Protection of Forest in Europe, 16-17 June, 1993, Helsinki. Folving, S. and Megier, J. 1992. Remote sensing applied in management of the Less Favoured Areas and landscape ecological mapping - experiences from the JRC Collaborative Programme. Proceedings UNESCO MAB Seminar: "Changes in agricultural systems: driving factors, effects on landscapes and ecological patterns. Caen, Normandy, September 9-11, 1991. Folving, S., P. Kennedy and N. McCormick (1995) EC Initiatives on European Forest Ecosystems Mapping and Forest Statistics. In: Proceedings to the 14th EARSeL Symposium on 'Sensors and Environmental Applications of Remote Sensing', Chalmers University of Technology, Goteborg, Sweden, 6-8 June, 1994. Harne, T. Andersson, K, Lohi, A; Kohl, M; Paivinen, R; JeanJean, H; Spence, I; LeToan, T; Quegan, S; Estreguil, C; Folving, S. and Kennedy, P. 1998. Validated forest variable maps and estimates across Europe using multi-resolution satellite data: Results of the first phase of the FMERS study p. 705-708. In Proceedings to: 27th International Symposium on Remote Sensing of the Environment, 'Information for Sustainability', Tromso, Norway 8_12th June 1998. Kennedy, P. J. (Ed.) 1997. Application of remote sensing in European forest monitoring. Proceedings, Workshop, Vienna, Austria, 14-16 October, 1996. EUR 17685. Kennedy, P. J., Folving, S. and McCormick, N. (1994) An introduction to the FIRS Project. In: Proceedings to International Confer- 380 ence on Satellite Technology and GIS for Mediterranean Forest Mapping and Fire Management', Aristotle University, Thessaloniki, Greece, 4-6, November 1993. Kennedy, P., S. Folving and N. McCormick (1995) European Forest Ecosystems Mapping and Forest Statistics: The FIRS Project. In: Proceedings to International Workshop on 'Designing a System of Nomenclature for European Forest Mapping', European Forest Institute, Joensuu, Finland, 13-15 June 1994. EUR 16113. Kennedy, P. J., Paivinen, R. and Raihuvua, L./Eds. 1995. Designing a system of nomenclature for European forest mapping. Proceedings, Workshop, Joensuu, Finland, 13-15 June 1994. EUR 16113. Kohl, M and Paivinen, R. 1996. Defmition of a system of nomenclature for mapping European forests and for compiling a PanEuropean forest information system. Space Application Institute, Ispra, EUR 16416 EN. McCormick, N. 1998. An integrated system for mapping European forestsusing medium- and high-resolution satellite imagery: the SILVICSsoftware. Joint Research Centre, European Commission. EUR Report.(In preparation). McCormick, N. and S. Folving. 1998. Monitoring European forestbiodiversity at regional scales using satellite remote sensing.In: Assessment of Biodiversity for Improved Forest Planning. Bachmann,P., M. Kohl, R. Paivinen (editors). Kluwer Academic Publishers,Dordrecht, The Netherlands. SAl Special Publications. 1995. Regionalization and stratification of European forest eco-systems. EC JRC Space Applications Institute Special Publications, S.PI 95.44. USDA Forest Service Proceedings RMRS-P-12. 1999