WP2 – Harmonizing growing stock, biomass and carbon estimation and forest state monitoring based on field and high-resolution remote sensing data

Objectives
The objectives of WP 2 are to develop new methods and models to produce harmonised data and information on wood resources at the individual tree and sample plot levels (tasks 1 & 2), and to elaborate general, flexible and powerful estimation procedures that combine NFI field data and auxiliary information from remote sensing and maps (task 3). These procedures shall integrate into an innovative system for providing up-to-date estimates of state and changes in forest ecosystems on a variety of scales and at the European level. Tasks 1 &
2 are devoted to the provisioning services of forests (volume, biomass, carbon stocks and changes), while task
3 applies to further ecosystem services including regulating, cultural and supporting services (in line with WP3 and WP 5).

Description of work and role of participants Work package leader: INRA; co-leader: IGN Background:
Providing harmonised estimates of volume, biomass and carbon stocks of forests at the European scale is still a challenging but also path-breaking endeavor. Currently, European countries periodically conduct their NFI and apply their own, country-specific, volume and biomass models, as well as their own up-scaling methods to produce statistics at the national scale from sample trees and plot level data. In addition, trans-national compatibility of the models is unknown. There is neither a harmonised European system of equations and expansion factors for volume and biomass estimation nor a common guidance to apply these models to NFI data.
In this context, the structure of the DIABOLO consortium offers a unique opportunity (i) to develop innovative sets of total or partial volume and biomass equations from data collected and compiled by NFIs and related research institutes (unprecedented European data set), (ii) to take advantage of new technologies such as terrestrial laser scanning (TLS) to develop readily available non-destructive assessment methods for application at the plot level, and (iii) to develop general and flexible estimation procedures using NFI field data and auxiliary information from various sources for the provision of consistent, harmonised and regularly updated forest statistics at the European, regional, national and local levels.

Approach:

Concerning volume and biomass assessments at plot level, distinct approaches will be followed for a) stem volume where harmonisation by adjustment of national data to a reference definition is the most efficient option; and b) total volume and biomass where standardization (common definitions, methodology and models) is necessary because of the scarcity of relevant data in many countries. Thus, a common procedure for producing harmonised estimates of stem volume (Cost Action E43 definition) for the whole Europe will be developed by using bridging functions in task 2.1, while in task 2.2 generic total volume and biomass equations for European tree species will be constructed to fulfil the requirements of robustness across site conditions and forest management practices across Europe.

About general statistical outputs, a flexible and powerful multisource methodology to produce harmonised and updated forest statistics at different scales in Europe will be developed, using NFI field data enhanced with various kinds of auxiliary information from remote sensing and maps. The interest of utilizing different kinds of auxiliary information is threefold: i) post-stratification based on existing or forthcoming broadly available

medium to high resolution (5-10 meter resolution) space-born sources and/or large scale forest maps can greatly improve, at low additional cost, the precision of estimates regarding slowly varying characteristics (e.g., forest cover, forest types); and ii) remote sensing mapping of large disturbances as developed in WP 4 can easily be used to update NFI field data after the disturbance; but iii) better resolutions (< 1 meter) are needed to track more detailed or rapidly and locally varying phenomena like tree size, local fellings and mortality, as the improvement of change estimates would require. The methodology developed in task 2.3.1 will be able to use these three kinds of auxiliary information, and numerical procedures will be established to compute anywhere in Europe plot-level forest indicators from very high resolution remote sensing data such as aerial or VHR satellite images (task 2.3.2). Furthermore, the possibilities opened by the association of VHR remote sensing sources and NFI field data will be investigated to integrate forest management inventories (FMIs) and NFIs (task 2.3.3).

Work breakdown:

Unless otherwise stated, all tasks and sub-tasks start month M1 and end month M48

 Task T2.1 – Harmonisation of stem volume estimates in Europe (TRL 7)
Task leader: BFW Other contributors: IGN, TI, and all NFIs involved in WP 2
In Europe, NFIs are established and conducted in almost every country. Altogether, European NFIs survey and maintain more than 500,000 sample plots on forest land. Many international and country-level forest statistics rely on the estimates calculated from these sample plot data. Since NFIs have developed more or less independently from each other and therefore use different inventory methodologies, the resulting estimates are not comparable and reflect country-specific conditions. National stem volume definitions differ concerning the dbh-threshold and the inclusion or exclusion of tree components like stump, stem top, bark and branches. Building on the experience gained through the Framework Contract with the JRC, to which contributed all NFI partners, bridging functions for harmonised stem volume estimations will be developed for the conversion of country-specific estimates into stem volumes that correspond with the reference definition of COST Action E43. Such estimates are urgently needed for international reporting obligations, forest and environment policy decisions at the European level, and wood supply related issues. Possible methodological approaches range from simple conversion factors to complex models and refer to tree-, plot- or stratum-level. Building upon information from participating countries and availability of relevant data, an harmonisation concept will be developed. Different options will be evaluated by countires according to the criteria of applicability throughout Europe, accuracy and local capacity building. After calibration using a representative data set, the performance of the factors and models will be tested and results will be compared across countries. Through the application of the developed factors and functions by countries, harmonised tree volume estimates at plot level will be available and will be used in task 2.3.1 to produce harmonised estimates of growing stock at the European level. The available expertise from the involved participants in establishing and maintaining such a common system and data from many countries will be a valuable basis for achieving harmonised stem volume estimates throughout Europe for the first time.
Start month: M1 End month: M36

 Task T2.2 – Compatible set of biomass and total aboveground volume equations for the main tree species in Europe
Task leader: ICAS Other contributors: INRA, UCPH-IGN, TI, UHUL, FR, BFW, IGN, FVA T.2.2.1 Development of continental models of total aboveground volume and biomass (TRL 4) Main contributors : ICAS, INRA, UCPH-IGN, TI, UHUL, FR, BFW, IGN
Two types of modelling approaches will be developed in parallel, both being much promising for the building of generic models at European scale: (i) the biological approach, which aims at developing a new theoretical framework for elaborating compatible volume and biomass equations for any tree compartment; (ii) the statistical approach, which aims at revisiting all prior equations available for the main tree (see Globallometree.org) to elaborate new and pan-European models using Bayesian technics or pseudo-data generation. While the main advantage of the statistical approach relies on its ability to provide robust equations for species where biomass data are scarce, the biological approach allows to better capturing the geographical patterns of tree species traits. For model building, measurement data will be preferred to pseudo-data. Hence, all available data will be collected from the contributing partners to constitute a harmonised volume and biomass database.
T.2.2.2 Development of algorithms for 3-D tree reconstruction and volume measurements from

terrestrial laser scanning (TRL 4)
Main contributors : IGN, INRA, FR, FVA
In complement to these modelling approaches, terrestrial laser scanner (TLS) is a new and promising technology to assess aboveground volumes and biomass at tree and plot level. The challenge is to combine (i) a cloud density approach for small branches and understorey and (ii) a 3-D tree reconstruction approach for the stem and big branches size distributions and angles to assess total volume of the trees and the plot. The work will be carried out across a range of European forests to provide data on the accuracy and reliability of various instruments (e.g. Faro, Leica, Riegl), acquisition methods and novel high- performance 3-D reconstruction methods (e.g., geometrical modelling, image processing) which will be available through the dedicated Computree open source platform.

 Task T2.3 Improvement of forest resources monitoring through general and flexible estimation procedures associating NFI field data with auxiliary data from various sources.
Task leader: WSL Other contributors: IGN, NFLI, FVA, UHUL, Metla, UCPH-IGN
Three topics are addressed: a) the further development of a common data analysis and output delivery system providing harmonised information on status and change of forest resources in Europe based on NFI field data and various sources of auxiliary data (T.2.3.1), b) the development of methods to extract plot- level information from very high-resolution (< 1 meter) remote sensing sources to be used in the common system mentioned under (a) (T.2.3.2), and c) to highlight efficient pathways to integrate forest management inventories (FMIs) and NFIs in support of an efficient and sustainable management and development of forest goods and services in Europe (T.2.3.3). The data analysis and output delivery system is generic and flexible, i.e. various types of NFI field data and auxiliary data are handled (WP 3, WP 4, WP 5, Copernicus High Resolution Forestry Layer and equivalent). Remotely sensed data on changes and disturbances in forest resources (WP 4) are important auxiliary data for always up-to-date information provisioning.
Shortcomings addressed We provide an innovative and flexible system for common reporting, develop techniques to produce auxiliary data aiming at greatly increasing the precision of forest information and at updating NFI plot information from different points in time into consistent and timely information about the state and change of forest ecosystems over Europe (time harmonisation). We highlight pathways for the efficient integration of sub-national inventories and NFIs in support of a sustainable management of forest ecosystem goods and services down to the forest management and planning level.
T.2.3.1 Common data analysis and output delivery system (TRL 5) Main contributors : WSL, IGN, NFLI, FVA, UHUL, Metla, UCPH-IGN
We further develop the statistical methodology of the ENFIN’s EFDAC E-Forest estimate to be able to incorporate (a) various auxiliary data sources to improve the precision of estimates and forest resource information, and to incorporate (b) NFI field and auxiliary data from various time points and using imputation and updating techniques to enable the system to deliver always up-to-date information on status and changes of forest resources in Europe, in conformance with existing European information systems. This task covers the development of estimators, as well as the technical implementation and case studies for testing the system by participating countries.
T.2.3.2 Development of plot data extraction technology from VHR remote sensing sources and provision of auxiliary data for task 2.3.1 (TRL 5)
Main contributors : IGN, FVA
The focus will be on stereoscopic aerial and VHR satellite images (resolution < 1 meter). The information will cover different parameters like canopy cover, mean canopy height, crown/tree numbers, mean crown sizes, proportion of broadleaves/conifers, tree height and diameter distributions, etc. The aim is to develop robust procedures to extract comparable parameters from various forest types in different regions of Europe, including non-provisioning services. These procedures will (a) provide a better resolution at small scales (small-area estimation), and (b) allow regular updating of forest inventories by detecting changes at plot to forest level. Procedures will be tested on key case studies in different countries and the data will be used in T.2.3.1.
T.2.3.3 Integrating FMIs and NFIs (TRL 4) Main contributors : WSL, NFLI, UHUL, Metla
We address the requirement for sub-national and local information provision for effective and sustainable use of forest resources based on NFI field and auxiliary data. The focus is (a) on information needs in

FMIs considering spatial and temporal scale of estimates, and (b) on up-to-date small-area estimation techniques suitable for reliable information provisioning for multi-purpose management of forest goods and services. It is expected that the cost of FMI can be largely reduced using reference data and auxiliary data collected in NFIs. Methods and good practices for integrating FMIs and NFIs are of considerable interest in many countries and providing harmonised methods would be of high value.