1) Link biodiversity users and Earth Observation experts
Prof. Per Wramner will lead the Diversity User Bureau (DUB). The DUB coordinates the Diversity User Group, a cross cutting activity which will run in parallel to the technical tasks and will focus all user interactions in one place. Based on his long term involvement in the CBD process and – on the other hand – his technical expertise, he will facilitate the dialogue between the biodiversity users and the EO experts of the team, making sure that the most suitable indicators will be generated, and keeping the users closely attached to the project.
The indicators will provide synthesised information of all status -, change - and trend maps and indicators. At the time of proposal writing it is too early to come up with a precise definition of these indicators (a provisional list is provided in the proposal), and to arrive at sound, meaningful and scientific solid definitions is the subject of the user interactions guided by the DUB. However, the indicators will function in such a way so as to describe and highlight, in an easy understandable, abstracted (high level) manner, the overall status and trends of inland waters and drylands based on the individual input indices and indicators. These high level indicators will aim at a differentiation of climatic drivers and (potential) human drivers of biodiversity loss (or improvements), in order to be relevant for policy development and implementation (high level).
2) Selection of best algorithms
Deriving the basic water quantity and water quality parameters for inland waters is a challenge. In particular the best quality atmospheric correction and the best quality retrieval of inherent optical properties (IOPs) and water constituents in optical complex waters is currently a methodological problem which has not yet been satisfactory solved. BC, BG and consultants together have a long term experience in this field, and we will perform a comparison of various algorithms, including our own adaptations of the Case2R and FUB algorithms to inland waters, using in-situ data from team members and simulated data as reference. The developments and findings of the recent years, through EU R&D projects such as Freshmon and the Round Robin intercomparisons of the ESA projects CoastColour and Ocean Colour Climate Change Initiative (CCI), show that there appears to be a general indication that iterative methods for inverting the radiative transfer equation perform best under all conditions. Additionally, these algorithms have the advantage that they can be parameterised using different SIOP models and are thus easily applicable to a global dataset of optically different inland waters.
EO based approaches to the assessment of dryland conditions and assumptions with regard to biodiversity are primarily dependent of the geographic scale. Global and continental assessments such as in this study, analyse time series data of above ground green biomass, which is estimated for fixed increments of time (most typically decades, months, years) using EO derived vegetation indices or bio-physical indices. Most commonly, the NDVI, the Normalised Difference Vegetation Index, is used as a proxy for NPP. We will refer to these NDVI based heritage methods on the one hand in order to maintain monitoring continuity to some degree, but, on the other hand, we will use more advanced indices, such as FAPAR, and model approaches which constitute an agreed basis for future assessments and for finally supporting policy development and implementation. Subsequently, efficiency indices will relate NPP indices to their driving forces. We will compare indices for rainfall and potential evapo-transpiration as two candidates and decide together with users and ESA which one of the two will be produced.
3) Software and production - building on experience and re-use/ extension of existing components
The software to generate the Diversity II Products will be complex and has to work on very large amounts of data efficiently; BC and GeoVille are well experienced and prepared for this challenge, due to their respective experiences and in their re-use of software components developed in previous projects. The processing chain includes steps from child product generation, merging of data from different sensors and sources, cloud screening, atmospheric correction, bio-optical inversions, indices calculation, spatial and temporal integration, change detection, indicator calculation, up to the final map generation. The production involves processing of more than 100 terabytes of input data. These tasks will be performed in a shared way by Brockmann Consult (pre-processing and inland water) and Geoville (Drylands), and both companies have suitable software and hardware capabilities in house. In particular BC has experience in processing the complete, global MERIS FR data set through the Land Cover CCI and CoastColour projects, and we will build the Diversity II production line on the basis of these mature production systems. The new algorithms will be developed largely using the BEAM toolbox, which provides many building blocks and an efficient environment for processor development. The Dryland processing at GeoVille will start from pre-processed and integrated data, provided by BC, and will also be based on existing processing modules (ERDAS), with the required extensions and new modules.
Algorithm development, software development and product generation will be performed in an iterative, agile process. This enables us to generate different product versions, reflecting the current development status, for validation and for feeding back into the development process. We will generate
- After month 7 (PDR): prototype products for 4 lakes and 3 drylands, generated with different algorithms for algorithm intercomparison
- After month 10 (CDR): prototype products for 10 lakes and 5 drylands, generated with the finally selected proposed algorithms
- After month 14 (QAR): qualified products for 10 lakes and 5 drylands, generated with the final algorithms, for qualification of the production system
- After month 20 (SDR): all products (300 lakes and 22 drylands) for assessment by users
Validation is a key for acceptance of the products by the users, and this is the vital to the success of the project. We will validate the basic EO parameters, i.e. over land the surface reflectances and the vegetation indices, and over water the water leaving reflectance, IOPs, chlorophyll-a concentration, Total Suspended Matter (TSM) concentration, turbidity/ Secchi Depth, as well as the indicators. Basic EO parameter validation will be supervised by CIBIO using in-situ data from project partners, consultants and public data sources. The validation protocol will be based on international standards, such as the MAVT protocol (now MERMAID protocol), which was also the source for the MERIS Lakes validation protocol and the MarCoast validation protocol. Standard statistics will be employed (match-up analysis), and we will also perform satellite product intercomparison, in particular where correlative in-situ data are difficult to obtain (e.g. for vegetation indices). A detailed validation of the indicators will be performed by CIBIO using their multi-scale species database and public data sources. A technique involving autocorrelation analysis and species distribution modelling will be used to generate maps of the probably presence for individual species. These are the basis for calculating the same indicators as derived from the EO data, for intercomparison. Map comparison algorithms will be calculated, Kappa, Fuzzy Kappa, and Fuzzy Inference System, to quantify levels of similarity and spatial agreement.
5) Communication and product dissemination
Guided by the Diversity User Bureau we will develop the internet presence for the Diversity project (WebPortal). This will provide background information about biodiversity in Inland Waters and Drylands, and about the project. All project deliverables will be published here and we will use the Web Portal to support the user requirements phase with an online questionnaire. At a later state of the project, the products will be made available through a WebGIS interface. The user handbook will become an online document, with a tutorial for working with the Diversity products, and presenting the Biodiversity stories to provide detailed material about the biology of the sites and the value brought by the EO products.
6) Preparing the future
The objective of the Diversity II Project is to support the implementation of the new 2011-2020 biodiversity strategic plan, and as such it requires a vision and planning that do not end after 24 months, when the project is completed. The project can only be successful if - by the time the projects finishes - a sustainable continuation of the activities until 2020 is in place or at least well prepared. This requires a critical review of the lessons learnt during the project, i.e. identified limitation of the approach and recommendation for improvements, as well as a cost-benefit-analysis, which is important when seeking further funding. With ENVISAT approaching the end of its lifetime, a technical analysis will be undertaken in order to analyse the possibilities, constraints and possible improvements of a continuation and further development of the Diversity II Products in the era of the Sentinels and Proba-V. This activity of the project will again be coordinated by the DUB and performed in close cooperation with EO experts and users.