NL FR EN
www.belgium.be

World Wide Watch by Earth Observation Services (WWW)

Research project SR/00/116 (Research action SR)

Persons :

  • Prof. dr.  DEFOURNY Pierre - Université Catholique de Louvain (UCLouvain)
    Coordinator of the project
    Financed belgian partner
    Duration: 15/10/2007-31/12/2008

Description :

The interaction between environmental change and human societies has a long and complex history. Despite a great spatial and temporal variability, a global perspective has begun to emerge in recent years to form the framework for a growing concern with regards to the human activities impact on the Earth System functioning, to the consecutive change on climate and biosphere and, finally to the various feedbacks to the human well-being. In parallel, the globalization of the world economy and the increasing awareness of global environmental issues also call for much more efficient information flows from objective record as obtained by Earth Observation (EO) to the public diffusion.

Recent advances in EO capabilities, in particular the impact of the very high resolution imagery, as well as the recent success of Google Earth have put the satellite remote sensing at the forefront of the global providers of land surface information. More specifically, the MODIS/MERIS medium resolution capabilities along with the 1-km SPOT VEGETATION instruments recently deliver several emblematic global products such as the GLC2000 land cover map, the Global Burnt Area 2000, the MODIS Fire product, the MODIS Rapid Response, the forthcoming GLOBCARBON and GLOBCOVER products.

Belgium has decided to play proactive role in the development of operational EO capabilities to deliver comprehensive relevant, high-quality information about the Earth System in the long term. The Global Monitoring for Environment and Security (GMES) initiative jointly supported by ESA and the European Union is a very dynamic framework supporting the development of an independent European operational monitoring capacity for the environment and civil security at regional and global level.

Based on significant experiences of the partners in the framework of GMES and to prepare the operational Belgian contribution to GMES Service Element programme, UCL-Geomatics and VITO have defined a program to design, develop, produce and maintain a suite of dedicated EO products for near-real time global monitoring. The present project corresponds to the phase 1 of this program which is defined to reply to information needs of different key global end-users.

UCL-Geomatics successfully tested a product prototyping method making the end-user actively involved in an iterative way to progressively tune the product in the framework of a pre-operational phase. In addition to its operational capabilities at global scale, VITO recently demonstrated the telecommunication efficiency to regularly disseminate added-value products directly to end-user on an operational basis. VITO has also acquired a multi-sensor expertise allowing combining different sensors to insure the service delivery independently of any specific EO source.

For Belgium this project is a great opportunity to build a partnership where complementary strong areas of competence of Belgian EO partners are combined, to tighten, consolidate and strengthen Belgian expertise in the European medium resolution EO scene. Furthermore, the continuous link between the end-users, the research lab and a production entity should allow for innovation or fast improvement of products and services over time.

Objectives

The first phase of the project aims:

to design and prototype an added-value products for monitoring Desert Locust Habitat and to design an addedvalue products for monitoring Agriculture Phenology, both based on the specific demands of the well identified end-users communities;
to assess the actual niche of the proposed products with regards to the existing one to properly communicate about these;
to operationally deliver the Desert Locust product to a well defined and active user community in near-real time STEREO I I Project N° SR/00/116– WWW Ini t i al repor t 6 and with the relevant recurrent time interval over a long and continuous time-period;
to incorporate the Desert Locust product in the FAO/ECLO operational decision support systems.

From a methodological point of view, the project has two transversal objectives:

- to validate an original product prototyping approach based on close and iterative interaction between the reference user and the research lab allowing to test and tune the final product during a significant pre-operational period thanks to a high-speed very flexible processing facilities
- to develop a new conceptual approach to design multi-sensor products transparent for the end-users with regards to the EO data sources.

Further objectives of this project and of this partnership are to help Belgium, by joining research and production forces at the
national level, to be recognized as an international centre of competence in the global land monitoring scene

Method

Research results demonstrated the quality of the medium resolution sensors currently in orbit, i.e. SPOT-VEGETATION, ATSR-2, AATSR, MODIS and MERIS instruments. Radiometric, geometric, and atmospheric correction performances allow considering new approaches in the data pre-processing as well as in the information extraction techniques. New operational products can therefore be developed on a short term basis.

A product cycle is composed of two parts, i.e. the development and the production, and includes for each product four different stakeholders closely connected, i.e. the production entity, the research team, the reference users (FAO/ECLO, FAO/GIEWS, UNEP-WCMC) and the users community. The proposed product development approach corresponds to a progressive and iterative prototyping strategy made of 4 steps with several loops allowing a fine tuning during a preoperational phase : (1) the product identification from the users community practices, (2) the algorithm design by the research entity, (3) the pre-operational prototype production and use in the application context while continuous contacts between research entity and reference end-users will allow iterative improvements, (4) the operational product ‘qualification’ by qualitative and quantitative assessment. In parallel to this fourth step, an independent benchmarking is needed to verify the performance of the new product with regards to the already existing ones. Once qualified the implementation of a product can proceed with the coding, the product quality control, the large scale dissemination planning, and the operational delivery. According to the complexity of the algorithm and of the processing requirement, the pre-operational products may be delivered either by the research lab or by the production entity.

The two main project originalities are the product development approach and the capabilities to derive added-values products from the best available EO source whatever the sensor. Unlike generic off-the-shelf EO products, very dedicated services jointly designed from the information content, layout and dissemination point of views are assumed to be readily adopted by the end-users while they remain cost effective from a production point of view when included in a products portfolio.

Product will be developed for at least two different EO medium resolution sensors among SPOT-VGT, MERIS, MODIS and AATSR. Of course the selection of the relevant sensors for a given product have to take into account the user STEREO I I Project N° SR/00/116– WWW Ini t i al repor t 7 requirements, to maximize the sensors compatibility (spatial resolution, spectral bands, observation geometry, calibration) with regards to the developed algorithm and to minimize the sensitivity of the product quality. Two or more sensors as EO sources will enhance the robustness and the reliability of the product delivery and should be as transparent as possible to the users. Different levels of fusion are currently foreseen to combine several EO sensors.

The difference in product quality and the fusion impact on product accuracy will be assessed by inter-product validation as well as comparison with independent data (e.g. ground-based measurements). Any fusion strategy will then be driven by a set of rules defining the conditions where merging appears as a plus or just a backup to insure a minimum product quality.

The three products considered for the two phases of the project has been selected because of their critical importance for the environmental monitoring in a context climate change, the explicit needs of key UN agencies and, the already existing scientific results and experience mainly based on SPOT-VEGETATION.

Desert Locust Product (reference user: FAO/ECLO)
Desert locusts are known to cause extensive crop damage and thus constitute a serious nuisance. It is recognized that the vegetation onset indicates the development of favourable conditions for the locust expansion. The key issue consists thus in monitoring the vegetation development, detecting its onset then forecasting possible locust habitats.

Currently, the monitoring system for this species depends on colour composites visual interpretations or on NDVI values analyses (fixing a NDVI critical threshold). The main inconvenient of this last method is an unavoidable confusion with some types of soil. Vancutsem et al. (2007) have recently developed an alternative compositing algorithm providing high quality temporal syntheses with a great spatial and temporal consistency in all the spectral bands. The detection method proposed in this project is based on this innovative pre-processing technique, on the use of a colorimetric system more adapted than the common RGB system and on the implementation of a new detection index. The methodology has yet been tested with SPOT-VEGETATION time series. Completed in collaboration with the FAO using ground observations, the validation demonstrated a significant improvements brought by this method.

The Locust product proposed in this project consists in an operational extension of this recently developed SPOT VEGETATION product to a multi-sensor product (SPOT VEGETATION, MODIS, and/or MERIS) so that it will combine the advantages of each of them (robust processing chain, daily coverage of the Earth’s surface and better spatial resolution). Indeed the FAO/ECLO found that 300-m resolution is really more relevant to the locust habitat detection for many areas.

Agriculture Phenology (reference user: FAO/GIEWS)
The monitoring of the agricultural growing season is of major interest for food security and early warning systems. Combined with rainfall distribution the satellite information on the crop growth provides a unique overview of the regional, national and sub-national situations. Furthermore, the inter-annual variability of the croplands phenology at the regional and national scales allows to interpret the current situation but also to study the possible trend related to climatic changes.
Traditionally, phenological metrics are based on a vegetation index such as the NDVI, with a frequency of 10 days, independently of the cover type and of the agro-ecological context. In the framework of this project, the products will focus on the characterisation of the phenology of agricultural vegetation and mainly deal with starting up and end of the vegetative period, date of the maximum development, growing and senescence rate, growth length, potential of vegetation and, amplitude. The originality of the proposed approach regards to the existing products relies on several elements: (i) the phenology metrics are "land cover specific" thanks to the use of GLOBCOVER map at 300-m resolution, (ii) a multi-sensors approach (VEGETATION, MODIS, MERIS) which provides the best information available with no need to combine different satellite sources of information, and (iii) the exploitation of the multispectral information and not only the vegetation index as usually done. This is possible thanks to the alternative compositing methodology allowing the production of spatially and temporally consistent information in all wavelengths.

Result

The product development cycle will be completed for the Desert Locust product and half way for the Agriculture product

Documentation :