Resumen de Development of burned area algorithms on a global scale
- español
El trabajo de tesis titulado "Desarrollo de algoritmos de área quemada a escala global - Development of burned area algorithms on a global scale" ha sido desarrollado y financiando en el marco del proyecto fire_cci dentro del programa de cambio climático de la Agencia Espacial Europea. El objetivo principal de esta tesis doctoral ha sido desarrollar un algoritmo para la caracterización de áreas quemadas (AQ) a escala global a partir de información del sensor MERIS. Dentro de la tesis se ha buscado contextualizar la relevancia del fuego a escala global. Se han revisado los métodos para caracterizar los incendios desde el espacio, llevando a cabo una revisión bibliográfica del estado del arte. Se ha desarrollado y probado el algoritmo de área quemada, basando su configuración final en los distintos métodos implementados y en los resultados de las pruebas realizadas. El algoritmo obtenido puede clasificarse dentro de la categoría de algoritmo híbrido, ya que combina la información obtenida del contraste térmico (proporcionada por el producto MODIS HS) y de los cambios temporales en las reflectividades de los datos MERIS. El algoritmo consta de dos fases: semillado y crecimiento. En la primera fase, se identifican los píxeles semilla, es decir los puntos más claramente clasificables como quemados. Para ello se obtienen de forma dinámica estadísticas locales (basadas en regiones de 10x10 grados) de forma mensual que permiten definir condiciones para clasificar los píxeles semilla. En la fase de crecimiento se realiza un análisis de los píxeles vecinos a estas semillas, estableciendo su carácter quemado si verifican a su vez una serie de condiciones. Se ha llevado a cabo un análisis y discusión de las estimaciones de área quemada obtenidas mediante este algoritmo a nivel global para los años 2006 a 2008. Estos resultados se han validado e inter-comparado con otros productos de área quemada. Se incluyen así mismo en la tesis las conclusiones obtenidas del desarrollo del algoritmo, y los posibles futuros pasos a seguir. El principal logro del trabajo realizado en el marco de este trabajo de investigación ha sido el desarrollo del primer algoritmo de áreas quemadas a escala global a partir del sensor MERIS. Esto permite obtener productos de AQ a mayor resolución que la proporcionada por las colecciones de AQ existentes en la actualidad, y mejorando la calidad de las colecciones obtenidas a nivel europeo.
- English
This manuscript contains the research work entitled “Development of Burned Area algorithms on a global scale”. This work was developed and funded in the framework of the fire_cci project within the European Space Agency’s Climate Change Initiative programme. The PhD aim was to obtain an algorithm to retrieve burned areas (BA) on a global scale from the MERIS sensor. The document is structured in the following chapters:
Chapter 1: Introduction: this chapter aims at putting in context the global relevance of fire and of accessing information on burned areas. The methods to characterize fire from space and the state of the art are reviewed here.
Chapter 2: Objectives: the aim of this chapter is to introduce the context of the PhD research as well as a more detailed description of the thesis objectives.
Chapter 3: BA algorithm development and testing: in this chapter an overview of general considerations required to build the algorithm is included. The methods developed and tested to obtain the algorithm to detect burned areas will be presented.
Since results from these tests helped shaping the algorithm final configuration, the structure in this thesis is not separated in methods and results. Instead, results from the different phases of the algorithm are introduced in this chapter.
Chapter 4: BA product assessment: burned area global estimates for years 2006 to 2008 obtained with the MERIS algorithm developed are included here. Results analysis, validation and inter-comparison with other BA products are introduced in this chapter.
Chapter 5: Discussion: analysis and discussion on the steps taken to develop the algorithm, results obtained and limitations found are included in this section.
Chapter 6: Conclusions and perspectives: main outcomes and lessons learned from the development of this algorithm as well as the way forward are introduced in this chapter.
The first two scientific publications result of this work are included in Annex 2 and Annex 3. Annex 2 is a paper published in Remote Sensing of Environment containing the main algorithm developments. Annex 3 is a paper submitted and accepted with minor corrections to Geofocus – Revista Internacional de Ciencia y Tecnología de la Información Geográfica. It summarises the main improvements made on version 1 that lead to version 2 of the algorithm.
The main achievement of the work carried out within this thesis is the development of the first algorithm based on MERIS data to obtain BA on a global scale, at higher resolution than the current BA collections, and improving the quality of existing European BA collections. Although the algorithm presents some limitations and future work is still needed to improve current results, the interest stands now on obtaining a longer time series, and on ensuring the continuity of these measurements, to monitor BA and to use this information in the context of climate change. In fact, this algorithm will be the basis for computing the full MERIS BA time series for the second phase of the fire_cci project (2002 to 2012). It will also prepare for the upcoming generations of the European Space Agency BA products, particularly for those based on Sentinel-3 OLCISLSTR sensors. The algorithm will also be adapted to other sensors, such as MODIS bands 1 and 2, with higher spatial resolution and revisit time. The availability of alternative BA time series would complement existing BA products, providing more robust estimations and improvements in the uncertainty characterisation of BA inputs for climate models. Furthermore, different BA products can provide diverse regional accuracies and therefore, merging outputs from global synthesis may greatly improve the way BA trends are currently characterised
