Use this URL to cite or link to this record in EThOS:
Title: New techniques and insights for burned area monitoring from remote sensing data
Author: Brennan, James Robert
ISNI:       0000 0004 9359 3828
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2020
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Satellite-derived datasets recording the global extent of burned area (BA) have been generated in the past two decades. These products provide vital information to fire-related disciplines. The algorithms used to produce these products generally share a low level of methodological consistency, are highly empirical, and lack some of the sufficient features required to produce consistent and error characterised long-term data records (LTDRs). This thesis sets out to re-examine the quality of the algorithms used to produce these products with the aim of advancing the quality of future BA products. While BA products have been validated and inter-compared, little is known about the individual qualities and sensitivities of their respective algorithms. A novel sensitivity analysis of six global burned area algorithms provides information on the key sensitivities determining detection of burned area from the observations as well as highlighting the limitations of present algorithms. This analysis highlights the need for quantitative quality information within BA products through uncertainty characterisation. Best practice frameworks for uncertainty characterisation have been developed for many Essential Climate Variables (ECVs) but not yet burned area. The thesis then goes on to develop suitable methodologies for estimating uncertainties in present BA products and future products. Global uncertainties of 4--6\% are calculated for current BA products, and the necessary error propagation frameworks adapted to be amenable to binary ECV records such as burned area. The information gained from the sensitivity analysis and advancement of uncertainty characterisation provides a framework to prototype multi-sensor algorithms suitable for producing uncertainty characterised LTDR BA datasets. This involves the application of a spectrally invariant model of the burn signal suitable for mapping BA in a probabilistic manner from optical sensors to provide consistent estimates through the satellite record.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available