Ledger, Martha
(2022)
Monitoring surface oscillation dynamics of tropical peatlands: a novel approach using APSIS-InSAR.
PhD thesis, University of Nottingham.
Abstract
80% of tropical peatland in Indonesia and Malaysia (15% of Earth's soil carbon) is drained for production of pulpwood and palm oil. Associated increases in peat decomposition and large-scale forest fires have led to widespread subsidence and deterioration of peat condition. However, quantification of subsidence and peat condition from these processes across SE Asia is challenging due to the scale and inaccessibility of dense tropical peat swamp forests.
Space-based platforms offer the opportunity to monitor these inaccessible environments with regular and efficient pan- regional measurements. A development in satellite interferometric synthetic aperture radar (InSAR), a technique that measures surface motion, has the potential to solve this problem. A new ‘intermittent small baseline subset’ (APSIS, formerly ISBAS) modelling technique, developed at the University of Nottingham, provides excellent coverage across almost all land surfaces irrespective of ground cover. This enables derivation of a time series of tropical peatland surface oscillations across whole catchments, regions and countries.
This project aimed to establish the extent to which APSIS InSAR can monitor seasonal patterns of tropical peat surface oscillations, and therefore physical peat condition, at North Selangor Peat Swamp Forest, Peninsular Malaysia. Firstly, a proof-of-concept study demonstrated how the APSIS technique can monitor peat surface oscillations under tropical forest canopy using C-band InSAR data, enabling continuous monitoring of tropical peatland surface motion ranging from 0.1 – 40 cm yr-1 at a spatial resolution of 20 m. Secondly, a ground-based study explored the relationship between tropical peat surface oscillations and physical peat condition across North Selangor. Thirdly, a statistical and conceptual comparison of surface motion time series produced by ground-based methodologies and the APSIS method was conducted. Finally, the spatial and temporal patterns of APSIS-derived surface motion was explored through use of machine learning to determine which peat swamp forest variables most strongly influenced surface oscillation patterns, and to what extent APSIS could inform physical peat condition at North Selangor.
The proof-of-concept study showed that C-band Sentinel-1 SAR could penetrate the forest canopy over tropical peat swamp forests and was applicable to a wide range of land covers. This presented potential for monitoring tropical peatland degradation and the impacts of management strategies with greater accuracy than L-band SAR. Results from ground-based methodologies showed that peat condition and tropical peat surface oscillation magnitude were significantly different between peat swamp forest condition classes and at different depths. Links between peat condition and surface oscillation magnitude were found, whereby more degraded tropical peat had greater elastic potential and therefore greater surface oscillation magnitudes. However, a regional-scale investigation showed that above-ground biophysical variables were overall poor predictors of APSIS-InSAR surface oscillation patterns. Further, comparisons between surface oscillation time series from ground-based methodologies and APSIS-InSAR showed no significant relationships. Further work was recommended to extend the period of measurement for all methodologies to include more full oscillation cycles, as well as tease apart the different components that contribute to the surface oscillation signal of each of the methodological approaches. Regional-scale variables indicative of the peat profile should also be incorporated into future study, particularly the inclusion of water table change, which has demonstrated a strong control on peat surface amplitude at North Selangor. The inclusion of a undisturbed site as a benchmark would also enable real quantification of the extent of degradation across North Selangor.
Continued development of remote sensing methods is recommended for effective and sustainable tropical peatland restoration and management at regional scales, including the support of global climate mitigation efforts. It is intended that the methodologies explored in this thesis will be built upon to better understand the regional-scale surface oscillation dynamics of tropical peat swamp forest environments.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
![[thumbnail of ML_thesis_AMENDED.pdf]](https://eprints.nottingham.ac.uk/style/images/fileicons/application_pdf.png "ML_thesis_AMENDED.pdf")