University of Twente Student Theses


Multi-frequency SAR data-based PolInSar inversion modeling for forest height retrieval

Rawat, Akshat (2021) Multi-frequency SAR data-based PolInSar inversion modeling for forest height retrieval.

[img] PDF
Abstract:With the threat of climate change and its impact on the plant and animal life on earth looming around us for over decades, it becomes necessary to understand the concept of climate change. An increase of the global temperature is a direct impact of the climate change process, resulting in subsequent rise in sea levels, stronger cyclones, and wildfires. The effects of these on animal life can be deciphered easily, but their effects on plant life is adverse and can only be understood with a proper study. Forests store in them huge amount of carbon, with the majority being stored in the trees and soil, constituting to the biomass. Using the height of the trees in the region, diameter at the breast height of all those trees, and the density, the above ground biomass is estimated. When a calamity such as forest fires, or deforestation occurs in a forest area, it helps in determining the amount of carbon released into the environment, and subsequent steps that must be taken to account for the loss that occurred during the event. This study focuses on determining one of the bio-physical components of the trees, the tree height using spaceborne Synthetic Aperture Radar (SAR) data, with the chosen area being the Manali forest ranges, Ram Bagh and Van Vihar in Himachal Pradesh, India. Data from multiple frequencies, X, C, and L band data are used in this study with the Polarimetric SAR Interferometry (PolInSAR) technique being implied as it has shown potential to determine accurate forest height results over many studies in the past (Aghabalaei et al., 2020; Chen et al., 2021; Denbina et al., 2018). Baseline simulation was carried to estimate the value of baseline compatible with the corresponding height of ambiguity, since prior knowledge of the forest height of the area was present. This was used to calculate the vertical wavenumber. Three Stage Inversion (TSI) and Coherence Amplitude Inversion (CAI) models were used in the study for determining the forest height. The modelled height was then validated with the field data of the field work carried out by Dr. Shashi Kumar in the year 2012. While the accuracy of the modelled height with CAI technique ranged between 84% and 92% for the three sets of data, it ranged between 91% and 95% for them when TSI technique was used. Further, the Global Ecosystem Dynamics Investigation (GEDI) product, 2019 global forest canopy height map was used for comparison with the field work as well as with the modelled height. It was observed at many places, the GEDI data had underestimated the forest height values, while at some places with lesser forest area, GEDI had not estimated the forest height. The changes in tree height due to temporal gap between the year of the field work and the acquisition date of the data pairs were not considered in the scope of the study. The forest ranges in the study are old and the trees are matured to their maximum height by the time field data was performed. The results suggested the use of TerraSAR-X data for the estimation of tree height in the region as its accuracy was most with the least RMSE value. Amongst TSI and CAI, TSI model was preferred since the inaccurate heights of urban and riverbed could be dealt with using a two-step improved TSI technique.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page