Ground deformation modelling based on causal factors in landslide-prone areas. A case study of Bududa district, Uganda.
MetadataShow full item record
The measurement of ground deformation has become a very critical part of landslide hazard assessment because before a landslide happens, ground deformation precedes which when reliably measured, communities at greatest risk can be determined. This leads to appropriate resettlement of communities which significantly reduces the loss of lives and property due to landslides. However, currently, the measurement and assessment of ground deformation have not considered the whole deformation process, which begins with the favourable causal factors, that progressively lead to the accumulation of ground strain, and finally to ground deformation. This is done through measurement of the causal factors, determining their influence on ground deformation and computing ground strain from ground deformation. This was accomplished through a quantitative approach that involved measurement of ground deformation using Interferometric Synthetic Aperture Radar (InSAR) and Global Navigation Satellite Systems (GNSS) techniques, computation of ground strain using the least-squares collocation method based on a covariance function and deformation tensor analysis of GNSS derived deformation velocities, and modelling the relationship between causal factors and ground deformation using multivariant linear regression modelling. Results indicate that spatial deformation magnitudes, velocities and temporal deformation in Bududa ranged from -3 to 6 cm, -1 to 4 cm/yr. and -4 to 7.6 cm in the Line of Sight (LOS) direction as measured by Small Baseline Subset (SBAS) InSAR from 2015 to 2018 well as spatial vertical deformation magnitudes, velocities and temporal vertical deformation ranged from 2 to 12 cm, 0.9 to 6cm/yr and -3 to 12 cm as measured by Persistent Scatterer (PS) InSAR from 2019 to 2020. GNSS showed Bududa to experience spatial horizontal and vertical deformation magnitudes from 0.4 to 7 cm and -10 to 2.5 cm and velocities from 0.1 to 2.3 cm/yr and -4 to 1 cm/yr from 2018 to 2021 at the installed GNSS monitor stations. Temporally, GNSS measured horizontal deformation ranging from 0 to 20 cm and the vertical from 0 to \pm 21cm. The ground strain slowly built up in the north, east and vertical directions with averages of 7.11 x 10 -5, 8.52 x 10 -5 and 5.17 x 10 -5 microstrain/yr respectively. The rotation, dilatation and maximum shear rates ranged from -100 to 100 0/k*Myr, -0.02 to 0.02 microstrain/yr and 0 to 2.5 x 10 -3 microstrain/yr respectively. The soils had average silt, clay and sand contents of 20%, 34% and 46% respectively and average infiltration, field capacity, saturation and bulk density of 0.397 cm/hr, 0.292 cm3water/cm3soil, 0.490 cm3water/cm3soil, and 1.349 g/cm3 at the GNSS stations. The correlation between rainfall and ground deformation ranged from R2 0.05 to R2 0.80 while moisture and ground deformation ranged from R2 0.06 to R2 0.88. As for the slope, the correlation between slope and ground deformation was R2 0.71. Ground deformation magnitudes and velocities as measured by InSAR and GNSS in Bududa district are high depicting the risk of landslides to the communities. The ground deformation in Bududa is a result of a slow build-up of ground strain, therefore communities are at risk but may not see anything to show risk at the moment at some places in the district. It has been observed that campaign GNSS and InSAR have the potential to reliably measure ground deformation in Bududa and also soil texture, slope, rainfall and infiltration can be used to predict ground deformation using multivariant linear regression. The communities located in areas of high ground deformation should be relocated. This information is vital to the Office of the Prime Minister and the Ministry of Disaster and Preparedness. It will be a good practice to make GNSS observations when InSAR is used. This will enable validation of InSAR derived ground deformation. The ground strain should be measured in addition to ground deformation to property characterise landslide hazards and soil texture, rainfall, slope and infiltration should be incorporated in the prediction of ground deformation. I have improved on knowledge on the understanding of the dynamics of ground deformation and strain in Bududa district. Ground deformation and strain have been extensively measured and computed for the first time in the district during this study. I innovatively combined InSAR and GNSS to measure ground deformation and compute strains and predicted ground deformation from rainfall, soil texture, slope and infiltration causal factors using multivariant linear regression. This builds on the knowledge to predict ground deformation from causal factors which have focused mainly on the use of bivariant linear regression. Further research would be exploring the use of longer wavelength SAR imagery in the L and P bands and the use of corner cube reflectors in InSAR processing. Single-frequency Continuous Operating Reference Stations (CORs) could replace campaign GNSS. More causal factors could be investigated in modelling ground deformation when their data is reliably available and lastly the model that predicts ground deformation from causal factors could be based on both InSAR and GNSS deformation and also used other models for example Artificial Neural Networks (ANNs).