UCLA

Quantifying tectonic uplift and erosion rates on various timescales is essential for understanding how tectonic and climate forcings interact to produce the landscape we see today. My dissertation is centered around three topics related to the erosion rates of the easternmost Himalaya in the Indian state of Arunachal Pradesh at decadal, millennial, and million-year timescales to better understand present-day natural hazards to long-term landscape evolution. In my dissertation, I quantify 1) landslide susceptibility, 2) landslide-derived decadal and 10Be-derived millennial erosion rates, and 3) plausible million-year timescale exhumation rates in the easternmost Himalaya to better understand climatic, topographic, and tectonic controls on surface processes and erosion over different timescales.

In the second chapter, I mapped landslide occurrences using satellite images, quantified landslide susceptibility using statistical and machine learning methods, and assessed the dominant controls of landslide occurrences in the easternmost Himalaya. Recently, deep neural networks (DNN) have been used in the application of estimating landslide susceptibility alongside physical and statistical models. However, DNNs are uninterpretable, making it difficult to determine mechanistic information about landslide controls in the modeled region. My landslide inventory was ultimately used to train an interpretable superposable neural network (SNN), developed and applied by my colleague, to model landslide susceptibility in the easternmost Himalaya. The SNN performed similarly to a state-of-the-art deep neural network, outperforming commonly used physically and statistically based models while revealing the relative importance of contributing controls. The analyses reveal that both strong slope-climate coupling and microclimates are dominant contributors to landslide occurrences in the region.

In the third chapter, I quantified landslide-derived decadal erosion rates over a 20-year interval from satellite images and determined millennial erosion using cosmogenic 10Be. Previous studies generally report a tectonic control on millennial erosion rates across the Himalaya. However, there exist well-understood and well-defined patterns of climate variation, tectonic deformation, and lithologic distribution along-strike of the Himalaya that allows for their disentanglement. In this chapter, I measured cosmogenic 10Be-derived millennial erosion rates of 12 basins from the range front to the hinterland of the Dibang and Lohit valleys. In addition, I compiled 161 10Be-derived erosion rates from the Garhwal, Nepal, and Bhutan Himalaya and grouped basins by dominant metasedimentary or crystalline lithology. I observe a clear correlation between climate metrics and erosion rates for basins dominated by metasedimentary lithology that is absent those dominated by crystalline lithology. Additionally, we find that the response of fluvial and hillslope erosional efficiency to climate differs between lithologies in the Himalaya. Furthermore, the high erosion rates and efficiencies observed in the easternmost Himalayan range front are likely facilitated by rainfall-induced landslides and efficient fluvial erosion and transport in metasedimentary lithology. Future studies may incorporate more extensive datasets including low-temperature thermochronometers, which may further elucidate the links among tectonics, erosion, and climate.

In the fourth chapter, I inferred the magnitude and spatial pattern of million-year timescale exhumation rates using five newly measured apatite (U-Th)/He sample ages from the Dibang Valley in the easternmost Himalaya. Fault activity along the active easternmost Himalayan range front is largely unconstrained over recent million-year timescales that are more relevant to our landslide and erosion rate analyses. Additionally, although the persistence of out-of-sequence faulting over recent million-year timescales has been proposed, little is known about the timing and magnitude of exhumation rates over this period. I determine plausible cooling histories using HeFTy inverse thermal modeling for two samples in both the range front and hinterland and infer exhumation rates over the last ~2-1 Ma assuming a simplified geothermal gradient. I observe that exhumation along the range front is highest along the Lalpani thrust and concentrated near the Lohit thrust in the hinterland. My findings potentially support persistent out-of-sequence faulting of the Lohit thrust that continues until the present though at a slower rate than that of the range front. Future studies might include additional thermochronology measurements in the range front to better constrain the spatial extent of high exhumation rates. Additionally, an improved understanding of the geothermal gradient would yield more accurate and reliable exhumation rate estimates.