Hyunglok Kim bio photo

Hyunglok Kim

Terrestrial Water Cycle
Remote Sensing
Bayesian Inference
Machine Learning
Deep Learning

Email Google Scholar LinkedIn Instagram Github ResearchGate Orcid

1 Research Background

Recent studies show that the terrestrial water cycle can be homogenized by human-driven modifications in land cover due to changes the quantities of water flow. Interpreting soil moisture variability, especially soil moisture memory, allows us to analyze the interaction between the land surface and low atmosphere because soil moisture has a longer memory than other hydrogeological variables. However, discontinuity and irregular sampling in soil moisture observations from sun-synchronous-orbit satellites limit calculations of soil moisture memory at different time scales, thus hampering our understanding of the fundamental processes that control the global terrestrial water cycle. Using numerous sources of remotely-sensed data with land surface models can be helpful in overcoming this limitation.

2 Research Objectives

Here, I assimilated frequent soil moisture observations from NASA’s recent constellation of eight micro-satellites, the Cyclone Global Navigation Satellite System (CYGNSS); and a sun-synchronous satellite, Soil Moisture Active Passive (SMAP), into a land surface model for use in analyzing the spatial and temporal dynamics of soil moisture memory. We calculate the average fraction of precipitation water using the stored precipitation fraction metric. Furthermore, in order to analyze the impact of environmental and atmospheric variables on the global water cycle at different time scales, I employed a sampling-based approach (i.e., the Hamiltonian Monte Carlo Markov Chain) and an approximation-based approach (i.e., variational inferences) with Bayesian multiple regression. In this study, I showed how different land cover types, environmental conditions, and ecosystems impact the soil moisture memory. This study can add value to our understanding of diurnal dynamics of soil moisture memory on a global scale.