Ecohydrology, hydroclimatology, and biometeorology
Evapotranspiration and land-atmosphere coupling
Evapotranspiration, which is the sum of soil evaporation and plant transpiration, plays a critical role in the terrestrial water cycle, energy budget, and carbon cycle. Understanding evapotranspiration is essential in assessing drought, plant water stress, and wildfire risk. However, estimating or predicting evapotranspiration is challenging due to the complex nature of the land surface and physiological effects such as stomatal closure of vegetation under water stress.
Recently, a novel theory (called SFE) has been proposed suggesting that near-surface atmospheric conditions can approximate evapotranspiration without knowing the land surface conditions (McColl et al., 2019). In my recent study, I demonstrated how this land-atmosphere coupling theory can be understood from turbulent diffusion processes and proposed a new physical equation that expresses evapotranspiration (Kim et al., 2021).
However, the original SFE theory has limitations in dry conditions where land-atmosphere equilibrium is difficult to achieve. To overcome this challenge, I have extended the theory and developed a new model called SFE-MEP. The below figure illustrates the results of my study, which shows that the SFE-MEP model can improve evapotranspiration prediction, particularly in dry regions, compared to the original SFE theory (Kim et al., 2023a).
Overall, my research provides new insights into the complex process of evapotranspiration and offers a more accurate method for estimating and predicting evapotranspiration, which has significant implications for managing water resources and assessing environmental risks.
Hydrologic responses to climate change
Understanding the hydrologic responses to anthropogenic climate change is critical for climate change adaptation. However, current knowledge on the differential changes in runoff, soil moisture, and near-surface humidity is insufficient. This knowledge gap has led to ongoing scientific debates regarding hydrologic changes under anthropogenic climate change, hindering appropriate climate change adaptation. For instance, there are long-lasting debates on the role of evaporative demand (or potential evaporation) under warming climatic conditions.
In my recent study, I proposed a novel approach to potential evaporation based on land-atmosphere coupling, which can be applied in warming climatic conditions (Kim et al., 2023b). The proposed approach provides a more accurate representation of hydrologic impacts, and it aligns well with climate simulations, suggesting the significant role of land-atmosphere coupling in hydroclimatic changes.
Furthermore, I have extended my theory and developed a simple and robust physical model that can explain changes in evapotranspiration using changes in the atmospheric state under anthropogenic climate change (Kim & Johnson, in preparation). This new model provides a more comprehensive understanding of the hydrologic responses to climate change, which is crucial for developing appropriate climate change adaptation strategies.
In summary, my research contributes to understanding the hydrologic responses to anthropogenic climate change and offers a more accurate method for predicting and adapting to the impacts of climate change on hydrology.
Machine learning applications
In the field of geoscience, machine learning algorithms, such as artificial neural networks and deep learning, have gained popularity.
For example, field observations may sometimes contain gaps due to unfavourable conditions that should be modelled statistically. Machine learning algorithms are particularly useful when the target variable is nonlinearly related to environmental conditions. In my previous study, I evaluated various machine learning algorithms and a conventional approach to identify the best practice for processing methane flux observations using the eddy covariance method (Kim et al., 2020).
However, despite their usefulness in geoscience, machine learning algorithms do not always conform to physical principles, limiting their generalizability and interoperability. To overcome these limitations, hybrid models that combine physical processes with machine learning approaches have recently gained attention. As part of this important progress, I am currently developing and evaluating hybrid evapotranspiration models using satellite observations.