"Earth Provides Enough to Satisfy Every Man's Needs, but Not Every Man's Greed."
— Mahatma Gandhi
People have long realized that unsustainable exploitation of the planet is ultimately self-destructive. The humanity must seek for sustainable development strategies grounded in science to maintain harmony with the vulnerable Nature.
My goal is to understand the fundamental mechanisms of the hydrological processes at different spatial and temporal scales, and bridge the missing gaps between the latest scientific research progress and the practical water challenges faced by human society. I seek to answer questions linked with the Earth’s natural hydrological processes and human’s water use activities using the combined methods of data-driven and numerical modeling techniques.
I am now developing a global 5arcmin groundwater model H08-GM (v1.0).
We are also building a System Dynamics model to evaluate the human water use stress, and investigate the various feedbacks within the human-water nexus
Above: Natural water cycle and human water use diagram in Beijing in 2022, estimated from the SD model (unit: 0.1 billion m3).
Above: A simplified illustration of the Beijing SD model. Parameters labeled in yellow represent the main input variables; parameters labeled in blue represent the main output variables; and parameters enclosed in pink boxes have varying values under different SSP scenarios in the future projection (2024-2050). Each arrow denotes a causal relationship, with the polarity (+/-) indicating the positive or negative feedbacks between variables.
My PhD thesis concentrates on the key processes in hydrometeorology -- soil moisture and evapotranspiration. I use remote sensing data, and mathematical and numerical modeling approach to understand the role of flux exchanges in the land surface processes.
By using satellite and meteorological datasets, I developed a two-tiered framework based on soil moisture – evapotranspiration coupling regimes to characterize land aridity relevant to regional extreme events.
Regional climate and extreme events are often characterized by the total availability of land energy and water conditions (e.g., Aridity Index, ratio of precipitation to potential evapotranspiration), which does not account for land surface water-energy interactions that are crucially important for plant growth and human activities. Compared to classical aridity metric such as Aridity Index, this two-tiered classification scheme provides complementary information in terms of spatial heterogeneity and temporal variability of land surface. The method proposed here can serve as a framework for land-atmosphere coupling evaluation. See the figures below and more in doi: 10.1088/1748-9326/ac50d4
Regions having similar accumulated precipitation could show different final soil water availability (topleft); Regions with different accumulated precipitation could end with similar soil water content (bottom right). These phenomenon calls for a framework to complement the current definition of Aridity Index (AI), which has been done in this study (righthand side table)
Using satellite observations, I established a hybrid model to separately characterize the high-frequency surface hydrological processes such as drainage and Stage-I ET and the low-frequency Stage-II ET process.
The background is that traditional methods built on Markov processes often mix the two completely different processes. The model was then applied at the global scale using the satellite observed soil moisture data and is proven to be able to detect reasonable distribution of land surface hydrometeorological regimes. This work highlights the importance of improving temporal representativity in current hydrometeorological characterizations. The results can serve as a satellite-based reference for validation of surface water simulations. See the explanatory figures below and details in doi: 10.1175/JHM-D-18-0141.1
Soil memory refers to the recovery time from interruption of soil moiisture (panel (b) in top left figure). Delworth and Manabe (1988) has used 30-day soil moisture data to estimate the memory time. However, taking estimates from 1-minute resolution data as the true value, the 30-day Markov fitting line compares well to the samples, but is poorly fitted to the true line; Using 3-day data, the samples compare good to the true curve, however the Markov fitting line compares very poorly either for the samples or the true curve (bottom panels in lefthand side figure). A hybrid model has been proposed to solve the problem in terms of time scale mismatch between data and theory (righthand figure).
Based on the hybrid model and the two-tiered framework, I evaluated behaviors of several major Land Surface Models (LSMs) in characterizing soil moisture hydrological processes.
The results show that current LSMs have significant biases in simulating surface hydrological processes compared to SMAP estimation. Such biases may be contributed by inappropriate representation of critical soil parameters such as wilting point and critical point. These results provide a satellite-based reference for LSM development. See the below figure and details in 10.1029/2022EF003215
Lefthand side figure: Comparison of soil memory time between multi-model mean SMM from six major reanalysis datasets (i.e., GLDAS-Catchment, GLDAS-Noah, MERRA2, NCEP-FNL, ERA5, JRA55) and SMAP (Top panels), and soil memory times comparison of the model itself (bottom panel). Righthand side figure: Soil wilting point comparison between SMAP (a) and models (b).
Based on the framework I proposed and model evaluation results above, I developed an optimization method to calibrate soil parameters in LSMs based on satellite observations.
The calibrated soil parameters are proved to be useful to improve land surface simulations in an example LSM (i.e., Noah-MP). See the figures below. The published article is in press and doi will be updated soon here.
Lefthand figure: The optimization procedure to calibrate soil texture data based on Shuffled Complex Envolvment-University of Arizona (SCE-UA) algorithm. Righthand figure: comparison of the model simulated soil moisture using default and optimized soil texture maps.