By coupling the global biosphere, hydrosphere, atmosphere, and cryosphere, Earth system models (ESMs) provide a unique testbed to disentangle the role of land heterogeneity in the global hydrologic cycle. However, existing modeling frameworks are not suitable for this task as they oversimplify the underlying drivers of spatial heterogeneity over land (e.g., topography). To address this challenge, our research develops and implements new conceptual frameworks that unlock the potential of hydrologic similarity to efficiently and effectively include the observed multi-scale land heterogeneity within macro-scale hydrologic models and ESMs. The following questions drive this research theme:
What are the primary drivers of land heterogeneity and how can they be efficiently and effectively represented in Earth system models?
How can hydrologic similarity be used to revisit the disconnect between field-scale and macro-scale hydrology?
This research theme was the motivation for the development of the HydroBlocks land surface model. Within HydroBlocks, sub-grid heterogeneity is defined through spatially interconnected clusters that are assembled via multivariate clustering of the proxies of the drivers of spatial heterogeneity (e.g., elevation and soil porosity) using global environmental datasets. Recently, these clustering concepts in HydroBlocks have been enhanced and included within the fourth generation of the Geophysical Fluid Dynamics Laboratory land model (GFDL LM4)—the land component of the GFDL ESM. Beyond the improvement of hydrologic models and ESMs, this research theme is leading to a novel conceptual framework that generalizes the concepts of hydrologic similarity; it has a strong potential to break the decades-long disconnect between field-scale, catchment-scale, and macro-scale hydrology. Moving forward, our research will extend these approaches on heterogeneity to include water management, urban areas, croplands, glaciers, and lakes, among others.
The ever-growing volume of environmental data from satellite remote sensing, field campaigns, and in-situ observation networks presents a unique opportunity to advance our understanding of the hydrologic cycle. However, the mere size and nature of these data create distinct challenges that limit our ability to synthesize and interpret its information. My research uses machine learning and high performance computing to harness these data sources to provide a more complete characterization of the observed heterogeneity. The following questions drive this research theme:
How can the petabytes of existing environmental data be harnessed and synthesized to more fully characterize the observed heterogeneity over land?
What are the primary landscape features that drive the spatial patterns of the water cycle? How do the characteristics of these features relate to local climate, geology, and topography?
One emphasis within this research theme is improving the description of the observed soil heterogeneity — a persistent weakness in hydrologic models; this led to the development of the POLARIS soil dataset. POLARIS synthesizes and harmonizes 100 years of disparate soil information to produce a probabilistic 30 meter soil series dataset over the contiguous United States; this dataset provides unprecedented horizontal and vertical probabilistic soil information. Another ongoing and future emphasis within this theme is to map over the globe the features over the landscape that play a critical role in the water cycle (e.g., channel networks and hillslopes); a challenge that has been ignored (and under appreciated) for too long in global hydrology and is indispensable towards improving our understanding of the water cycle.
Improving the representation of the physical environment within ESMs and synthesizing the petabytes of global environmental data makes it possible to investigate the role of heterogeneity in the hydrologic cycle in ways that were unthinkable before. To this end, our research uses the recently enhanced GFDL ESM to assess the sensitivity of the hydrologic cycle to multi-scale land heterogeneity at temporal scales ranging from hours to hundreds of years over the globe. The following questions drive this research theme:
What is the unique role of each driver of spatial heterogeneity in the terrestrial water cycle? How do these roles interact to explain the observed spatial structures of the water, energy, and biogeochemical cycles?
What is the sensitivity of the Earth system to the multi-scale land heterogeneity at temporal scales ranging from hours to decades? How does this improved characterization of heterogeneity impact weather and climate predictions?
Within this theme, we have previously explored how soil moisture spatial patterns can have deep ties to the fine-scale land cover, topography, soil, and meteorology. This was followed by another study where it was observed that hillslope characteristics can play a critical role in dampening hydrologic extremes in water limited regions. Due to the fine-scale coupling of the water and biogeochemistry within the model, this dampening also emerges in the carbon cycle; this suggests that in water-limited regions, strong topography can increase ecosystem resilience. Currently, the approach in this study is being used over the globe to run a suite of heterogeneity-enhanced offline (only land) and online (coupled Earth system) model experiments. Moving forward, our research within this theme will focus on using these and future model experiments to gain insight into the role of heterogeneity in land-atmosphere interactions, hydrologic extremes, and ecosystem resilience at local, regional, and global scales.