Ecohydrology research | Environmental Data Scientist | Interest in research in support of informed policy and decision making
I'm postdoctoral researcher at Duke University.
My research interests span a range of disciplines – dryland ecohydrology, fire ecology, and climate effects of forest loss.
I aim to bring together observational data and modelling approaches – spanning simple 'toy' models, physics-based hydrological modelling and machine learning / data science – to develop new methodologies for data sparse environments and learn from available observations.
My current research aims to understand the physics of storm runoff in dryland environments, and how runoff at the timescales of individual storms relates to growth and development of plant communities on yearly to decadal timescales.
Selected article: Crompton, O. V., & Thompson, S. E. (2021). Sensitivity of dryland vegetation patterns to storm characteristics. Ecohydrology, 14(2), e2269.
Vegetation patterns and dryland vulnerability
Drylands – including arid and semiarid regions –make up 40% of the earth's land surface and support some 2 billion people. As the climate warms, these regions are increasingly vulnerable to land degradation and desertification.
Ecology and hydrology are tightly coupled in these water-limited landscapes – storms generate overland flow, which redistributes water from bare soil to vegetated areas, providing additional inputs that sustain vegetation. Land degradation may flip the hydrology from resource capturing to resource shedding, where overland flow remove water and resources from the landscape.
To study dryland degradation from a hydrodynamic perspective, my research uses hydrological models to study how overland flow mediates threshold-like behaviour between capturing and shedding states. For example, how can readily available information about vegetation patterns and storm climatology be used to assess ecosystem productivity and health?
Hydrodynamics of overland flow in patchy landscapes
Describing resistance to overland flow in drylands is challenging because these flows traverse complex terrain with spatially varying surface roughness and infiltration rates. For example, infiltration rates are typically low in bare-ground areas due to the formation of surface crusts and higher under vegetation due to root activity and protection of the soil surface against rain-splash by the canopy.
While a lot is known about flow resistance through vegetation canopies in channel flow settings, storm runoff in drylands is different: (i) bed resistance cannot be neglected for the very shallow flows generated by storm runoff, and (ii) spatial patterns in roughness and infiltration rates between vegetated and bare soil areas means that effective resistance observed at hillslope scales differs from local flow resistance.
My research addresses fundamental questions about overland flow in patchily vegetated environments:
Can easily observable data – such as a hillslope hydrograph – be inverted to infer the partitioning of resistance between canopy drag and bed shear stress? How do contrasts in resistance and permeability between bare soil and vegetated areas influence the resulting flow? Under what conditions does the greater resistance provided by vegetation versus the contrast in infiltration rates control the flow dynamics?
Selected article: Crompton, O., Katul, G. G., & Thompson, S. (2020). Resistance formulations in shallow overland flow along a hillslope covered with patchy vegetation. Water Resources Research, 56(5)
Physically-informed data science: temperature effects of tropical forest loss
Like natural evaporative air conditioners, tropical forests lower surface temperatures, and deforestation-induced surface warming is known to extend beyond deforested zones to undisturbed forests. In tropical countries such as Indonesia, Brazil and the Congo, rapid deforestation may have accounted for up to 75% of the observed surface warming between 1950 and 2010. With more than 40% of the world’s population living in the tropics, keeping forests intact is vital to protect those who live in and around them as the planet warms.
This raises a number of policy-relevant questions: If forest conversion is to take place, it is possible to plan that conversion to have smaller warming effects? Can the cooling provided by intact forest to surrounding farmland help to make a `business case’ for preserving intact forest?
To answer these questions, this research uses satellite data over Indonesia, Malaysia and Papua New Guinea to ask how contextual factors –i.e, the spatial extent of forest loss and its level of fragmentation – influence observed surface warming.
Crompton, O., Corrêa, D., Duncan, J. A., & Thompson, S. E. (2021). Deforestation-induced surface warming is influenced by the fragmentation and spatial extent of forest loss in Maritime Southeast Asia. Environmental Research Letters
See our article in The Conversation!
Interactive web app applying study findings: https://treeheat.azurewebsites.net/
Areas of forest cleared for oil palm plantations, in Bawa village, Indonesia. Indonesia is the world’s largest producer of palm oil. EPA/HOTLI SIMANJUNTAK
Selected article: Crompton, O., Corrêa, D., Duncan, J. A., & Thompson, S. E. (2021). Deforestation-induced surface warming is influenced by the fragmentation and spatial extent of forest loss in Maritime Southeast Asia. Environmental Research Letters.
Wildfire restoration in the Sierra Nevada mountains
Devastating fire seasons have become a new normal in California, due in part to the lack of preventative measures like low-intensity fuel removal controlled burns, lack of thinning, and a long-term history of fire suppression. Worldwide, fire severity and extent are increasing, due to climate change. The use of managed wildfire to reduce these hazards is a globally relevant strategy for forest management. The implications of manipulating or restoring forest fire regimes, however, are challenging to determine because of the scarcity of sites subject to long-term fire regime restoration. Natural wildfire has been restored to only a handful of areas in the Sierra Nevada, and the long term consequences of this restoration are not fully understood.
This research synthesises long-term research conducted in the Illilouette Creek Basin (ICB) and Sugarloaf Creek Basin (SCB), where fire restoration was implemented in the late 1960s/early 1970s. To examine the long term consequences of this restoration, I used a simple dynamical model to explore the forest cover dynamics for these basins, finding that forest cover has not yet reached steady state in ICB.