My research focuses on complex network approaches to understand ecohydrologic interactions, models, and water resources systems. Spatial scales in these systems range from a single flux tower that measures heat, water, and carbon fluxes between
the landscape and the atmosphere, to precipitation patterns across the United States. Temporal scales range from interactions that occur in seconds or minutes, to persistence across multiple seasons.
Below are several examples of research areas within my group.
Ecohydrologic interactions in the Critical Zone
This work seeks to explore interconnectivity within the “critical zone”, the highly active region of Earth’s surface between bedrock and canopy. We look at environmental time-series variables, such as temperature, humidity, and
heat fluxes, as interacting nodes in a network. Links between these nodes characterize complex dependencies. We explore how these variables are dependent on each other at different time scales and under different conditions. For
example, evapotranspiration depends on multiple drivers such as wind, water availability, and temperature, all of which also interact with each other.
Haiku version:
change a single thing
or shake up the whole order
is there a difference?
Precipitation persistence
This work addresses the predictability of precipitation occurrence, based on the knowledge of past precipitation at various locations. For example, if we know that it rained 3 days ago, how much does that reduce our uncertainty of whether it will
rain today? Similarly, if it rained directly west of us yesterday, does that help us predict whether it will rain today? These questions are related to the persistence of precipitation, or its patterns in both time and space.
Haiku version:
it rained yesterday
for today, we hope this means
less uncertainty
Casuality and modeling
This work addresses different aspects of how ecohydrological models behave. Rather than just finding a best performing model, we might want to know which model best represents a certain process, or how sensitive a model is to different combinations
of inputs or forcings.
Haiku version:
a model models
tangled webs of connections
…hard to untangle
Software:
Temporal Information Partitioning Networks (TIPNets), a Matlab GUI to characterize interactions using information theory-based measures, available on Github
