Scaling Evapotranspiration Across Thermal Scales
How much of the variability in dryland ET disappears when you coarsen land surface temperature — and which of seven ET models notice.
Role
Lead author (poster), co-author (EGU oral)
Tools
Python SLURM NEON Google Earth Engine
Evapotranspiration is the largest terrestrial water flux after precipitation, and we mostly estimate it from land surface temperature observed at whatever resolution the satellite happens to offer. In drylands, where a single pixel can contain both a transpiring canopy and bare soil at forty degrees, that choice is not innocent.
This strand of my work asks how the variance in land surface temperature propagates into ET estimates as you move across scales — from centimetre-scale UAS thermal imagery up to the kilometre-scale pixels most models are actually run on.
Computer Vision in remote sensing
My primary hypothesis lies in using different convolusion filters (e.g. a wavelet decomposition) across different resolutions to narrow down the spatial patterns, clusters, components, and configurations of an ecosystem that affect the landscape heterogeneity. By inferring the ecosystem processes underlying these convolusion patterns and remote sensing products, we can inch closer to a mechanistic understanding of what natural processes and feedbacks affect the uncertainty in our evapotranspiration estimation across scales.
Model intercomparison
To test this properly, seven ET models are cross-validated against NEON tower sites: PT-JPL, PT-JPL-SM, STIC, MOD16, TSEB, SSEBop and GeeSEBAL. Each makes different assumptions about how surface temperature relates to water stress, so each degrades differently as resolution coarsens.
Assembling the meteorological forcing data for that comparison turned into a project in its own right — see the forcing pipeline.
Where it has been presented
- Caylor, K., Amin, S., Morgan, B., Trugman, A. (2025). Capturing Fine-Scale Variability in Dryland Evapotranspiration Through Multi-Scale Thermal Image Analysis. EGU General Assembly, Vienna. EGU25-16088.
- Amin, S., Morgan, B.E., Trugman, A.T., Caylor, K.K. (2024). Scaling Evapotranspiration: Impact of Land Surface Temperature Variance Across Ecosystems. AGU Fall Meeting, Washington, D.C. H33I-1037.