Estimating Arctic Tundra Soil Water Content Variability and Relationship to Landscape Properties Using Above- and Below-Ground Imaging

Abstract:

Estimating the spatiotemporal distribution of soil water content is crucial for ecosystem understanding because of its strong influence on vegetation dynamics, surface-subsurface energy exchange, water and heat flux, microbial activity and biogeochemical mechanisms. In particular, quantifying Arctic ecosystem feedbacks to climate requires advances in monitoring soil moisture in sufficiently high resolution over modeling-relevant scales, and in understanding its coupling with landscape and soil characteristics.

As part of the DOE Next-Generation Ecosystem Experiments (NGEE-Arctic) we investigated diverse strategies to estimate the soil water content spatial distribution in Arctic tundra using various proxies including electrical conductivity from electrical resistivity tomography (ERT) and from electromagnetic induction imaging, dielectric permittivity from time-domain reflectometery (TDR), and vegetation index from low-altitude multi-spectral imaging. In addition to occasional campaigns along a 500x40 m corridor, high temporal resolution is achieved through continuous monitoring of surface and subsurface dynamics along a 35 m transect using ERT, temperature loggers, TDR, and visible and NIR imaging from pole-based cameras. This was conducted at the NGEE Barrow, AK site.

The results of this study inform on the complementary nature of and trade-offs between various approaches with regard to accuracy, resolution and coverage. In short, while point-scale measurements are relatively hard data, ERT data improves spatiotemporal monitoring of soil water content and state, and low-altitude aerial imaging can be used to extend predictions at larger scale using field-dependent relationships. Importantly, this study enables the identification of spatiotemporal links between soil and landscape properties (such as water inundation, vegetation, topography, thaw layer thickness, water content, temperature, and snow thickness). While some of these properties show strong co-variability, other features show more spatially and temporally variable relations or more complex linkages. Overall, identifying such links is crucial for extrapolating strong knowledge at specific sites over larger scales and for improving parameterization of models simulating ecosystem feedbacks to climate.