Relating tree water content and satellite SAR data in boreal forests

Forest water dynamics greatly influences the earth’s climate, weather patterns, ecosystems and biodiversity. With a changing climate, the occurrence of forest fires, insect outbreaks and windthrows are expected to increase.

Information about tree water content is essential to monitor forest vitality, understand on-going changes and mitigate negative consequences.

Currently, there is no method for estimating tree water content variations over large spatial scales with adequate temporal resolution.

Optical techniques are limited by clouds, require sun illumination and are not directly sensitive to water in bulk tree structures.

The only viable solution is to use microwaves and earth observation from space.Microwave remote sensing is known to be directly sensitive to plant water due to the high dielectric constant of liquid water.

In the proposed project, we will use active microwave techniques from space which can provide regional and global spatial coverage together with fine resolution using synthetic aperture radar (SAR).

New satellite SAR systems, with improved spatial coverage, offer opportunities for providing the needed tree water data.

Although it is known that SAR is sensitive to tree water content, there is a lack of understanding of the underlying processes and research is needed to establish how forest tree water is related to, and can be estimated from, satellite SAR data.The goal of the project is to understand, develop forward and inverse models, and establish the link between satellite SAR data and tree water content.

We will make combined use of data collected by satellite SAR systems, tower-based radars, an eddy covariance flux tower and in-situ sensors.

We will use satellite data from current and future SAR satellites operating at L-band (25 cm wavelength) and C-band (5 cm wavelength).

Most importantly, the project will use data from the tower-based BorealScat radar, which has produced time series at 5 min intervals since 2017 at a test site in southern Sweden and will be moved to northern Sweden in 2021.

The radar tower will be relocated to the Svartberget experimental forest, Sweden’s most well-studied forest site in terms of hydrology, tree water relations and forest-atmosphere water exchange.

BorealScat is an advanced multi-frequency radar which produces information about different scattering mechanisms from ground and up throughout the canopy.

Along with data from meteorological sensors and cameras, the forest water status will be quantified using soil moisture sensors, the flux tower measurements and novel tree sensors (sap flow sensors and digital point dendrometers).Based on the available data, the first step in the project is to develop forward models which are able to predict backscatter data observed with the satellite SAR systems and BorealScat.

The forward models will include tree physiology and build on existing tree water dynamics models. Input data to the models are data from the weather station, in-situ sensors and flux tower. The second step is to develop inverse models, i.e. that estimate tree water content from the satellite SAR.

This step will be more challenging than the first step since the temporal sampling (revisit time) of the satellite data is coarser (longer). We will use dual-polarization SAR data in the inverse model as well as meteorological data.

Two different approaches for inferring water content will be investigated and evaluated, based on hidden Markov and Bayesian models, respectively.The proposed project is for a four-year PhD student and is part of a collaboration between Chalmers University of Technology and the Swedish University of Agricultural Sciences (SLU) in Umeå.

The PhD student will be supervised by a team of leading experts in radar remote sensing of forests, satellite SAR, tower-based radars, forest radar scattering and plant physiology.