Site icon Conservation news

Scientists predict tree death from drought in California’s Sierra Nevadas

A dead tree in the Sierra Nevadas in California. Photo by Rhett A. Butler.

A dead tree in the Sierra Nevadas in California. Photo by Rhett A. Butler.

  • A study in the Sierra Nevada mountains in California shows that remotely-measured changes in the canopy water content (CWC) of conifers can be used to forecast tree mortality.
  • Water content in tree canopies can be remotely monitored using laser-based images from aerial surveys.
  • Changes in the CWC in conifer forests during droughts correlate well with tree mortality.
  • After estimating canopy water content from past years using a deep learning model, researchers were able to accurately predict tree death during a recent drought.

As the world heads toward more frequent and severe droughts, forests will increasingly suffer from water scarcity. In this scenario, finding ways to predict how trees will respond to water stress is becoming increasingly important.

It is now possible to look at large swaths of forests in incredible detail using aerial and satellite images. The technology goes well beyond simply monitoring deforestation. Remote sensing tools can be used to tell different tree species apart that live in the same area or to measure specifics such as the chemical composition of their leaves. Many of these features are useful to describe how droughts are impacting forests. However, we still lack models that can predict future tree mortality based on past events.

In a recently published study, scientists Phil Brodrick and Greg Asner show that changes in the amount of water in the canopy of conifers in Sierra Nevada over the course of five years correlate well with tree mortality rates during the 2016 drought.

“This work is unique in that we show how canopy water content can be used to anticipate tree mortality a year before it occurs,” Brodrick wrote by email. “As you might imagine, this advance warning could be of significant use to forest managers, conservation groups and policy makers that may be interested in reacting to drought effects.”

Brodrick and Asner work in the Department of Global Ecology of the Carnegie Institution for Science, in Stanford. In their study they used images taken with the Carnegie Airborne Observatory, a modified aircraft equipped with laser-based sensors for remote monitoring. Flying over the conifer forests of Sierra Nevada in 2015 and 2016, these sensors allowed them to measure the water content of the canopies and to identify dead trees over an extension of more than 26,000 hectares.

The Sierra Nevadas in California. Photo by Rhett A. Butler.
The Sierra Nevadas in California. Photo by Rhett A. Butler.

Since they lacked similar images from previous years, they had to estimate the canopy water content between 2010 and 2015. To do this, they first gathered a large data set from satellites and other geographic information systems covering the area over the whole period. Then, they used a deep learning model that compared the geographic information from 2015 and 2016 with the canopy water content measures collected during the flights. After this “training,” as it is called in the scientific jargon, the model learns how to correlate these two data sets and is able to estimate canopy water content simply by feeding it with geographic information.

When the figures from the time series of canopy water content are sided with those of tree mortality during the 2016 drought, a pattern emerges: the areas where trees have less water in their canopies and where the canopies lose more water over time are also the areas with higher death rates. 

Several details were considered during the process. For example, locations with reported fires between 2010 and 2017 were excluded from the analysis. Also, the study focused specifically on conifers, such as pine (Pinus sp), juniper (Juniperus sp) and cypress (Family Cupressaceae).

“We were careful to limit the scope of this work to conifers, as some species (several types of oaks, for example) have the potential to recover after browning or losing leaves,” Brodrick said.  These changes in the number and color of the leaves, which happen naturally over the life cycle of many trees, but not conifers, could affect the measures and confound the results.

However, according to Brodrick, the approach used here could be applied to evaluate the vulnerability to drought in other types of forests, although it would be necessary to tweak some of the parameters.

An interesting place to further test the model could be the Amazon Basin, which has suffered intense droughts in 2005, 2010 and 2015-2016. Asner, who has worked extensively in the area, believes that, should longer droughts occur, this methodology could be useful.

“If we start getting more frequent drought in the Amazon, I am very confident that we will be able to use airborne canopy water content monitoring to track tree stress and to predict where trees will die,” Asner said, also over email.

The Sierra Nevadas in California. Photo by Rhett A. Butler.
The Sierra Nevadas in California. Photo by Rhett A. Butler.

For the moment, a limitation to their approach is the need to fly over the area with the Carnegie Airborne Observatory. However, that could change in the near future, as Asner is currently raising funds to put the technology to orbit, replacing aircraft with satellites.

“Once in orbit, we will be able to monitor canopy water content and a wide variety of other forest properties every two weeks, everywhere on the planet,” he said.

Banner image: A dead tree in California’s Sierra Nevadas. Photo by Rhett A. Butler.

Ignacio Amigo is a freelance journalist based in São Paulo, Brazil. You can find him on Twitter at @IgnacioAmigoH.

Citation:

Brodrick, P.G., Asner, G.P. (2017) Remotely sensed predictors of conifer tree mortality during severe drought. Environmental Research Letters, 12, p.115013

FEEDBACK: Use this form to send a message to the author of this post. If you want to post a public comment, you can do that at the bottom of the page.

Exit mobile version