The rhododendron that can be tracked from space

In the alpine tundra of the Changbai Mountains in Northeast China, on the border with North Korea, climate has warmed significantly over the last few decades (at a rate of 0.28 °C/decade, from 1959 to 2017, to be precise). It’s a pity that we can’t go back in time to see how this has affected the mountain and its ecosystem! Or can we…?

A unique time series

In a new paper, led by Shengwei Zong and published just now in Remote Sensing and the Environment, we dug up an extraordinary time series of 54 (!) years of satellite data, starting with declassified KeyHole camera data from a US Defense satellite from 1963, and then all the way up to the most recent, high-resolution satellite imagery. The goal? Analyzing these images for changes in snow and vegetation cover on the mountain. For the latter, we focused specifically on a unique little shrub: Rhododendron aureum, an alpine shrub that conveniently stands out in autumn as it’s the only shrub in the region that keeps its green leaves when winter approaches.

Fig. 1
Changbai Mountains, on the border with North Korea, providing the back-drop for our study. Noteworthy is the top-right picture (A), where you can see patches of Rhododendron (in dark green) stand out from a brown autumn vegetation on the mountain.

Rhododendron in retreat

In short: we show that climate change has advanced the melting of snow in spring over the last 54 years of satellite data, while it resulted in a general upward retreat in the distribution of R. aureum over the last 30 years (of applicable satellite data). This makes sense, as the species needs a good spring snow cover to survive against cold temperatures! Take away the spring snow, and conditions get much harsher.

Fig. 10
We used our model – calibrated on the current modelled link between Rhododendron and snow cover (a), to predict its future distribution as climate keeps warming (b). While we already observe a small upward retreat now (not shown here), we expect significant further declines in area (c) and shifts in elevational optimum (d) in the future.

So, our study provides a good example of how climate change is actively affecting species distributions under our very own eyes (or at least those of our satellites), in this case through changes in snow cover. We can expect those changes to be happening all around us. However, as long-term biodiversity monitoring is still rare and far-between, these changes are often hard to proof.

Complexity abound

We here show that the increasingly long-term satellite record can provide a great alternative to long-term surveys on the ground to record such changes over decadal periods. However, and we can’t stress this enough, accurately reading satellite data and getting this information out of them is NOT easy. You’ll see this if you browse through the paper, which has an impressive methods section (cheers and applause for Shengwei Zong for persisting with the analyses), and has gone through a few rounds of deep peer review helping us to clean up loose ends and making the methodology more air-tight.

Fig. 2
Did I say complexity? This figure gives an overview of the methodological steps taken to extract Rhododendron aureum (RA) distribution data and link it to environmental data through species distribution models.

Just to give one example of the complexities: in the Changbai Mountains, we were pretty lucky to have fairly large patches of one specific species – the Rhododendron of interest – that stand out on satellite images in autumn, while the rest of the vegetation is green. And even then, it is far from straightforward to tell your algorithm which color of green represents your species of interest, and which is noise, as often enough there are other plants intermingled with the Rhododendron.

Fig. 4
Linking the ‘greenness value’ of the satellite images (here called ‘NDVI’) to distinguish between a large cover of herbs (brown line) or Rhododendron (green line) in autumn

And then I’m not even talking yet about the aligning of images from different satellite sources, which gets especially tricky when one dives back as deep in time as the KeyHole camera data from back in the days when accurately monitoring environmental change was most certainly not the main goal of the imagery.

In short, we hope this rather complex methodological paper holds lessons for ecologists and remote sensing specialists in the future, and can help put us further on track to use satellite data to analyze vegetation change in these times of increasingly rapid global changes.

More details:

Zong et al. (2021) Upward range shift of a dominant alpine shrub related to 50 years of snow cover change. Remote Sensing of Environment

This entry was posted in China and tagged , , , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s