"This place has changed a lot since I moved here."
"It's not what it used to be."
"You should have been here 10 years ago."
"You should have been here yesterday."
We knows these familiar laments of life in a resort town. Whether someone moved to Steamboat Springs in 1986 or 2006, the refrain is the same. It was at its best the day I moved here.
Do you see the paradox here? If we all want our sleepy mountain town to be the same as it was the day we moved there, and we moved there on different days, in different decades and different millenia, then we all have a different vision. The only commonality is that vision is squarely planted in the past. This phenomenon is so pervasive there is even a scientific name for it. It's called "last settler syndrome." Yes, there are scientists at universities whose entire career is built on researching the sociology of, well, us. And if you want to feel even less important, last settler syndrome was first described in 1977.
Of course, Steamboat has changed. Routt County has changed. I think that a common vision can rise out of the constant evolution of our surrounding landscape. And zooming out to look at the bigger picture might help put things in perspective. It feels like it's changed, but what do the data say? How much has it changed? From what to what? When? And why?
This month, we're diving into the National Land Cover Database, and I try my hand at answering a few of these questions with these notorious data.
The National Land Cover Database, 2016, Routt County, Colo.
Learn data science fundamentals by
following along with this post
DATASET: National Land Cover Database
LEARN: Raster analysis, animation,
for loops and treemaps
What is Routt County made of?
The National Land Cover Database is a robust and long-running satellite-derived spatial land cover classification database. In other words, it's a detailed map of what's on the land. It covers the entire United States, dates back as far as 1992 and is the basis for an astonishing amount of land change and ecology research.
The 2016 NLCD data say that this is what Routt County is made of: mostly forests. And shrubs.
Let's break it down a little more. NLCD has a standard set of 21 land cover classes that give us more specifics, like the kind of forests, and level of development.
I've spent hours looking at these charts. I like rolling them around in my mind and thinking about how this reflects what I've seen in all my years driving around the Yampa Valley. What I've seen but also what I've missed. That the human footprint is so small compared to the natural landscape is comforting. What surprises you about Routt County land cover? Inspires you?
The 2016 data is the most recent NLCD release (that's how long it takes the USGS to produce this amazing product!). When we look back in time, we can look at how these data compare to our perceptions of land change in the Yampa Valley.
We see the development and growth with our own eyes, every single day. The data back that up. While developed area is a relatively small percentage of Routt's land cover, medium intensity and high intensity development each increased 20% since 2001. That's astonishing. Most of that happened in Steamboat proper, which is depicted in that animation to the right. So why does this matter? In the NLCD, medium intensity development means at least 50% of that area is impervious surface - like concrete, roads and buildings. High intensity development is greater than 80% impervious. Water can't drain through impervious surfaces, so it quickly runs off until it finds a non-impervious surface like a nearby lawn it can soak into. Increased impervious surface has significant implications in watershed health and river flows. As we pave over more and more natural space, we need to keep an eye out for our beloved Yampa River. But impervious surface area isn't the only threat to the Yampa River watershed. Most developed area in Routt County is open space, and that class also grew a couple percent. Open space development includes areas like golf courses, single-family housing developments with large lawns and parks. And cultivated crops increased 72% (!!!). That's more than 6 square miles worth of new cropland. When water runs off of developed open space and agriculture, it takes with it all of the fertilizers and chemicals humans like to use. Yes, our footprint is expanding, but perhaps more importantly, we're using that land more intensely.
What the hell is happening to our forests?! Don't panic. I have a hypothesis as to what's causing this decline in forested area. You may have also noticed the increase in grassland area. Here's my humble opinion: I think that NLCD's classification process (which analyzes Landsat data, so this is the perfect follow-up to last month's post on Landsat imagery) is capturing beetle kill losses and subsequent fire or wind events that render conifer forests unrecognizable. With enough trees dead and downed, that might look a lot more like a grassland to an algorithm. As this cycle is part of a natural disturbance event (although likely aided and abetted by climate change), a lot of these trees will grow back over centuries. Deciduous trees like aspen will be the first to pop back up, then lodgepoles eventually shade them out and take over.
Routt County lost .4 square miles of wetlands. Just a couple percent. It may not sound like a lot, but it's a trend that we need to keep an eye on. Wetland loss is highly concerning to ecologists in the West. There are too many reasons why to list, but unlike forests, wetlands can't really come back. Once we lose them, they're gone. They're fragile and support some very special species that can't find what wetlands offer in any other ecosystem. They affect bird migrations and watershed health. Their collapse resounds through the rest of the landscape.
In just 16 years, the landscape changed dramatically, whether we saw that change occur or not. Whatever vision of Steamboat you're holding onto in your mind, watch these animations a few times and remember that it's going to keep doing that. For a while.
For more on working with NLCD data, how I came up with these graphs and numbers, and to try your hand at it yourself, see this month's tutorial on Github.