Eyes on Earth Episode 50 – Delaware River Basin

JOHN HULT:

Hello everyone and welcome to another episode of Eyes on Earth. We're a podcast that focuses on our ever-changing Earth. And on the people here at EROS and across the globe who use remote sensing to monitor and study the health of Earth. I'm your host John Hult.

The Delaware River Basin is spread across four states on the East Coast. And supplies water to 15 million people. That represents the largest inter-basin withdrawal of water east of the Mississippi River. Climate change and the socioeconomic response to it, will alter lives, livelihoods, and landscapes in the basin over the next eighty years. And teams at EROS were among those working to understand what that might look like. EROS is home to a project called Forecasting Scenarios of Landscape Change or FORE-SCE for short. The project models change at Landsat satellite scale by considering factors like historical landcover and land use patterns, population growth, climate change, industrial shifts and more. Now the group recently released a GIS data set that looks both forward and backwards in time. To help understand the dynamics of the Delaware River Basin. The work was part of a larger collaborative effort across the USGS to refine projections of change by pooling data sources to build better models. On today's episode of Eyes on Earth, we are talking with Terry Sohl, the Integrated Science & Applications Branch Chief at EROS and the head of the FORE-SCE Team. We are also chatting with Jordan Dornbierer, an EROS contractor who is the lead author of the Delaware River Basin Data Set. Terry, Jordan welcome to Eyes on Earth.

TERRY SOHL:

Thank you, John. Thanks for the opportunity.

JORDAN DORNBIERER:

Thanks for having us.

HULT:

Terry let's start with you. Let's talk a little about FORE-SCE in general terms. What is it? What are you doing with FORE-SCE? When did you begin doing this and where have you applied it in the past, like before the Delaware River Basin?

SOHL:

We've been doing this awhile now. It goes back to a NASA proposal that was funded in 2005 or 2006. The whole idea at the time was to try to evaluate feedbacks between climate change and land use change. And we wanted to do it not just looking at remote sensing data, but we also wanted to look into the past and into the future. So that's the first time that we really realized that there's a lot of applications that have a need for land use and land cover that goes beyond the satellite era. So is there a way that we can take a modeling approach and extend those data both backward and forward in time. We're trying to enable analyses of climate change, of biodiversity, hydrology, health, and human welfare issues. That are all reliant on these very long-term time series that we can help model and foresee. We completed the Great Plains right before the Delaware River Basin. So that's the entirety of the Great Plains. We've modeled the southeastern United States. The Pyrenees and the border of Spain and France. We actually had a PhD student come over from France for a summer and we helped her to model the Great Pyrenees. We've also modeled on a national scale with a land carbon project. What we try and do is make it an approach that's applicable regardless of scale from regional to national.

HULT:

So, you have actually done relatively large chunks of land before. The Delaware River Basin was not way off the mark from the kinds of scales, regional scales that you had worked in the past.

SOHL:

Oh no. Not at all. The Delaware River Basin was kind of a target of opportunity if you will. We have the capability to model at regional all the way to national scales. Depending on the resources. The way that we treat the model, we kind of look at it as a puzzle. We're trying to account for all these vast number of factors that drive land-use change. We can't tackle them all at once. You kind of have to do them one at a time. The Delaware River Basin was a nice opportunity to try to add a couple more pieces to that puzzle that we have not tackled before. Particularly related to urban change, water use and water availability. We knew that we needed to beef up in the model.

HULT:

Jordan, let me turn to you on this question. I'm interested to know what kinds of things you were building into this model for the Delaware River Basin. What are we looking at here? Urban growth, agriculture, industry, water ... what are some of the things you had to put into this model?

DORNBIERER:

Well, as far as some of the newer things that we were looking at urban growth and generally anthropogenic change. So, any change in the landscape due to human activity. We had found fairly recently, when we moved from a 250-meter resolution to a 30-meter resolution that the fidelity of landscape patterns we're representing could be improved in some ways. We want to match what you would see in NLCD or LCMAP as precisely as we can.

HULT:

So NLCD (National Land Cover Database), LCMAP (Land Change Monitoring Assessment Projection). Land cover data sets. Land cover mapping projects at EROS. You kind of want to mirror that, is that right?

DORNBIERER:

That is exactly right. We take those contemporary land cover classification data sets, those are our starting points. And then we try to extend that paradigm into the future or into the past. FORE-SCE is really good at doing something that we call meeting demand. So, you can tell it, "I want to see this much change in the Delaware River Basin of urban that would have reverted to agricultural land moving back into the past." And it will do that. But to do it in the right place and with the right pattern is more challenging.

HULT:

Terry, is there anything that you want to add to Jordan's answer there, in terms of what you saw in the Delaware?

SOHL:

One of the things that we're trying to do is answer some specific questions in the region. And the improvements with regard to the urban modeling as well as some improvements in how we represent the intersection of water use, water availability and land use ... everything that was done there facilitates those types of questions. For an area like the Delaware River Basin, as you mentioned, John, There are 15 million people that depend on drinking water in that region. And one of the issues that we were asked to look at was drought. Back in the 1960s there was a very severe drought. It really reduced the flow of the Delaware River. When the flow of the Delaware River is reduced, that allows salt water in the estuary to kind of creep upstream. It got very, very close to the freshwater intakes of the cities of Philadelphia, Camden. Some of the big cities that really rely on that river for freshwater. And so one of the questions that we're hoping to answer is, "what if that drought happened now?" With a new landscape, with a few million more people depending on that Delaware River for water, with a hotter and potentially changed climate. Under that scenario, what if that drought hit again? Would we be able to handle that? All the model improvements that Jordan tries to tackle with the FORE-SCE model are critical for being able to try and address those questions.

HULT:

So, looking into the future, you have to be aware of what happened in the past. And this is going to help answer some of those questions. This is a possibility and we know that because it has happened in the past. We can model into the future. And that is what you are trying to get at. I mean, maybe initially I thought you guys were just doing this for fun ... but that's what you're after, right? You are looking to answer questions that are going to have an impact on people's lives and the ecosystems.

SOHL:

Oh yeah. Most definitely.

HULT:

So, this data. This goes back hundreds of years. Right, Jordan? This goes back much further then Landsat. How do you look into the past, in the pre-satellite era? How does that work?

DORNBIERER:

We look at as much historical data for the region as we can get. We have a guy on the team who specializes in this. He will dig up historical data sources, and then we will use those data sources to create, I'm going to use that word again, demand. It's kind of the word we use inside FORE-SCE for a prescription of change that we'll feed into the model. For a historical reconstruction we'll create demand. And in this case, we want to go back to pre-settlement, which we determined to be approximately 1680, and every land cover class represented on the landscape, gets a prescription.

HULT:

So, demand is basically you guys telling the model how much change there needs to be in any particular kind of land cover class.

DORNBIERER:

Yeah.

HULT:

Ok. Got it.

DORNBIERER:

And if you plotted that out, you would see that the agricultural classes rise up going back in time. When the economy was much more agricultural than industrial. You see the urban land cover diminish going back in time, and you see a lot of it displaced by forest. Going into the future, we are often creating what looks like a spaghetti graph, under different climate scenarios. It's a stark contrast to the kind of simulations we are attempting to create going into the past. We want to converge on what was most likely to be there historically-not a variety of histories. Given that, unless you are into multiple universes-which we're not doing on this project-we are going for just one history and we are trying to nail it.

HULT:

You're trying to get what was actually there. Just get it right one time. Because there is only one history that actually occurred. 

DORNBRIERER:

That's right.

HULT:

But the future is wide open.

DORNBIERER:

Yeah, as far as we know. We might as well think about the future as multiple universes. A lot of different things could happen. We try and pick the one that's going to be it. You're really making yourself vulnerable to being wrong. Going into the past, we have the one. So, we are using as much data as we can to kind of converge on that. I won't go thru the actual data sets besides saying that the historical census both population and agriculture are big ones. And then we turned up some new ones. There was a group that created a historical settlement data that was spatially explicit. That was quite helpful.

HULT: 

Interesting. Terry, what do you want to add to that one?

SOHL:

When we get requests for applications of FORE-SCE, half the requests are for future projections and about half refer to historical landscape reconstructions. What's most exciting to me about this work is, we've modeled to the past before, but only back to the 1930s. This is really the first time that we've had the opportunity to go all the way back to pre-settlement. Before there was a lot of European impact on the landscape. Seeing those fluxes in agriculture, forests and wetland and urban land ... it's just such a dramatic story of how impactful human beings can be on the landscape. 

HULT: 

And that's of course what you're looking at going forward. Let's talk about that a little bit. Let's talk about looking forward. What sorts of data were you looking at?

SOHL:

We're using that scenario, like Jordan said, to look at this spaghetti-like graphic of the future, where the uncertainties are very high. So for an individual land cover class like cropland it may go up. It may go up a lot. It may stay stable. It may decline. It all depends on the scenario. Whether economic conditions are suitable for cropland or if the demographic demand is there for those agricultural commodities. The challenge is trying to represent all of those driving forces. All the way from climate to local policy. To soil conditions, topography. It really is a very large set of data sets that go into the scenarios. But we do have a specific suite of scenarios that we have run both for the Great Plains and for the Delaware River Basin. I don't know if this is where Jordan, you would want to jump in.

DORNBRIERER:

Yes. I can certainly elaborate. You've the Intergovernmental Panel on Climate Change. You've got groups of experts taking all these different variables-policy, demographics, climate, economics in their domain of expertise-and bringing them to bear on global models, often called Integrated assessment models, that look at different future pathways that the planet could take. One of these models, these Integrated assessment models that we have used consistently and goes into our recent work is the Global Change assessment model. We used three of their versions under different climate change mitigation potentials. One where nobody is trying. There is no mitigation. And then two different levels of representative concentration pathways.

HULT:

You're looking at scenarios wherein climate change is happening. And the people on the ground either do nothing at all, a moderate amount of response or they do a lot to deal with that. Is that right?

DORNBIERER:

Yes. That's going to have impacts on agricultural expansion or contraction. On population expansion or contraction. Will cities grow or will the population maybe level out? Maybe you see more suburban and exurban growth. And we are using expert analysis and taking that into our model. We aren't making those decisions ourselves. So another set of scenarios is based on the Department of Energy's Billion-Ton Report. And that's asking the question, if the United States was to produce one billion tons of dry biomass annually, using biomass for energy and fuels and other product and displacing, they're estimating, approximately 30% of reliance on fossil fuels for those same things ... If they were to do that, where would that biomass come from? So, this is agriculture and economics driven. And then the final one that we did, just to mention it, is a business as usual. Pretty straightforward. We just extrapolate trends from the existing chunk of contemporary land cover data sets.

HULT:

How might this guide research and land management to know what might happen in 2100. And separately, how do we know that we can trust this?

SOHL:

That's a great question. Particularly that last question, John. You know from a trust perspective because this is scenario based. One of the beauties of doing scenario-based modeling, is that you can never be told that your wrong. The scenario is dependent on some very specifically defined conditions, and so when we run an IPCC scenario like Jordan was talking about, we have specific assumptions that are related to that from a climate perspective, from an economic perspective, from a demographic perspective. And what we're saying simply is, if those combinations of driving forces occur, this is what's going to happen to the landscape. The problem is it's very difficult to predict how each one of those individual driving forces are going to act in the future. Much less how they are going to relate to each other. That's why we do the multiple scenario approach. From the standpoint of having information that are useful, I look to the Delaware River Basin and some of the work of the Delaware River Master. The Delaware River Master, coolest title in the world by the way, was established after the droughts of the 1960s to provide some guidance for management of the Delaware River and the reservoirs. The releases of those reservoirs which have a big impact on a wide variety of processes, including drinking water and others. Kendra Russell, the Delaware River Master, is the one that tries to manage those water levels. In cooperation with the various stakeholders in the region. By informing the future of landscape and climate change, we can look at the interplay between climate, between land use, water use, water availability. How those all interact in specific scenarios. And so, from a planning perspective. If we do go back to the conditions of that drought in the 1960's, where by 2050 we have 20 million people that are dependent on that basin for drinking water instead of 15 million. Can the basin handle it? It's not just from a drinking water perspective. Whether it's irrigated agriculture that's dependent on the Delaware River. From an ecological perspective, there's temperatures differences that are caused by the management of those reservoirs. From a standpoint of energy development, those dams produce hydroelectric power. So trying to inform how climate change and land use impacts river flows, provides information on how much water is going to be available for energy development.

HULT: 

Any closing thoughts? Do you want to talk a little bit about anything that you are looking at in the future?

SOHL: 

One thing we are looking right now is something that just came out today from the Biden administration. They release a report called America the Beautiful. And it outlines a vision for what is often called 30 by 30. So that's an effort to conserve thirty percent of the lands and waters by 2030. And it has some very ambitious targets. What this document outlines is an approach that helps us work with local conservation groups. Helps the government partner with other governmental agencies. To try to identify an approach to conserve that amount of landscape by 2030. A lot of work that we're doing in the Delaware in the coming year is to try and answer questions related to this 30 by 30 initiative. Such that we can anticipate and help plan for what are the best lands to conserve in terms of bang for the buck. While also, reducing negative impacts.

HULT:

There's a lot of work left to do, right? Jordan, what would you like to leave folks with?

DORNBIERER:

For me, a big part of what makes this work really interesting is the opportunity to get the model to do these things. To solve those problems. If you know anything about data science, artificial intelligence, machine learning, they're pretty hot field right now. And it just so happens that we get to use some of these new computational tools and environments. It's an area that's changing a lot. And so it's fun to address these problems using some pretty shiny new tools. Getting the models to talk to each other. Getting FORE-SCE to grapple with the sorts of problems that we've been talking about, it's nice to have a well-stocked tool bag. And so that's a pretty fun aspect of working on the project.

HULT:

Great. So, it's a technologically challenging task in a new and exciting field. Or a rapidly expanding and exciting field.

DORNBIERER:

Yeah. Absolutely. There is so much support for that kind of work across projects and science at EROS . It's a good place to be when there are challenges in front of you and support behind you.

HULT:

We've been talking to Terry Sohl and Jordan Dornbierer about the past and future of the Delaware River Basin. Thanks for joining us guys.

SOHL:

You bet. Thanks John. It's been great.

DORNBIERER:

Thanks for having us, John.

HULT:

And thank you listeners for joining us as well. Be sure to drop in again for the next episode of Eyes on Earth. You can find us on our website at usgs.gov/eros. Or you can find us on Apple podcasts or Google podcasts. This podcast is a product of the U.S. Geological Survey/Department of the Interior.