Sources/Usage: Public Domain. View Media Details

In virtual presentations late this summer to the USGS Rocky Mountain Region 2020 Science Exchange Workshop, and to the Environment for Visualizing Images (ENVI) Analytics Symposium, Doucette focused on:

·       How the Land Change Monitoring, Assessment, and Projection (LCMAP) initiative developed at EROS closely parallels the objectives of EarthMAP, and

·       The capacity building needed to enable the EarthMAP vision, which includes data management infrastructure and adapted compute architectures; optimizing the utility of on-premise and cloud resources; determining the utility of existing datasets and identifying deficits; and rigorously investigating the utility of Artificial Intelligence and Machine Learning (AI/ML), particularly Deep Learning, to support decision making.

Doucette takes a complementary approach to the conversations that EROS Research Physical Scientist Terry Sohl has as a member of EarthMAP’s Program Management Team (PMT). The PMT and its representatives from each USGS mission area have focused on identifying integrated use cases to demonstrate EarthMAP concepts, with preliminary emphasis on integrating land change and water flow modeling in the Delaware River and Colorado River basins.

“From my perspective, the PMT’s main function is to focus on putting together the integrated domain science strategy, whereas the capacity-building effort emphasizes aspects of data science and engineering,” Doucette said. “In a sense, Terry and I approach our EROS contributions to EarthMAP with overlapping and complementary viewpoints. Given just how multi-faceted EarthMAP is, it makes sense to come at it from different angles.”

A long-term vision

EarthMAP is the USGS’ long-term vision for delivering actionable science information through integrating observations and projections of the current and future state of Earth systems—all done at scales and timeframes needed to support decision making. It fits into USGS Director James Reilly’s vision for the USGS, which involves the wide-scale integration of disparate scientific datasets and models into user-friendly tools for predicting:

·       Water availability

·       Land use and ecosystem change

·       Mineral and energy stores

·       And the probability of hazardous events like wildfires, flooding, hurricane impacts, or invasive species infestations.

At the Rocky Mountain conference, Doucette talked about the history behind LCMAP and how it closely parallels the objectives of EarthMAP. The obvious resemblance is the “MAP” in their names —Monitoring, Analyses, and Prediction for EarthMAP, and Monitoring, Assessment, and Projection for LCMAP. Though the origin of LCMAP predates EarthMAP by a couple of years, Doucette said that if the naming of one didn’t influence the other, then it’s quite an inspired coincidence.

He likes to think of LCMAP as a “special case” of EarthMAP, in that LCMAP’s focus is about understanding land surface change, whereas EarthMAP seeks to integrate a broader range of Earth observation inputs. Given LCMAP’s “head start” in development, the “Monitoring” component of LCMAP has already matured to an operational dataset and capability on a national scale.

The release of LCMAP’s operational land change monitoring product in June of this year can provide a very tangible foundation toward use case development for several EarthMAP scenarios, Doucette said. “We’re working proactively with Core Science Systems (CSS) leadership toward getting that word out, which means doing more interfacing with the key EarthMAP players. We need to better understand and demonstrate the tactical utility of current LCMAP data products across mission area needs—Water, Ecosystems, Hazards, and Energy and Minerals—and adapt our ongoing development activity accordingly.”

Migrating to the cloud

Doucette also discusses how the next phase of LCMAP will explore how to migrate development and operational activities into cloud environments to further facilitate integrated capacity building and science use case development.

“As we scale out the use of computational resources for LCMAP, we’ll be able to facilitate incorporating additional remote sensing observation data into our analyses, such as Sentinel-2, and quite likely several other datasets,” he said. “Furthermore, in the cloud, we believe we’ll have more effective access to the latest AI/ML services, which is one of the fastest growth areas in the cloud in recent years.”

That same capacity building need, of course, applies to EarthMAP as well, especially with the data and compute power needed to support national-scale applications. It would likewise provide us with capabilities to take on global monitoring applications for some of our partners, Doucette added.

As he speaks at various venues, Doucette said he hears people talking about their particular projects, their specific roles in those projects, and how they see their work potentially being connected to the EarthMAP vision.

Those conversations reveal a central challenge of EarthMAP—establishing a top-down orchestration of activity across an organization whose culture has evolved into a broad collection of distributed science centers over many years, he said.

“It makes it a bit more challenging for individuals or smaller groups to propose how their work could fit into a unified vision like EarthMAP,” Doucette said. “And then navigating the programmatic machinery through which to actually make a meaningful contribution, of course, presents the biggest challenge.”

The challenge of integrated science

Establishing a multi-disciplinary science environment and culture sounds nice in theory, but how to effectively pull that off in practice is hardly clear-cut, he added.

Doucette said they’re faced with the same kind of challenge at EROS. In conversations last March about becoming more integrated as a branch at the Center, he said some staff were not convinced that “integration” per se was warranted, let alone finding consensus on how it should be defined in practical terms.

“If we find ourselves promoting change for change’s sake, then we’ve missed the point. However, technology is always changing, and because the pursuit of science relies on it, the business of science must continually adapt to it,” he said. “The trick is to adapt more efficiently across shared science and user needs.”

As an analogy, he described how in the course of his lifetime, he’s witnessed data media forms evolve from magnetic tape, floppies, HDDs, CDs, DVDs, flash memory, to streaming from a cloud. And lamented, he added with a smile, that at some point he had to cut loose from his cherished cassette tape deck and vinyl records.

Doucette believes cloud computing is the new medium through which we can completely rethink how we do “integrated” science, facilitating the investigation of ideas in ways that could simply not be done before. “But even cloud computing will eventually evolve into something else that we can’t predict, because that’s what technology does,” he said.

He thinks the evolution of LCMAP serves as an ideal entry point into the broader EarthMAP discussion for the entire branch and Center, and that focus areas like NLCD, Fire Science, Vegetation, Water, Climate Dynamics, Coastal Dynamics, Early Warning, LSIS, S2, ECCOE, and partnerships, have already started those conversations.

“Part of what we’re talking about involves adapting our cultural mindset,” Doucette said. “While that doesn’t happen overnight, it can be encouraged through shared goals and smartly designed incentives.”

The conversations, and the actions that result from outreach venues, will hopefully move the ball forward, Doucette said. He believes strongly in an aggressive outreach strategy and looks for opportunities to share ideas like he did at the Rocky Mountain Region workshop, and the ENVI Analytics Symposium. As he does, he will talk about the lessons learned and potential capabilities from LCMAP, about capacity building, and about how EarthMAP can leverage these ideas.

“The EarthMAP concept has tremendous scientific merit,” Doucette said. “We have a real opportunity to influence its trajectory with what we’ve accomplished, and where we’re going, with LCMAP.”