Upstream - Enabling coastal adaptation and resilience

Using remotely-sensed data and machine learning to optimize adaptive measures that protect coastal ecosystems, communities and related infrastructure endowments

Coastal ecosystems are no longer merely “at risk.” The severe and accelerating impacts of an increasingly unstable climate and sea-level rise are being actively felt across the entire globe. The looming, long-term consequences of climate disruption can no longer be avoided, and we must prepare accordingly. Fortunately, we have at our disposal a growing and diverse array of resilience-enhancing green, grey and hybrid adaptive measures to help reduce climate disruption impacts on coastal ecosystems, communities and related infrastructure endowments. The effective deployment and scaling of these measures faces challenges, however, due to the human resource and capital intensive nature of manual design, monitoring and optimization processes.

Conventional means of understanding the climate crisis, for example, require hundreds of human hours to: 1) travel to impacted coastal regions to collect ground data that quickly becomes stale; and 2) the subsequent modeling of projected impacts using labor-intensive and thus costly geospatial tools. Although the advent of drones has helped, their deployment at scale to collect data also remains cost prohibitive. These costly, front-end processes ultimately reduce the availability of funds for monitoring and evaluation to assess and optimize the effectiveness of adaptation strategies. Collectively, these dynamics create barriers to scaling and make it all but impossible to adapt proactively to ever-changing ground conditions.

To ramp-up and optimize the impact of both public and private sector investments in adaptive measures designed to protect and prepare both populations and natural habitats to future climate regimes, we need a more cost-effective, scalable way to use geospatial data and related GIS technologies and tools. The ever expanding presence of global satellites represents an important part of the solution by enabling the collection of geospatial data, most of which is publicly available, across vast temporal and geographical areas in a more cost effective manner. The use of satellite data to analyze ecosystems’ response to adaptive measures presents two challenges, however. First, raw satellite data must be transformed into uniform and meaningful data sets using a model or algorithm. Second, individual satellites collect data frequently or with high spatial resolution, but not both.

Upstream’s platform overcomes these challenges by leveraging ongoing advances in machine learning to improve both the resolution and model accuracy to the levels required for more widespread use and adoption. Our large-scale data processing pipelines automatically input satellite imagery into machine learning models, bringing iterative and rapid software development processes to a historically slow-moving market. These client-specific models, and associated design and operational advancements, make the platform highly versatile, cost-effective and user friendly. The Upstream platform is therefore ideally positioned and uniquely suited to play a key role in enabling coastal stakeholders to make more informed choices regarding the adaptation strategies available to them, and actively monitor these strategies once implemented to optimize their long-term effectiveness.

Upstream has successfully harnessed technological advancements in, and synergies between, remote sensing, computer science and machine learning to create customizable platforms that addresses water-related conservation and development challenges. The power of Upstream’s technology lies in its ability to cover large geographies with a high temporal cadence in a way that is cost effective and non-intrusive. We do this by combining numerous satellite sources into one “Satellite Ensemble,” supplementing one satellite’s weakness with another’s strengths. This allows us to penetrate clouds and employ cloud-masking models, providing more complete coverage and higher levels of analytical accuracy than our competitors.

Upstream leverages three technological innovations to process satellite data in a more cost-effective and efficient manner to overcome the constraints of conventional geospatial analysis: machine learning, cloud infrastructure and satellite imagery. Collectively, these innovations transform satellite imagery from multiple sources into uniform, customizable data layers that are updated, processed and analyzed in near real-time autonomously whenever new satellite images are captured. The autonomous, highly customizable and thus cost effective nature of Upstream’s platform lends itself to scaling up geographically and out across industries and applications.

Over the past two years, Upstream has established a firm organizational, operational and commercial foundation to enable our platform’s rapid growth, and thus impact. With a stable, back-end software platform and capable team now firmly in place, our efforts are focused on expanding our partnership  base through the development and optimization of partner-specific dashboards focused on the conservation and monitoring of water resources and related watershed, wetland and coastal ecosystems. Over the next 12 months, we aim to secure 20 new partnerships with a cross-sectoral mix of conservation organizations, private developers and government entities in both developed and emerging markets.

Our platform’s scaling and impact maximization over the next 3-5 years will revolve around strategic, overlapping partnerships with leading global conservation organizations, corporations and host governments in the geographies most at risk from climate change. In the global south, for example, we developed tools to enhance the water-use efficiency of key agricultural commodities while simultaneously monitoring the integrity of linked ecosystems to enhance the resilience of millions of smallholders. As additional solutions are developed, they will be scaled and expanded to include other water-focused applications in vulnerable regions with an initial geographic focus on developing Asia and Africa.

The Upstream Dashboard is a web-based platform readily accessible to diverse audiences through a customized and easy-to-use online and mobile applications. From smallholder farmers and their NGO enablers to water resource and forestry managers and infrastructure planners, the platform is designed to make data exploration, analysis, stakeholder collaboration and decision making easier, cheaper and more transparent. Within the context of the Solve mission to support coastal communities, we envision using the platform to: 1) raise awareness and build support for coastal restoration and conservation activities; and 2) optimize the design, implementation and rolling monitoring of coastal restoration and adaptation programs.

Upstream serves three complementary audiences: 1) national and international decision-making organizations - i.e. national and state-level governments, NGOs, and multilateral development institutions - that need rapid baseline and constantly updating geospatial data and corresponding analysis over vast regions historically and in near real-time; 2) corporate stakeholders that manage natural resource and commodity-anchored supply chains; and 3) public, private and non-profit water managers, utility companies and infrastructure developers that need planning and design support to incorporate future climate regimes. Upstream has active and evolving partnerships with organizations that directly touch the lives of millions of direct beneficiaries.

Upstream’s coastal-focused dashboard will initially be deployed through partnerships with at least ten Gulf Coast communities to support coastal restoration and conservation programs. Our custom machine-learning models will enhance the design, financeability, implementation, monitoring and rolling optimization of restoration projects in these communities to reduce the risk of coastal flooding, protect infrastructure endowments and improve systemic resilience. These benefits will be realized rapidly as the platform is deployed, and be scaled to a target at least 100 communities within 3 years. Scaling will be driven by the cost-savings our platform delivers in comparison to conventional geospatial analysis and monitoring techniques.

Upstream’s strengths include our demonstrated capacity to innovate, our lean and close-knit team of professionals with complementary backgrounds, and our commitment as a public benefit corporation to pursue commercial opportunities that benefit our planet. Our cross-disciplinary expertise and decades of combined field experience in environmental management, machine learning, big data, international development and public policy ensures that our customized decision-making tools are grounded in reality, user friendly and impactful. Further, our distributed feature teams enables faster iteration, lower overhead and gives us access to wider talent pools than our competitors, positioning us for rapid scaling over the next three years.

Upstream is a Software as a Service (SaaS) company with a revenue model that emphasizes long-term service contracts with NGOs, governments and resource managers, supplemented by shorter-term consulting contracts. We anticipate that consulting revenue will gradually increase as the number of our commercially available dashboards expands, lending itself naturally to scaling and improved financial performance. Once dashboards are developed, for example, we can deploy at little or cost while simultaneously freeing up financial and human resources to develop new, revenue-generating dashboards. We balance our for-profit structure with our incorporation as a public benefit corporation, holding impact above profit and often helping underfunded organizations in cases of high priority impact. To capitalize partnerships with traditionally underserved or under-resourced communities and grassroots organizations, and in so doing fulfill our public benefit vision, we are also developing an impact fund with several large, US-based family foundations. Because of our low operating costs and small team, we project profitability towards EOY 2019.

Solve’s unique emphasis on community outreach and engagement made an impression on Upstream, as it aligns with our dual desires for greater grassroots engagement and expanded coastal-focused impact. Becoming a part of the Solve community will: 1) help improve the accuracy and impactfulness of our existing, coastal-focused tools through feedback from a community of active and informed practitioners; and 2) expand our impact by introducing us to a network of communities and on-the-ground doers - including infrastructure developers and policymakers - that can benefit directly and materially from our platform’s more cost-effective quantified decision-making and geospatial analysis capabilities.

Machine learning and advanced remote sensing techniques are still relatively novel among geospatial practitioners and the decision-makers they empower. Trust-building and targeted, user-oriented education campaigns will lead to their more widespread acceptance as integral and versatile components of the geospatial toolbox that help generate more informed decisions and greater impact at lower costs. Without regulatory support from the Army Corp of Engineers, EPA, and other key stakeholders involved in coastal restoration work, the benefits of Upstream’s platform to aid in project planning and monitoring will not be realized. Solve can assist in facilitating these relationships and in related advocacy.