SWCA and Alteia collaborate on vision AI solutions for environmental applications

Environmental consulting company SWCA is collaborating with Alteia on vision AI solutions for its clients in the energy and industrial sectors

Alteia provides a flexible artificial intelligence (AI) application development platform that complements SWCA technologies, and extensive experience developing and deploying environmental applications at scale for a wide range of industries.

The collaboration focuses on enabling and deploying AI-based applications on the version of the Alteia software platform deployed for SWCA called SWC.ai.

These applications use digital technologies and computer vision to help SWCA clients overcome environmental challenges and move their projects forward. In the face of rapid environmental, economic, and societal changes, SWCA’s purpose is simple: to preserve natural and cultural resources for tomorrow while enabling projects that benefit people today. 

This software platform and its related applications are quite complementary to traditional GIS solutions. Of course, there are a lot of parallel features but the high-level target is altogether different: SWC.ai is very strong when parsing and analyzing massive amounts of asset-based visual data. The objective is to operationalize AI/Machine Learning (ML) on visual data at a massive scale and contextualize information easily between a wide range of data sources.

Additionally, Alteia and SWCA are exploring opportunities for collaboration on solutions that help achieve net-zero carbon emissions and decarbonize energy, construction, and industrial sectors, including emissions management.

Case study: Major Energy Operator and SWCA demonstrate high value by leveraging SWC.ai

Companies that operate natural gas or petroleum pipelines in the United States must restore the land on which they’ve installed new lines. They must verify they’ve complied with the terms of their land lease and any national or local environmental regulations.

The monitoring of these restoration projects is traditionally a manual process conducted by skilled survey crews. In an effort to be a more digitally enabled company, a major energy operator partnered with SWCA data and environmental experts on a new approach to data collection and analysis.      

SWCA data specialists acquired data from a combination of sources including satellites and drones and processed it through SWC.ai to deliver on this pipeline lifecycle management project. 

Lease agreements and local regulations often require environmental monitoring.

Companies that operate oil or gas pipelines work under regulatory or contractual requirements to restore land in the right-of-way where they’ve laid new lines. 

When operators sign property lease agreements, the contracts often require operators to restore environmental disturbances that occur when they install pipelines. Federal and state-level authorities often impose added requirements. 

Similar mandates apply when oil and gas companies shut down wells or pipelines, but the process is called reclamation rather than restoration.  

For both restoration and reclamation projects, operators must restore a site’s stability and ecosystem to comply with contracts and local regulations. 

Operators may be obliged to establish a native plant community that meets standards for density and foliage production. Requirements may include specifications for vegetation cover, density, vigor, resiliency, diversity, control of non-native species, and free of noxious weeds.

Operators may also be required to: 

• Preserve and salvage topsoil.  
• Establish controls to prevent loss of topsoil to wind and water erosion.

To meet these obligations, pipeline operators fund land-restoration projects and monitor the results. 

They verify their regulatory and legal compliance by collecting survey data about the environmental characteristics of the land before installation and after restoration or reclamation.

It’s a big task. The United States operates nearly 3 million miles of natural gas pipelines that connect sources to consumers. Almost 70,000 miles of those pipelines were for bulk distribution in 2017. This is the latest year for which numbers are readily available. 

The U.S. mileage of bulk gas distribution pipelines increased by an average of about 1,500 miles a year between 2013 and 2017. Most of those new miles required monitoring of land-restoration projects.  

Challenge: Safely monitoring every inch of nearly eight miles of pipeline

Until recently, surveying crews collected such environmental data manually. Pipeline operators typically coordinate their own surveying crews or hire third-party contractors. 

But the traditional surveying process is not exempt from the questions that arise with digital transformation underway across all market sectors.

• Is there a better way to collect more comprehensive data?
• Is there a better way to collect more accurate data?
• Is there a more cost-effective and efficient way to collect and analyze data?

In remote or mountainous areas, it may be impractical or unsafe for surveying crews to perform detailed site surveys. In such cases, crews may sample data from accessible locations within a site, capturing readings at intervals of a hundred feet or more. Manual data capture is also subject to human error.

Moreover, the COVID-19 pandemic elevated two more concerns: workplace safety and labor shortages.

During surges in the rate of infection, the risk of contagion made it unsafe to deploy surveying crews. Lockdowns prevented teams from going into the field, and subsequent labor shortages made matters worse. 

Pipeline operators had to find new ways to monitor their sites.

Solution: Collect data remotely and process it using vision AI 

During the pandemic, SWCA proposed a pilot project for a pipeline operator to gather local environmental data through remote sensors. 

The plan was to capture the data through unmanned aircraft and satellites. The use of aerial imagery to collect site data is not new in the pipeline industry. Operators have often used aerial photogrammetry and geospatial mapping to supplement the data gathered by survey crews. 

Here’s the new part. SWCA would process the data through SWC.ai in an automated way, leveraging its data contextualization capabilities and vision AI algorithms.  In fact, one challenge with using vision AI at scale is that images are unstructured data. Still or video images are in diverse formats and may come from a wide variety of sources, including smartphones and other still or video cameras on the ground. 

Alteia has spent years developing and refining an artificial intelligence platform that quickly gathers, processes, analyzes, and contextualizes massive amounts of visual data. 

Results: Better data and increased efficiency at a lower cost 

Compared to other methods the energy operator used to survey and collect data, the proof of concept (PoC) delivered a richer data set, more objective data, increased efficiency and lower costs, and influenced a faster response to address project problems.

It also showed that the artificial intelligence system, powered by Alteia, effectively integrates and analyzes survey data for environmental restoration projects.

How the Proof of Concept worked

The two companies located their vision AI pilot at a restoration site in the US. The pipeline operator had recently installed 7.7 miles of new 42-inch diameter pipeline. 

SWCA collected visual data from two sources: 

• Satellite imagery bought from commercial sources
• 2D photos taken by drones that the firm has deployed 

Both sources provided images of each area in two formats:

• Conventional 2D still photography with geolocation data
• Near-infrared (NIR) spectroscopic images 

The data was processed through SWC.ai. In fact, the software combines data from multiple sources, then fuses it to create an accurate 2D digital map of the site. 

Next, the platform analyzes the data to calculate vegetation coverage. It does so by using an index called F-Cover, which measures the density of vegetation within a square meter of ground. The index shows the percentage of an area covered by vegetation. 

The software then plots F-Cover values on a geo map. The map shows average vegetation density for a designated area that can be as small as a pixel.  

By comparing before and after images of the same areas, we can see the extent to which the pipeline operator has met their obligations to restore vegetation.

What is F-Cover?

F-Cover is the percentage of plant material that covers the soil surface when observed from straight above. As an example, monitoring F-Cover on agricultural fields from the early season onwards presents an indication of the rate of crop development and vigor. Increasing F-Cover signifies development of leaf area or above-ground biomass. 

F-Cover also provides a measure of the susceptibility of the soils in a field to erosion. Crops with high F-Cover at early developmental stages better intercept incident radiation and rainfall, thereby increasing soil shading and decreasing soil evaporation. This can also be used to estimate irrigation requirements.

F-Cover can be estimated from digital images taken of a field from a vertical position looking downwards. The device used can be a digital camera or smartphone. The picture is interpreted into “crop/plant area” and “soil area” from which F-Cover is estimated. 

Many industries use F-Cover, including Power & Utilities, Oil & Gas, Agriculture & Forestry, Mining, and more.

Challenges overcome

This PoC demonstrated that there is a better way to collect more comprehensive and accurate data in a more cost-effective and efficient manner. 

Alteia’s vision AI software applied the same algorithm to every single pixel of the satellite image, calculating the indices in the exact same way every time and providing a truly consistent analysis of vegetation cover. 

Lessons learned

The PoC confirmed two hypotheses:

• Remote sensors, including cameras mounted on satellites and drones, can provide the data needed to help fulfill the obligations of its leasing agreements. Data collection and analysis were more accurate than for human crews monitoring on the ground.
• SWC.AI, built on Alteia’s vision intelligence technology, can process that visual data and turn it into actionable insights. A single user of the platform can complete all the tasks in minutes. 

The PoC also showed that the use of vision AI can identify problems fast, without the need for crews to patrol rights of way. 

At one site, vision AI showed a loss of vegetation in an area the team had restored. The system identified the problem because it spotted declining chlorophyll levels. 

SWCA sent in a crew.  They discovered that a nearby landowner had allowed their livestock to graze on the grass in the right of way. The livestock ate the new grass down to the dirt. 

The process of identifying the problem was much faster and more efficient than having teams patrol the right of way. 

Next steps for SWCA

The SWCA Data Acquisition team is exploring additional data types for enhanced contextualization and decision making such as satellite imagery, terrestrial LIDAR scanning devices, 360-degree cameras, and mobile data capture applications. In six months, the SWCA team has over 70 projects on the platform and is servicing over 30 enterprise clients across the energy, mining, and land development sectors.