Land Cover

Building More Accurate Flood Models with Ecopia Land Cover Data

October 11, 2022
Post by
Ecopia AI

Enhancing Stormwater Resiliency in a Rapidly Changing Climate

The City of Peterborough, Ontario is not unfamiliar with floods. On July 15, 2004, 220 mm of rain fell on the city over the course of nine hours, causing extensive flooding and an estimated $50-100 million in damages to infrastructure and property1,2. While this flood occurred almost two decades ago, the impact it had on the community is still felt today as city planning officials continue to develop flood resilience and mitigation strategies in the midst of a changing climate.

Peterborough implemented various stormwater management plans in the years following the 2004 flood, but found they lacked a true digital representation of their community that they could use to effectively model potential flooding scenarios. There were three main issues with the data Peterborough was using to understand flooding: 

  • Detail: The data Peterborough was able to source was made up of basic line features representing rivers, creeks, and other watercourses. These layers were helpful indicating the general flow direction and depth, but did not provide the information about surrounding surfaces that is necessary to develop accurate flood models. The City relied on zoning data to infer where impervious surfaces were located in relation to watercourses rather than deriving exact surface information from imagery data, resulting in low quality estimates being fed into flood models.

  • Consistency: The data Peterborough was using was collected by many different sources. As a result, the features the team had to work with lacked consistency and were of mixed quality, having been collected using different methodologies and at different times. This made it extremely difficult to model flood conditions with confidence.

  • Freshness: Inconsistencies in data collection also resulted in data that quickly became stale. Relying on piecemeal data collection from various sources, Peterborough never had a true, up-to-date representation of watercourses in their city. Instead, flood data contained a mix of vintages, with some features having been collected and verified recently enough to be accurate, and others collected years ago and not updated. 

Faced with a changing climate, increasing flood incidents, and the memory of 2004’s catastrophic damage, the City of Peterborough set out to enhance their climate resiliency and flood mitigation strategies by building an integrated flood model (IFM) with a level of detail, consistency, and freshness they could be confident in. 

Leveraging Ecopia AI’s Land Cover Data for Advanced Flood Modeling

The City of Peterborough partnered with Ecopia AI (Ecopia) to digitize the features of their community needed to build an accurate IFM and continuously update it to reflect the constantly changing landscape. The resulting IFM is a three-way coupled system incorporating all storm sewers, watercourses, and surface drainage across the entire city. Storm sewers and watercourses are modeled using traditional 1D methods and then connected to 2D surfaces to accurately define overland flow, depths, and velocities for a range of stormwater events in ways that 1D features alone could not. Ecopia’s complete coverage of features digitized from high-resolution aerial imagery has not only enabled Peterborough to accurately model watercourses, but also further understand how the surrounding land and infrastructure factors into flood events.

Ecopia AI ingested high-resolution imagery data (left) of Peterborough and digitized it into vector features (right) so hydrologists can develop accurate flood models.

To effectively model how a flood will impact a specific area, hydrologists need more than estimated land cover derived from zoning laws. Using Ecopia’s detailed land cover dataset, which digitizes and classifies features from imagery using artificial intelligence (AI) and machine learning (ML), Peterborough is able to understand exactly where certain types of land cover exist in relation to water features. This level of detail enables Peterborough’s hydrologists to optimize the IFM based on how water can be expected to interact with different pervious and impervious features.

For example, Ecopia’s detailed land classification empowers Peterborough to calculate surface roughness and flow resistance measurements that correspond to actual land cover captured in imagery data. Essentially, the Peterborough team has developed a 2D surface mesh representing land cover, with each grid of the mesh containing critical information related to water depth and velocity in the event of a flood. Building this 2D mesh requires accurate knowledge of surface roughness (also known as Manning’s n-values) which Peterborough was unable to derive from land zoning documentation alone. With Ecopia data, Peterborough can assign distinct values based on actual flow resistance for the different types of land cover throughout the City.

From left to right: Peterborough’s flexible 2D mesh to model surface flow; the IFM’s surface roughness layer derived from Ecopia’s land cover data; the output layer of flood extents based on the other two layers and including the area’s pipe network.

Peterborough used to rely on estimates of where impervious and pervious surfaces were located; with Ecopia they can not only know for sure where these vastly different areas are, but also further segment impervious surfaces to enhance their models with more detail. The team now classifies impervious surfaces as either directly or indirectly connected to the city’s sewer systems in order to develop more accurate runoff coefficients. With this segmentation, directly connected impervious surfaces like roads, driveways, and parking lots have a higher coefficient than indirectly connected impervious surfaces such as roofs and sidewalks. This separation better simulates runoff response and improves the accuracy of the IFM in stormwater management and planning, a huge improvement from previous methods. 

“With the IFM, we have a far greater understanding of flood risks in our community. Our flood reduction capital program uses this intelligence to identify future projects, test a range of scenarios, and prioritize work, all with the goal of achieving the highest level of flood reduction with limited capital funding. Land-use planning is also better informed with the IFM, resulting in future development that is protected from flood risk and limits or eliminates exacerbating flood risks for other areas of the City.” - Ian Boland, Senior Watershed Project Manager for the City of Peterborough

Accurately modeling the complexity of land cover in urban areas like downtown Peterborough was top of mind for city hydrologists as they implemented Ecopia’s data. Much of the damage during 2004’s flood, as well as subsequent stormwater events, was concentrated in urban areas with varying types of impervious surfaces. To plan for future stormwater events and develop a higher resiliency for the community, Peterborough leverages Ecopia’s building footprint dataset in their 2D mesh representation of the city. The digital terrain model of cells containing buildings is artificially raised to signify how water will move around those structures in a flood event and into other cells. 

The team then uses a kinematic wave method to calculate rainfall runoff catchment based on the previously discussed different segments of impervious surfaces (directly and indirectly connected). In Peterborough’s more rural areas with less impervious surfaces, a Soil Conservation Service (SCS) Curve Number (CN) is used to determine flood risk and stormwater resiliency. The CN indicates the runoff and infiltration potential of soil derived from Ecopia’s land cover data, allowing hydrologists to more specifically model the effects of stormwater events on pervious surfaces.

Why Ecopia?

As the Peterborough team evaluated different sources of data for their flood modeling and stormwater management efforts, they sought a dataset that was detailed, consistent, and fresh enough to develop climate resilience strategies with. Because of the guaranteed >95% precision of features and the frequency of updates, Ecopia stood out from the beginning as a source of truth Peterborough could rely on to accurately improve and scale their planning.

“Ecopia’s ability to efficiently extract all land cover features, whether manmade or natural, enables us to develop flood models that represent reality. The planimetric level detail map was critical for our stormwater engineering consultants, Jacobs, to help support the development of our IFM.” - Ian Boland, Senior Watershed Project Manager for the City of Peterborough

Ricardo Santaella of Jacobs added:

“The outputs from Ecopia will be very helpful in the future as it can be used to track for development and land cover changes. These modifications can then be used to update the model and evaluate the implications in terms of flood risk exposure at a local or global level. One of the main benefits from the Ecopia product is that manual delineation of polygons is not required, thus saving a significant amount to update a previous dataset.”

The AI and ML used by Ecopia to extract land cover features from imagery means that data for Peterborough is updated on a more frequent cadence than before, and delivered in a consistent format that simplifies the flood modeling process. With Ecopia data, the team no longer needs to spend time combining data of varying quality from disparate sources, and can instead devote their energy to further enhancing their stormwater management plans. The detailed classification of land cover provided by Ecopia, as well as the human-like precision in which each feature is digitized, gives Peterborough the flexibility and confidence needed to build the most advanced flood models and better plan for future stormwater events.

With Ecopia’s data, the City of Peterborough has a comprehensive view of how water features relate to both pervious and impervious surfaces. Explore a sample of data from Peterborough below:

What’s Next?

Since implementing Ecopia into their stormwater management plans, other departments within the City of Peterborough have started to leverage the data. When departments are unsure of existing renderings of building footprints, they turn to Ecopia’s highly-accurate datasets. Other departments have used the land cover classifications to analyze driveway and parking availability throughout the city, and are looking into applying the data for solar energy installation plans. Peterborough is currently developing public-facing mapping applications that feature Ecopia’s data to educate constituents about flood risk, and are also relying on the data to develop detailed reports for city planning efforts. 

As communities continue to evolve city planning efforts in the face of climate change, the importance of having an accurate and up-to-date representation of the world will only grow. To learn more about Ecopia’s stormwater management solutions, or how we are supporting geospatial data creation in other municipalities, please contact Brandon Palin, Senior Director, Public Sector & International Development.

Request a Demo

Related Blog Posts

Subscribe to get our latest posts straight to your inbox!

If you're ready to leverage groundbreaking advancements in artificial intelligence, let's chat. For more examples of Ecopia's extraction capabilities, please view our samples.

Ready to get started?

If you're ready to leverage groundbreaking advancements in artificial intelligence, let's chat. For more examples of Ecopia's extraction capabilities, please view our samples.