Using Global Climate Model Projections to Improve Stormwater Resiliency

When planning or designing infra­struc­ture, Global Climate Models (GCM) are a powerful tool to help us understand future climate conditions such as sea level rise, temperature, and rainfall. Precip­i­ta­tion output, however, presents a particular challenge because these models typically focus on precip­i­ta­tion changes to large regional areas over months, years, or even decades. Although informative for many projects, GCM output is poorly suited for stormwater appli­ca­tions that require hourly data during storm events. Breaking down these global models to more granular data often requires highly specialized and complex processes that are not accessible to most munic­i­pal­i­ties or utilities and the results are still often poorly repre­sen­ta­tions of local conditions. 

Recently, a team of climate resiliency experts from CDM Smith and the Philadel­phia Water Department discovered a practical approach to downscaling GCM output for local stormwater systems. As part of the Philadel­phia Water Department’s Climate Change Adaptation Program, the team needed future precip­i­ta­tion projections to assess the impacts of changing storm size and intensity on stormwater and wastewater collection systems and infra­struc­ture. By itself, the downscaled GCM data could only be broken down into daily outputs and couldn’t give accurate insight into individual storm events. Also, when compared to current local rain gauge numbers, the GCM outputs over­es­ti­mated the number of consecutive wet days and under­es­ti­mated the intensity of the rainfall.

To address the need for more realistic data, the team combined existing rain gauge information and the daily GCM outputs to create a future precip­i­ta­tion hourly time series. This time series was used as a basis for a stochastic (randomly determined) weather generator to explore potential variability in projected precip­i­ta­tion.

“We were able to align the GCM ensemble data to data from Philadel­phia rain gauges from 1995-2015,” says CDM Smith senior vice president Mark Maimone. “This matching allowed us to take the 2080-2100 GCM predictions for the Philadel­phia area and shape them into a more accurate estimate of hour by hour rainfall as it might occur at the end of the 21^st century. ”

This method is compatible with commonly used tools such as hydrologic and hydraulic (H&H) modeling and Intensity-Duration-Frequency (IDF) curves commonly used in the engineering, planning, and design of urban storm sewer systems.

Step 1: Define Problem and Require­ments Figure out what specific challenges does your stormwater system need to address—is it combined sewer overflows, surface water quality issues, street flooding? You also need to assess your current stormwater management practices to determine what results are needed, such as IDF curves for design, sub-hourly precip­i­ta­tion time series for continuous modeling, hyetograph generation, and so forth.

Step 2: Collect and Pre-Process Data Gather observed rainfall records near the area of interest, ideally hourly or sub-hourly data points, to evaluate and aggregate extreme rainfall conditions. It is important to have at least 20 to 30 years of observed records to capture natural variability in rainfall. Then use a resource such as the U.S. Bureau of Reclamation’s Downscaled CMIP5 Climate Projections to obtain downscaled GCM output on a daily time scale. 

Step 3: Determine Global Climate Model Ensemble There are a few different approaches to selecting the ensemble of GCMs for your needs. It is recommended to use a minimum of 10 GCMs to capture a diverse set of models. One of the most common approaches is the “envelope approach” which capture a range of plausible projected precip­i­ta­tion changes. The method used in Philadel­phia combined the envelope approach with an approach based on past performance.  

Step 4: Choose Planning Horizons Decide a baseline period of at least 20 years to represent a comparison of “current conditions”. From there, choose one or more future time periods where projections are needed based on your system’s assets or relevant regulatory require­ments, such as MS4 compliance.   

Step 5: Develop Future Projections Develop future projections using the delta change factor method, using existing rain gauge precip­i­ta­tion data and downscaling the GCM ensemble to sub-daily time steps. This will allow you to create a future hourly precip­i­ta­tion time series.

Step 6: Apply Projections to Planning and Design Using the time series from step 5, develop or update the tools needed for urban drainage planning and design application (e.g. IDF curves, hyetographs, "Typical Years", etc.). 

From sewer overflows to basement back-ups and street flooding, this practical method can be used to simulate an array of future conditions. In turn, this can greatly inform stormwater planning and modeling. This framework can also be applied to other disciplines and appli­ca­tions where high-resolution precip­i­ta­tion time series are required, such as riverine flood modeling, drought planning, and even water quality modeling – which can then inform climate resilient design guidelines for infra­struc­ture projects.