Solar Sanctuaries

The capstone workshop constitutes the culmination of Columbia University's Sustainable Development undergraduate program as it affords students the opportunity to draw on the knowledge we have gained in the classroom to collaboratively craft practical solutions to critical problems for real-world clients.

For our capstone project, we are working for the NYS Council of Churches and Crauderueff & Associates. Our clients have received a pre-development grant from the New York State Energy Research and Development Authority (NYSERDA) to support the use of solar development within the framework of compliance with Local Law 97 (LL97), a piece of ambitious climate mitigation legislation that places carbon caps on most buildings larger than 25,000 square feet (nearly 50,000 properties in NYC). We have been tasked with using GIS technology to estimate the potential for Solar PV to offset LL97 fees for Houses of Worship across NYC.

To maximize the financial incentives available to our clients, we will focus on investment zones identified by the state and the federal government (Disadvantaged Communities (DAC) and Federal New Market Tax Credit Zones (FNMTCZ) respectively). Our scope is limited to rooftop solar due to time constraints (excluding parking lot covers and other solar development areas).

We began our project by conducting extensive research about LL97 itself. LL97 is a notoriously complicated law that establishes different carbon caps, and different “pathways” to reduce emissions, for classes of buildings across NYC. Even though the law has been written and passed, it will not enter into force until 2024–that means that most people, even experts, don’t know exactly how it will be enforced. A number of questions remain unanswered: will Houses of Worship be subject to fines? Which Houses of Worship will qualify, and which will not? What about mixed use buildings? These unknowns highlight the significant limitations of our analysis. Our team interviewed a number of legal experts on real estate law and climate legislation to gain the best understanding we could on the subject. 

Furthermore, we struggled to understand exactly what qualifies as a House of Worship. In New York, most buildings are multi-use, so a church might exist on the first floor of a building with offices on top. Or, maybe a religious school has a chapel in it–is the whole building a House of Worship, or just the chapel? What if the school and the chapel are located on the same tax lot? These real world issues forced us to make a set of assumptions based on our interviews with legal experts on LL97. In the end, we assumed that “covered” Houses of Worship were: 

• Buildings over 25,000 square feet with “worship facility” listed as their primary use

• Buildings over 25,000 square feet with “worship facility” listed as their secondary or tertiary use, IF this use was also allocated at least 25,000 square feet. 

We also completed extensive background reading on solar site suitability analysis and the types of criteria analysts use to determine which sites are good for solar. We learned from the literature that most site suitability uses some kind of weighted ranking system, but our client was only interested in a binary system (i.e, he was still interested in reaching out for solar development even if the site was mediocre due to the NYSERDA funding mechanisms). From the literature, we decided to focus our suitability analysis on rooftop pitch and solar irradiation. We omitted more complicated factors like aspect, seasonal radiation, etc. because we were not employing a weighted ranking system.

In the early stages of our project, we formulated the following research questions to guide our analysis:

1. Which buildings used for worship are covered under LL97, based on LL84 data?
2. How many buildings used as Houses of Worship are suitable for solar development? Which of these qualify for tax incentive programs based on their location within designated Disadvantaged Communities by NYSERDA and/or within Federal Tax Incentive Zones?
3. What is the solar potential for these buildings?
4. How can we mobilize web GIS to create educational tools for building owners around NYC  to come into compliance with LL97?

Our project used the following data sets: 

•  LL84 Energy and Water Data  (reported 2020 for calendar year 2019)

•  Federal Opportunity Zones  Shapefile (2018) 

•  NYC Disadvantaged Communities  (2023)

•  NYC LiDAR data  (2017)

•  NYS Building Footprints  (2016)

•  NYC Historic Landmarks  (2018)

Originally, we hoped that our client (the New York State Council of Churches) would be able to provide us with a list of properties for analysis. However, we soon realized that this would not be possible and used publicly available self reported data from LL84 to identify Houses of Worship. 

Our biggest challenge in working with this data was the inaccuracy of self-reporting. It was impossible to join the LL84 data sets to others using the Add Join tool because over 50 Building Block and Lots (BBLs) in the data set were inaccurately reported, along with a number of incorrect addresses and other line items. We worked around this by manually correcting these data points one by one and completing a spatial join. However, it is important to note that potential inaccuracies are another key limitation of our analysis.  

We also struggled to work with the data size, particularly with the LiDAR data. We did our best to work around this by clipping the data as much as possible (to NYC at the very least, but around the Houses of Worship when possible). In the end, the LiDAR data were too heavy for the virtual machines and we ended up using a computer at GSAPP. 

With these existing data sets, we also generated new data for our solar site suitability analysis. After extensive searching, we couldn’t find a publicly available DSM of NYC that we could use to analyze rooftops. Instead, we created one using the LAS to raster tool in ArcGIS. We used the resulting DSM to create a hillshade data set and solar radiation data set as well.

LL97 grandfathers in covered building from its predecessor, Local Law 84 (LL84). Because LL97 is so new, at the time that we started our analysis there was no publicly available data on which buildings were covered, so we used the covered building list for LL84. This data set also includes operational metrics for each covered building. After cleaning the spreadsheet we geocoded each building in ArcGIS Pro and visualized them alongside publicly available shapefiles of DAC and FNMTCZ in NYC. Then we extracted all buildings that used 25,000 square-feet or more for worship and mapped them.

Our exploratory data analysis highlighted the iterative nature of GIS analysis. We first visualized all covered buildings in NYC and clipped to a selection of use type, “worship facility.” We redid this analysis several times after realizing that: 

1. We had been using data collected in 2020, which would inaccurately underestimate energy use for commercial buildings. 
2. New expert interviews highlighted that  buildings listed as secondary or tertiary worship facilities were only “covered” when those secondary or tertiary uses met the 25,000 square foot requirement.

The  map  below is our answer to question 1. It displays Houses of Worship across NYC and includes pop-ups containing operational energy and emissions data, using pre-pandemic metrics, which can be accessed by clicking on a point.

Step 1: Identify scope of analysis

We knew we needed a Digital Surface Model (DSM) to calculate incoming solar radiation per roof, but there is not one publicly available for NYC...so we decided to make our own using LiDAR point clouds from the US Geological Survey (USGS).

Below is an interactive rendering of the raw point cloud, to get a sense of what we were working with. This single tile contained over 11 million points.

Using the ESRI workflow "  Estimate solar power potential ," we executed our analysis that revealed solar power potential per building for just one tile. The church of interest in this study area, depicted as a red point, was found not suitable for solar development.

To wrap up our analysis we found the area of suitable rooftops and selected rooftops with 30 meters squared or more of suitable area. This is what designated the house of worship within this study area as not suitable for solar development.

Then we used the field calculator to convert incoming solar radiation to electric power using standardized constants.

Our final product, below, represents a yearly average estimate for solar power potential per rooftop. Click on the yellow and orange shapes to view the outputs of our analysis.

We tried to repeat this process for all 608 tiles using Model Builder on ArcGIS Pro. Unfortunately, while we developed a functional model for a single LAS file, we were unable to successfully troubleshoot the implementation of an iterator which prevented us from scaling up our analysis.

We found that there are 229 Houses of Worship in New York City that will be subject to LL97 in 2024 or 2030, and need to begin retrofitting their buildings to reduce their emissions. Out of these 229 buildings, 87 Houses of Worship are located in tax incentive zones (Federal Opportunity Zones or NYC Disadvantaged Communities), where they can receive direct payments to finance building solar projects. 17 Houses of Worship are considered historical landmarks and are ineligible for rooftop solar, but may be candidates for community solar subscriptions or other types  of installations (parking canopies, garden arrays, etc.). 

Unfortunately, we do not know how many sites across New York City are suitable for solar installation nor how much power they could collectively generate. However, we did document a workflow that a professional GIS analyst could implement for the client.

• Self reported data is unreliable!
• We were challenged by limits to computing power, time, and our own computer science knowledge when it came to processing heavy data.
• Client communication is key. Explaining what GIS is and what we can do with it was necessary to ensure the clients could give clear direction on the desired product. Good science communication is key to crafting a mutually beneficial partnership.

If we had unlimited time to work on this, we would continue to troubleshoot the model builder to complete our site suitability analysis and determine how much power these churches could produce. 

We also recognize that a very similar project has already been accomplished with Google’s  Project Sunroof . We wonder if there is a more efficient way to accomplish a similar result by using python code to search and save the results for each result to an excel file. 

Another interesting direction to take this analysis, combined with other results from the workshop group, would be to map potential PV output and building owners’ awareness of LL97.  Can we quantify how much solar power is lost due to unawareness or undereducation around LL97? Are owners with larger industrial buildings more likely to be aware of LL97 and their options?