Many have used the past year to take a step back and assess where their interests lie. For me, that has meant discovering my interest in data analysis and data science. I’ve been fortunate to be able to take several online courses to foster this interest and was recently looking for an opportunity to apply what I have learned to my work at SBC on the Climate & Energy team. After a few brainstorming sessions, I developed the idea to build a dashboard for the initial exploration of possible relationships between annual electricity consumption, time, and annual cooling degree days for counties in California. The basic premise I wanted to explore was what, if any, relationships may exist between how hot a year was in a particular county and how much electricity is consumed by the county. I chose to explore these relationships for several reasons.
There were 30 years of publicly available data for all counties in California
Temperature variation be can a major driver of energy demand for cooling (and heating)
How Hot Was It?
There are a plethora of possible variables that could be used to measure how hot a year was: average daily temperature, median daily temperature, number of extreme heat events, etc. I chose to use the annual number of cooling degree days because it is specifically related to energy use. According to weather.gov, “Degree days are based on the assumption that when the outside temperature is 65°F, we don’t need heating or cooling to be comfortable. Degree days are the difference between the daily temperature mean, (high temperature plus low temperature divided by two), and 65°F. If the temperature mean is above 65°F, we subtract 65 from the mean and the result is Cooling Degree Days. If the temperature mean is below 65°F, we subtract the mean from 65 and the result is Heating Degree Days.” (source: )
Degrees are typically used to help compare an individual building’s energy efficiency performance across different time periods, whereas this dashboard looks at the electricity consumption for an entire county.
How to Use The Dashboard
The maps shows annual residential and non-residential electricity use for each county, with shade representing per capita electricity use and size representing aggregate electricity use
The Year slider in the top left corner is used to select which year’s data is displayed on the maps.
Hovering over any of the bubbles will display that county’s data for the selected year.
Clicking any of the bubbles on the map will filter the rest of that dashboard so that only that county’s data is displayed.
The top right chart explores potential trends in county’s per capita electricity use over time using linear regression. The color of the data points in this chart represent the relative number of cooling degree days for a given year.
The bottom right chart explores potential trends in per capita electricity use based on the annual number of cooling degrees experienced by a county.
Like the maps, hovering over a data point reveals its precise values. Hovering over the line of best fit reveals characteristics about the line and its fit. In general, the higher a line’s R squared value, the better it fits the data. Lines with p-values greater than 0.05 are statistically insignificant and should be disregarded.
The current view is filtered to show only Nevada County data. To highlight a specific county’s data without filtering out other counties’ data you can use the “Highlight County” search bar in the top right corner.
Hopefully these instructions make it easier to begin exploring the dashboard’s features but, I think the best way to get familiar with it is to dive in and play around with it yourself, so go ahead and give it a try!
In general, the trend between Year and Per Capita Electricity use is more statistically significant than Annual Cooling Degree Days for a County. This claim is supported by the generally lower p-value of the trend lines in the top right chart than those of the bottom right chart. The general cutoff value for statistical significance uses a p-value of less than 0.05. Some possible reasons for this finding could be that the year variable captures variance attributable to trends in cooling degree days (annual number of cooling degrees could be increasing over time) in addition to variance attributable to energy efficiency installations (more energy efficiency technologies are being installed and implemented over time) and behind the meter renewables installation (more residential and small commercial solar energy are being installed over time).
Future Exploration
This initial exploratory dashboard exercise presents numerous opportunities for further exploration. Of greatest importance would be a closer examination of the independent data sets and linear regression models to determine if all the assumptions of linear regression are being met. If they are not the accuracy of the model’s fit could be exaggerated. Once the underlying assumptions of linear regression have been confirmed, I believe there is opportunity to develop a more robust regression model that incorporates data on number of energy efficiency installations and behind the meter renewables capacity, with the ultimate goal of creating a more accurate model that can improve the model’s accuracy and possibly even be used for predicting future electricity consumption levels.
Core to Sierra Business Council’s mission is our commitment to being a triple bottom line organization so it’s only natural that we would lead an effort to help small businesses adopt these principles. A triple bottom line business considers more than just economic results or profit. They also value their impact on society and the environment. We often refer to these businesses as “sustainable” or “green” and research shows that they typically deliver superior customer service, score higher in job satisfaction, and perform better financially.
As you know, California witnessed its worst wildfire season on record in 2020. Over 4 million acres burned, costing us more than 10,000 homes and buildings and at least 33 lives (not to mention the long-term negative health impacts of the dense smoke experienced across the state). Wildfire is a natural part of California’s landscape, but today’s wildfires are out of balance due to a century of fire suppression and misguided forest management. As climate change accelerates the risk of extreme wildfire, 2020 could be the start of our new normal. In 2021, we have an opportunity to keep that from happening. If we take action now, we can help protect our communities and restore the health of our forests.
I grew up climbing on granite slabs at Donner Summit, and the rock formations fascinated me. Everywhere I went in the mountains, I found myself mesmerized by the colors, textures, and stratigraphy lines that painted the landscapes. Having grown up in Northern California in an outdoors family, the concept of conservation was ingrained very early. “Respect the playground; if you want the beautiful places you love to remain intact, then do your part.” At that point in my life, I knew I wanted to do something that allowed me to be outside and in the field solving problems (or something to that extent). Naturally, I began my academic career pursuing a degree in geology.
You may have recently heard that head of the EPA, Scott Pruitt, announced the agency intends to roll back Obama era fuel economy standards for vehicle fleets. There has been some confusion, including my own, about what this announcement actually means compared to how it is being portrayed in headlines.
On January 22nd, President Trump approved a tariff on imported solar cells to safeguard US producers from cheaper international prices. The tariff will begin at 30% the first year and decrease annually by 5% over the next four years.
The maps shows annual residential and non-residential electricity use for each county, with shade representing per capita electricity use and size representing aggregate electricity use
Like the maps, hovering over a data point reveals its precise values. Hovering over the line of best fit reveals characteristics about the line and its fit. In general, the higher a line’s R squared value, the better it fits the data. Lines with p-values greater than 0.05 are statistically insignificant and should be disregarded.
