ReNewport Project Manuscript

From Norsemathology
Jump to navigation Jump to search

Title

An Air Quality Dashboard for the Citizens of Newport, Kentucky

Authors

Nelum Hapuhinna, Tyler Massie, Daniel McGee Marin, Ethan Davis, Andy Long

Abstract

In this paper we describe a dashboard designed to effectively communicate a variety of air quality data and results based on statistical analysis of that data to citizens of Newport Kentucky (and the surrounding Cincinnati area). A wide range of publicly available data is used to generate geographically based maps, time series of real-time and historical data by location, and predictions of air quality (as well as air quality alerts).

The code for the dashboard is open-source, and easily transferred to other areas of the country. The models underlying some of the displays is local to Newport, however, and the Ohio River valley. Nonetheless, many features will transfer easily to offer useful information.

Introduction and History of the Project

ReNewport is a non-profit organization whose goal is to enhance the quality of life of the people of Newport, Kentucky. One important issue to everyone is air quality: ReNewport partnered with RISE (Research Innovations using Sensor Technology in Environmental Justice Communities -- RISE Communities program) to install a dozen PurpleAir monitors which collect various measures of air quality around downtown Newport.

Our mission was to analyze the data collected, and to present it to the citizens of Newport in a comprehensible fashion via a web-based dashboard. We became involved when the director of ReNewport reached out to Kristy Hopfensperger, Director of the Environmental Science Program and Professor of Biological Sciences at Northern Kentucky University for assistance. She in turn approached members of the Department of Mathematics and Statistics (Andy Long and Nelum Hapuhinna), who agreed to participate and formed a working group featuring three students (Tyler Massie and Daniel McGee Marin of Northern Kentucky University, and Ethan Davis of Highlands High School).

The students brought their considerable programming talents primarily to the design and mechanics of the dashboard itself, but were also involved in the development and implementation of some of the statistical techniques for generating maps based on the data collected.

One of the challenges is the variety and ephemeral nature of the data on which models are based: while we had historical data from the EPA, and some long running PurpleAir data, monitors from PurpleAir pop in and pop out, meaning that at any one time only a portion of the data is actively contributing. Furthermore, we had the good fortune to have a plethora of various air quality data from the summer of 2024, from a GroundWork Study that ran concurrently to our own. The challenge was, however, that this data did not persist past the summer. So the challenge was to build models that used a variety of air quality data, but from monitors that dropped in and out.

The Dashboard

Design Criteria

Minimum Viable Product (MVP)

PurpleAir provides its own dashboard, which served as a Minimum Viable Product (MVP); our objective over the eight weeks of our research project was to duplicate that dashboard (at least). However we were able to far exceed the basic dashboard that PurpleAir provided.

User Interface

Because this dashboard was to serve the citizens of Newport, Kentucky, our goal was to create a dashboard that would be intuitive, and simple enough that anyone could use it.

Competition

R, R Shiny, Leaflet

Grafana

"Themes"

Minimalist/Mobile

Expansive/SpaceShuttle

The Database

Data Wrangling

Data Types and Sources

Monitors dropping in and dropping out

Missing and Censored Data

Using transient data to help develop models

Making models adapted to flexible data

Modeling

What is "Air Quality"?

AI and Deep Learning

Traditional Statistical Techniques

Timeseries Analysis

Variogram analysis

Kriging, Cokriging,

Conclusions

Acknowledgements

Northern Kentucky University's Center for Integrative Science and Mathematics provided summer research funding (UR-STEM) for Hapuhinna, Long, and Massie; NKU's Department of Mathematics and Statistics provided funding for McGee Marin. We gratefully acknowledge their support.

We thank Kristy Hopfensperger of NKU for tirelessly organizing those groups working on the important issue of making air quality information accessible to citizens.

Appendices

References

Time Series Analysis in Meteorology and Climatology: An Introduction: Claude Duchon, Robert Hale; ISBN: 978-0-470-97199-4; December 2011; 262 pages.