Author: Madeline Carlini

Hello potential CDS intern! We are the FloodBud team, with Maddie Carlini (Colby ‘21), Kutub Gandhi (Rice ‘20), and Jade Wu (UNC Chapel Hill ‘20) and we worked as part of the Smart Sea Level Sensors (SLS) team in the City of Savannah and Chatham County. The complex geography of Chatham County causes intricate differences in flooding patterns across the region, affecting local residents in different ways. So far, a single tide gauge at Ft. Pulaski has been used to monitor the entire Georgia coastline. However, this one gauge does not provide enough granularity to capture trends in flooding across the entire region. The SLS project is working to quantify the intricacies of this tidal system. 

Over the past year, the SLS  team has deployed over 30 sensors across Chatham County, with a goal of over 100 installed by 2020. Each sensor records water level, air temperature, and barometric pressure measurements every five minutes. As the sensor network expands, there is a growing need to monitor and maintain the sensors. Thus far, researchers must manually check the datastream for each sensor to assess its health. This process can become inefficient and time consuming. We aim to devise tools that will (1) streamline the monitoring of the sensor network for researchers and (2) make the sensor data more accessible to the public. 

In order to assist with sensor maintenance, we built an anomaly detection model to flag sensors that are outputting anomalous data. To give a primer on what our data looks like, refer to the plot below. The top graph displays normal, health sensor data, with normal fluctuations in the high and low tide. The highlighted chunks in the bottom plot show examples of anomalous data that we would want our model to detect. 

The issues with the sensors fall into two main categories: issues with the sensor itself and environmental signals. Issues with the sensor involves fuzzy, incorrect, or missing data points. On the other hand, environmental signals are more complex and are caused by weather events. Our model aims to flag both kinds of anomalies. 

The general idea of our model is to take Ft. Pulaski water level readings and apply vertical and phase shifts based of off each sensors’ output to create a unique fitted function for each sensor. We then use the past three days as our testing period. We calculate the least squared difference between the testing period data and the fitted prediction function for each sensor. The testing comparison are made over one hour, one day, and three day time windows. We then flag sensors if the test errors for any of the time windows are above a set threshold. 

We came up with two main deliverables to answer our research questions. To address sensor maintenance, we created an email alert service that sends our research mentor the sensors that are currently flagged for being anomalous. The email includes the names of the seniors sorted into groups for type of anomaly. In addition, plots of the sensors for the flagged time period are attached so that a human can check for what kind of event occurred. 

Our second deliverable is a public facing website to host the data visualization and exploration tools that we created. 

The website connects to our anomaly detection model by displaying the current day’s flagged sensors. The user can also see what kind of anomaly each sensor was flagged for. 

Thank you for tuning in! Hopefully by the time you read this, our website will be up and running 🙂

 

Things are coming together!

 

Our visualizations are going on their own website and we have an email system that updates the maintenance crew on the sensors that are causing trouble. While they aren’t fully pat and polished, they are extremely close. In addition, we have cleaned up or visualizations and added further interactivity to the plots. Our webpage now includes a leaflet map that displays circle icons for each sensor for added spatial information. The user can click on a sensor’s icon to toggle plotting the data. The icons also change from outlines to solid to indicate which are currently plotted. The tool can be used by Dr. Clark and citizens alike, including exploration features that can further highlight and inform people of coastal flooding in Georgia. 

Our main challenge is putting our materials on a server to be accessible to the public – our final product will look something like what’s pictured above. 

 

With our final week coming up, we are planning to make our final presentation and poster for the showcare Wednesday afternoon. After that, we hope to only have documentation and commenting left to clean up! 

 

This week we were able to take a trip to Savannah and Chatham County to explore the area. Mainly, we helped our mentor, Dr. Clark, carry out firmware updates on some of the sensors as well as scout locations for new sensors and gateways. Visiting the area gave us context to better understand the sources of the data that we have been working with all summer. We knew these sensors by names and ID’s, now we actually got to see where they were placed and why those locations were chosen. In addition, we had the opportunity to meet with Kate Ferencsik and Nick Deffely from the Savannah office of sustainability to learn more about the community engagement side of the Smart Sea Level Sensors project. 

With the end of the program approaching, we tied up some loose ends and gave some thought as to what we want our final deliverables to look like. For one thing, we’ve been working on the visualization (front-end) and event detection (back-end) separately; an immediate to-do was connecting the two components. Now, the front-end has an option to display an anomaly layer, which will show the anomalous sensors for that specific day. We’ve also been working off a static csv to feed into the visualization, which we’ve now updated to the live Sea Level Sensors API. 

 

In terms of final deliverables, we’re thinking of two complementary parts: one public-facing website and one tool intended for Dr. Clark. The former will be a primer on the context of the project and will give a brief description of our research question/methods, with a final pane displaying the front-end aspect of our visualization. For Dr. Clark, we’ll be sending him a more detailed anomaly report through an automated email, informing him of which sensors were anomalous for that day as well as their respective error values. 

 

After we met with Dr. Clarke and Dr. Cobb, our two advisors for the project, we not only gained a better understanding of the project, but also of our scope over the next ten weeks. Currently, residents of Chatham county are relying on a singular gauge in Fort Pulaski for flooding predictions. Although this sensor is fairly accurate, the problem is that it’s only fairly accurate for that specific region. During hurricane Irma and Matthew, this gauge predicted that the magnitudes of the two storms would be roughly equal. However, the deployment of temporary sensors by the USGS showed that, for certain areas, Irma showed more severe flooding than Matthew. This difference was not captured by the Fort Pulaski gauge, thus the need for the current network of sea sensors. With a better understanding of the project motivations, our research question thus comes in two forms:

• Why is a sensor network necessary as opposed to the singular Fort Pulaski gauge?
• How do we justify the network’s permanence? What nuances and insights can we learn from it?

Communicating this information poses a challenge because there is a plethora of datasets to consider. Namely, how can we intuitively, but also meaningfully, incorporate space (geography) and time into our visualizations? What should we consider when building a visualization that should highlight the nuances of the geography?

We began the week by creating a storyboard of the motivations and desired outcomes of the project to organize our ideas. We then brainstormed ideas for possible visualizations that aim to communicate the importance of the sensor network. The primary tools we intend to use are two javascript libraries called D3 and Leaflet; D3 for visualizing the sensor data we have, and Leaflet for placing that data on the maps. We are all beginners in the world of data vis, but we’re having a ton of fun playing around with the features the libraries have to offer. After creating some initial visualization examples, we met with Dr. Ben Shapiro to get his feedback on our ideas. He answered our questions on how to best express time and spatial components of our data and helped us sketch further options for more complex visualizations. We plan to spend the rest of the week updating our models based on Dr. Shapiro’s feedback and to continue to build the first versions of our brainstormed ideas.

 

Coastal flooding and sea level rise is a rapidly growing threat, especially for regions like Chatham County. On the Georgia coast, 60% of major flood events have occured since 2015, with both flood intensity and frequency increasing. Due to these changing environmental conditions, a better framework needs to be developed to better prepare and inform residents during three crucial time horizons. (1) Before a flood: providing informed, data-driven decisions of high-risk flood areas before a flooding emergency (2) during a flood: real-time information that can drive emergency responses and (3) after a flood: detailed assessments of compromised infrastructure.   

Our work this summer is based out Chatham country and the city of Savannah. We are working on an extension of an ongoing research project that uses sensors to monitor water heights at various locations along the coast. Currently, there are 30 sensors deployed (shown in the map below), with the goal of having 50 installed by the end of the summer. Water depth, air temperature and humidity data from these sensors are available via a public REST API.

The complexity of Chatham’s river systems combined with the increasing effects of climate change results in complex flooding patterns. We will focus on analyzing the sensor data to discover any underlying trends in water depths. More specifically, we plan to compare closely grouped sensors to detect more detailed patterns in flooding such as minor variations caused by local geography. Through data analysis and mapping visualization tools, we hope to gain insight into the occurrence of floods and the predictors surrounding flood events, ultimately using this information to create useful visualizations to communicate the patterns to the community.