WEmap x Zonehaven
Wildfire Information Delivery for Non-Native Speakers
Overview
This project aims to augment Zonehaven AWARE by researching and visualizing the linguistic struggles of vulnerable communities in Marin County, CA. In this project, we, the WEmap team from UC Berkeley, collaborated with FireSafe Marin and Zonehaven and were supported by Wonder Labs’ Reimagining 2025: Living with Fire Design Challenge Program. This is an open-source project that might benefit all efforts in wildfire risk reduction. See the full project at: bit.ly/wemap_2021.
As a UX designer, I led the user research, ideation, and prototyping process to help improve accessibility to zone information for non-English speakers by incorporating multi-language features to Zonehaven AWARE. As the project lead, I also kept regular communication with stakeholders, including supervisors from Zonehaven and Wonder Labs, academic advisors from UC Berkeley, and community partners from Marin County.
Team
Virginia Wong
Yuquan Zhou
(Both GIS researchers)
My Role
UX Designer
Project Lead
Timeline
June-August, 2021
My Contribution
Problem research, User Research, Synthesis, Ideation, Prototyping, Communication with stakeholders
What Zonehaven told us
The linguistic aspect of information delivery
needs immediate attention
In the previous project, our team identified the lack of communication between first responders, support groups, and the community (evacuees) at different stages of wildfire which prevents successful resource allocation and drastically delays evacuation responses to wildfire. Through Wonder Labs, our team was able to connect with Zonehaven, a platform that shares a similar vision with our proposal, and an opportunity to integrate our interest in helping vulnerable communities into Zonehaven AWARE.
While the goal of Zonehaven AWARE is to provide the most reliable source for first-order evacuation updates and preparation resources, all information on Zonehaven AWARE is currently in English. However, the minority community usually speaks other languages. According to the recent Census tract, 14.7% of the population in Marin County speak other languages, such as Spanish, Asian, and Pacific Island languages including Chinese and Japanese, or Indo-European languages like French. This linguistic problem can directly affect the effectiveness of the platform within the community and also how information is delivered.
Number of limited English Speaking households at Census Tract Level, Marin County, 2019
Source: Team WEmap created in QGIS
This project aims to augment Zonehaven AWARE by researching and visualizing the linguistic struggles of vulnerable communities in Marin County. Specifically, based on our GIS analysis, we decided to focus on Spanish households in San Rafael as our target group of user research.
As an open source research project, we also aim to provide insights and recommendations that could be benefit other products besides Zonehaven, to improve information delivery for the non-natives.
Research
Understand different linguistic needs
We created our community survey in English and had it translated into Spanish with the help of a Spanish-speaking liaison from the Public Communications Division at the County of Marin. We sent the survey out both online on some Reddit groups with 798 responses and in-person with 20 responses and 5 interviews. I went to Cardenas Markets (330 Bellam Blvd, San Rafael) with the Spanish-speaking liaison for the in-person community engagement.
798 submissions online
Photo from the in-person session
Photo from the in-person session
20 in-person surveys + 5 interviews
Our next step was to clean up the data from surveys and generated insights from the data. Our team originally received 798 online survey responses. After cleaning out duplicate responses and contradictory responses, the number of valid responses is 703. We focused on the non-English speakers (86 people) whose native language is not solely English to generate insights.
Key Insight #1
People expect different information in alerts
Based on our interviews, people usually search for different information by themselves upon receiving the alerts.
Key Insight #2
Language needs vary among residents to understand the alerts
We focused on the non-English speakers (86 people from the survey) whose native language is not solely English, and learned that non-native speakers have different abilities in English and therefore have various needs regarding the languages
Survey data from all non-native respondents
Key Insight #3
Alerts should attract people’s attention immediately
From interviews, we understood that sometimes email alerts don't attract people's attention. From non-native speakers' submissions, we understand that only 20% of people usually find out emergencies through alerts.
Survey data from all non-native respondents
Analysis
.png)
Synthesis
Identify user behavior archetypes
From the data, I realize that the difficulties people were experiencing have a close correlation to their English proficiency, and how they receive alerts. After laying out all our insights on affinity maps, I identified 3 behavior archetypes for non-native speakers regarding their experiences to receive wildfire alerts, which are translators, autodidacts and late receivers.
Analyze users' emotions & feelings
Our next step was to analyze the 3 types of users' emotions and feelings upon receiving emergency alerts through creating empathy maps.
Map out 3 groups of users' journey
Translators and autodidacts are worried upon receiving the alerts, but they both become calm later when translators help autodidacts understand the alerts - helping people makes translators feel good, and confirming the information helps autodidacts become calm.
However, late receivers usually don’t receive and understand the alerts as fast as the others. They need help with translation and information search, and they feel worried if they are exploring by themselves because of their limited language skills.
How might we deliver timely and relevant information
in a language people understand?
Concept #1
Community support: translators → late receivers
When an emergency happens, late receivers usually need help from others. Translators or native speakers could sign up on Zonehaven to be volunteers to help them - call late receivers via phone, translate the alert in a language that the late receivers understand, or provide help with mobility.
Late receivers
need assistance
Volunteers
sign up and help
Concept #2
Translation: information shown in
zone's popular languages
After analyzing people's translation data, Zonehaven will show the most popular language in this zone on the information card. English will still be the default language. Users are able to translate the information card into another language by clicking on the “translation” button.
.png)
Translation charts: frequently-used phrases
in alerts in various languages
To help people with fast translation, translation charts including frequently-used phrases in alerts could be created on Zonehaven platform. The charts will help with the instant translation of alerts. When there is an alert, people can choose their preferred language to view the translated alerts and information.
.png)
Concept #3
Resources: wildfire-related info links on alerts
To help people with information search, displaying multiple wildfire-related resource links in different categories to the users is extremely helpful. All three groups of users would benefit from the information.
.png)
Fire Maps
Traffic
Air Quality
Concept #4
Visualization: display the alerts using doodles
Visualizing the alerts in doodles would be helpful for people to help people understand the alerts and this has been confirmed in the surveys that 43% of people think visualization is helpful. However, considering that people might have different interpretations of the same doodle, doodles might make the information delivery not as accurate as it should be. A method to create proper doodles fast and in high quality would be needed to achieve this concept.
.png)
Wireframing
Community support: sign up for help
Late receivers request assistance
When an emergency happens, late receivers usually need help from others. They could sign up on Zonehaven to request assistance when an emergency happens, such as to receive a phone call to get notified, or to be translated by someone speaking the same language.
.png)
Volunteers sign up to offer help
To help late receivers, translators or native speakers could sign up on Zonehaven to be volunteers to help them - call late receivers via phone, translate the alert in a language that the late receivers understand, or provide help with mobility.
.png)
Translation: get prepared for alerts
Show zone information in the popular language.
After analyzing people's translation data, Zonehaven will show the most popular language in this zone on the information card. English will still be the default language. Users are able to translate the information card into another language by clicking on the “translation” button.
.png)
Let people know what might appear in alerts.
Using the translation chart to get people prepared for what might appear in alerts. People are able to understand the different status of their zone that might happen, and what information they could get on Zonehaven AWARE.
.png)
Translation Chart in Spanish
Resource links: deliver information efficiently
.png)
Instant translation
Show the most popular secondary language in the zone as a shortcut and others in a drop down menu.
More useful links
When an emergency happens, displaying wildfire-related resource links and most-searched topics on alerts in different categories on the zone info card is extremely helpful.
.png)
Limitation & Assumptions
What needs more research
In this project, we assume all non-native English speakers, regardless of their native languages, have similar needs in terms of understanding English alerts. We also assume people expect to receive alerts in their native language, which is one of our criteria during the cleaning up process. These assumptions helped us set boundaries within the problem space. However, more research is needed to prove the statements and understand the actual needs of the community.
Impact
Making alerts accessible to everyone
This open-source research created augmentation concepts for Zonehaven AWARE and our deliverables was well received by the stakeholders. By publishing this project, we hope the insightful recommendations could provide valuable inputs to other products’ consideration on non-native English-speaking communities. We hope this research will ignite dialogues and have a positive impact on making alerts accessible to everyone.
My Learnings
Communication with stakeholders - this is an important part I learned in this project as project lead. We were fortunate to receive a grant from Wonder Labs to develop this project with Zonehaven. In May and June, we had several meetings to build up this partnership and to agree on a new proposal with a smaller scope than the original WEmap idea in this summer project. After the project started in June, I kept regular conversations with our supervisors from Zonehaven and Wonder Labs, academic advisors from UC Berkeley, and community partners from County Marin. Seeking a balance among these stakeholders is a precious learning experience for me.
Community engagement and user research - user research on Spanish-speaking communities was challenging because none of us speaks Spanish fluently. To better understand the needs of the underserved communities, we sent out surveys not only online, but also in person outside a grocery store in San Rafael. During the in-person sessions, I was able to conduct interviews with our Spanish-speaking liaison from the County of Marin, and I also felt the real language barriers between me and other non-native speakers. The learnings and findings from the in-person conversations were precious. If there’s more time, we would do another in-person session focusing on interviews and user tests.
A real product vs. an academic project - it’s so different to work on a real product! We received a lot more feedback from stakeholders than we expected on each step we moved forward. For instance, I came to understand the importance to identify different user behavior archetypes because it was impossible to have only one persona for a disaster response product. Data hygiene of our survey responses and collaboration we two other researchers on insights and assumptions was another important lesson for me. I came to understand that for a real product, research and insights on real problems are far more important than creating fancy mockups.