Outputs

Digitizing with AI Workshop: RapidID

Led by: Drishtie Patel & Subbu Subramanian

Drishtie Patel and Subbu Subramanian from Facebook ran a short mapping workshop that introduced the use of RapidID in the digitization process of road networks on OpenStreetMap. RapidID is a tool, developed by Facebook using machine learning techniques, to map roads more efficiently.

Making Maps with Blender

Led by: David Garcia

David Garcia led us into a cartography-oriented session, showing us how to use terrain topography layers such as Digital Elevation Models in QGIS to then render them in visually appealing ways in Blender. The result are more aesthetically pleasing maps that have a topographical base map look.

The image of Valles Marineris on Mars on the right was created by Jeff Underwood using Blender, following the workshop by David.

Mapathons on OSM

By: Adityo Dwijananto, Sheila Amalia, Tri Selasa Pagianti, David Garcia, Gautam Dadhich, Raksmey Ang, Kunio Pratopo, Ashrika Sharma, Giuseppe Molinario, Robert Soden

Multiple mapathons were conducted to add missing geographical information of Chiang Mai onto OpenStreetMap. The Nonghoi neighbourhood is one such area where much of these work was done.

Reflection: Building Vulnerability Mapping with ODK

By: Rebecca Bicksler, Alina Bill-Weilandt, Ashrika Sharma, Kunio Pratopo, Raksmey Ang, Wahaj Habib

Two approaches to field surveys have been explored during the field lab. A field survey, led by Becca Bicksler, Alina Bill-Weilandt, and Ashrika Sharma, was conducted using Open Data Kit (ODK) to collect information on building vulnerability to floods in the Nong Hoi neighbourhood. This survey spanned across three weeks, starting with the survey design process where the team created and tested a list of survey questions. The actual field survey and data analysis took about 1.5 weeks to complete.

Vulnerability Mapping in Nong Hoi Poster

By: Rebecca Bicksler, Alina Bill-Weilandt, Ashrika Sharma, Kunio Pratopo, Raksmey Ang

During the field survey information related to various building attributes were collected. The characteristics such as the elevation of the structure relative to ground, number of stories and building heights which are important factors in accessing the vulnerability of the buildings to a flood event were the focus. The data collected was in a GIS readable format. The field data was collated with the existing OSM data available and a query based GIS analysis was performed to access the vulnerability in the Nong Hoi community.

Chiange My Map: Participatory Mapping

Led by: David Garcia

Often mappers forget what they are really working towards and what “spatial-intelligence” is really about. To get back to the first step and think about why mapping is so effective and important, “Mapmaker” David Garcia led us through an exercise of “participatory-mapping for the mappers”. In slowing down to add our own experienced features to the map, drawing and coloring them by hand, we had time to tangibly come back to the experience of understanding the relationships of things on the map: a road, a river, a canal, a building, a park – how they all spatially relate. This tactile break allowed us a rare moment of dialogue about the basics of cartography and geographical information systems. This was appreciated by many of the mappers at the un-cnference, from those with little experience to those with multi-decadal GIS and remote sensing backgrounds; all appreciating the respite from computer work.

Critical Infrastructures Mapping in Nong Hoi

Led by: Sheila Amalia and Tri Selasa Pagianti

Field survey on critical infrastructure mapping was carried out in the Nonghoi neighbourhood. Led by Sheila Amalia and Tri S. Pagianti from Humanitarian OpenStreetMap (HOT OSM), this survey was conducted using field papers and OsmAnd. Critical infrastructures that are important for disaster response were collected and uploaded to OSM over the span of two weeks.

Video compilation of acquired images using Drones and 360° Camera

Compiled and edited by: Wahaj Habib, Maricar Rabonza

Showcased in this video are the products of exposure mapping using novel technologies – drones and 360° cameras. A fly-over animation of the 3D context model was developed from the drone images obtained in Mae Chan Chiang Rai. This video also exhibits the street-view images with 360°camera acquired at the same location and uploaded to Mapillary. Such visualisations present the level of quality and efficiency that these technologies can contribute in exposure mapping.

Nature-based Solutions Factsheets

By: Laddaporn Ruangpan, Perrine Hamel, Missaka Hettiarachchi, Bhola Saha, Bhen Aguihon, Ashrika Sharma, Shelly Win

The Nature-Based Solutions working group has produced 4 fact sheets on the potential of using nature-based methods (from the basin scale to the the neighbourhood scale) to tackle the problem of flooding in the Ping River Basin.

The Flood Table

By: Karen Barns, Laddaporn Ruangpan, Alex Curran, Ashrika, Jennifer Joy Chua, Rebecca Bicksler, Perrine Hamel, Missaka Hettiarachchi

The nature-based solutions team built an interactive flood table for the Art & Science Exhibit. This piece aims to explain the concept of nature-based solutions for flooding problems by allowing the audiences to understand the interactions between the natural systems and society.

Persona Cards for Nature-based Solutions

By: Tian Ning Lim, Rachel Siao, Pamela Cajilig

Inspired by the discussions surrounding the use of personas and the complexities of urban flooding, Lim Tian Ning and Pamela Cajilig, with the help of Rachel Siao designed a deck of persona cards. The aim of these cards is to help users understand the complexities of urban flooding. This cards will be published on Figshare. 

Application of HEC-RAS Model for Introducing Flood Mitigation Measures at the Upper Ping River, Chiang Mai City

By: Shelly Win, Thanasit Promping, Raksmey Ang, Kunio Pratopo

The objective of this study is to investigate the benefit of introducing channel modication and levees as flood mitigation measure in the upper Ping River. The HEC-RAS— an opensource flood model — was used to produce flooding scenarios. A poster was used to summarize the findings of the study.

The Flood Green Guide Workshop

Led by: Missaka Hettiarachchi, Perrine Hamel

Missaka Hettiarachchi from World Wildlife Fund (WWF) led a workshop on the Flood Green Guide and conducted a serious game aimed at educating decision makers about nature-based solutions. The Flood Green Guide was developed by the WWF to support local communities around the world in using natural and nature-based methods for flood risk management. Participants of the workshop were requested to provide feedback to improve the educational activities around the Flood Green Guide.

Stories from Mae Chan

Produced by: Rachel Siao, Karen Barns and Thanasit Promping

In Mae Chan district, our participants Rachel Siao, Karen Barns and Tor Thanasit interviewed elderly residents on urban flood problems in Chiang Mai. Their shared experiences talk about common flood mitigation strategies employed by the locals and their feelings toward them.

Field Work with OASIS Lab in Mae Chan – Flash-Flood Early Warning

Led by: Paskorn Champrasert, Autanan Wannachai, David Lallemant, Giuseppe Molinario, Thanasit Promping

Field Lab participants went with the OASIS Lab of Chiang Mai University to Mae Chan community near Thai-Burmese border to explore the possibilities of installing a new flood sensor prototype developed in the previous week. The OASIS lab has worked with the Mae Chan community for many years, as it regularly suffers from flash floods. They have installed a flood-sensor upstream and designed web-based platform to inform community leaders and the public of imminent floods.

Flood sensor prototyping workshop

Led by: Autanan Wannachai, Paskorn Champasert

A new flood-sensor prototype design was developed with Prof. Paskorn, Autanan and the OASIS lab of Chiang Mai University. Prof. Paskorn shared with participants the complexity of operational flash-flood early warning systems. Designing effective systems requires novel sensing technology, but also designing maintenance procedures at the local level, information systems to inform downstream communities of imminent floods, confront issues of trust in technology without which warnings are not headed, and many more. We focused on improving the current system by adding additional optical sensors to increase reliability of the early-warning system. We also explored filters and machine learning methods to better understand the relationship between water level increases upstream and flooding downstream. The prototyping workshop was followed by a feasibility testing and information gathering in the field, focusing on the Mae Chan community near Chiang Rai.

Miki: The Smart Flood Sensor Photobooth

By: Autanan Wannachai, Wahaj Habib, Maricar Rabonza, David Lallemant

The improved flood-sensor designed through the field lab uses an ultrasonic sensor to measure changes in water level. If it increases past a certain threshold, the camera is triggered to take a photo for aditional reliability before the early-warning goes out. For the art & science fair, the sensor was repurposed into a photo-booth. The ultrasonic sensor is used to identify if someone is standing in front of the sensor, which then triggers the photo-booth camera. Pictures were automatically uploaded to a server, which people could download from the barcode provided at the entrance. Add a instagram frame and props for extra fun!

Machine Learning Vision Paper Outline

By: Dennis Wagenaar, Alex Curran, Mariano Balbi, David Lallemant, Alok Bhardwaj, Gizem Mestav, Gautum Dadhich, Carmen Acosta, Emir Hartato, Giuseppe Molinario

Increasing amounts of data and machine learning algorithms to analyze this data are causing changes in almost every aspect of our lives. This trend is expected to continue as more data becomes available and machine learning algorithms become smarter. It is also affecting and will continue to affect flood risk and impact assessments. Such assessments are being used to plan mitigation measures, prepare and respond to floods and help with flood recovery. For both predictive and descriptive, flood hazard, exposure and impact assessments, machine learning methods have the potential to improve accuracy reduce calculating time and reduce model development cost. Machine Learning methods have especially in academia been applied for a long time, but the methods are often not feasible yet for practical application. It is expected that in the future more applications become feasible and many process models and observations will be replaced by Machine Learning. This paper reviews the current use and future potentials for the application of ML in these areas.

Data Training: Application of deep learning for flood extent detection

By: Alok Bhardwaj, Raksmey Ang, Kunio Pratopo, Gautam Dadhich, Thanasit Promping, Carmen Acosta, Feroz Khan, Giuseppe Molinario, Robert Soden

This group attempted to develop deep learning architecture to estimate flood extents using microwave remote sensing in an automated fashion. Flood labels are required to run deep learning models to estimate flood extents. Flood labels is hard to generate and is a perfect candidate for community mapping. The project involves digitizing 4 to 5 land cover classes including flooded areas. The input data is a Digital Globe imagery for a flood in Phitsanulok in 2017. This flood event is selected because it has both microwave and optical data with one-day difference. Microwave remote sensing data is used as input to deep learning models for which labels are created using optical data in a QGIS environment.

The Application of Deep Learning to Estimate Flood Extents for Urban Areas using SAR Amplitude and Coherence Time-Series Data: Challenges and Recommendations

By: Alok Bhardwaj, Sang-Ho Yun, Emma Hill

AGU Abstract Submission:
The application of deep learning to estimate flood extents for urban areas using SAR amplitude and coherence time-series data. Specifically, we investigate the use of a temporal convolutional neural network (TCNN) to identify flooded and non-flooded regions for different land-cover classes. We explore the labels needed to successfully apply the TCNN and showcase strategies to tune hyper-parameters of the TCNN 

Framework development: Spatial Transferability of Damage Models

By: Mariano Balbi, David Lallemant, Dennis Wagenaar, Maricar Rabonza, Alex Curran

One key challenge in the widespread use of flood-damage models is lack of data. While models are typically callibrated in data-rich contexts such as Europe and the United States, they are nonetheless applied in widely differing setting, with little attention to the potential biases this may introduce. Bringing together leading flood damage modelers, we explored methods to assess the applicability and transferability of flood damage models from context to context. We identify variables most important in determining flood damage. From this we explore various methods for measuring ‘similarity’ between contexts. We test this approach using an extensive dataset of typhoon damage in the Philippines. Damage models are calibrated for four different regions, and then applied to all others to predict damage. This enables us to study the perfomance of transfered models with respect to their “similiarity” measure.
During the field lab, participants discussed this issues, exchange and discussed relevant references, identified promising methodological approaches, identified and obtained relevant data-sets to test these methods, conducted preliminary analysis and drafted outline for an upcoming academic journal paper.

Infographics on Machine learning, Deep learning and Decision Trees

By: Mariano Balbi, Dennis Wagenaar, Alex Curran, Laddaporn Ruangpan, Maricar Rabonza, Lim Tian Ning

The team created three fact sheets on machine learning for the Art & Science Exhibit.

Artistic use of AI to combine Paintings and Flood-related Photos

By: Alex Curran, Maricar Rabonza, David Lallemant, Mariano Balbi

The team combined art and machine learning, under the theme of flood and Chiang Mai, by using the Deep Dream Generator. Artistic styles of well-known art pieces were applied to images of floods and Chiang Mai city using the Deep Dream Generator.

An Overview of Disaster Governance in Thailand with a Focus on Flooding: Chiang Mai Case Study

By: Gurusaravanan Manoharan, Xavier Asuncion

This brief aims to provide an overview of disaster governance in Thailand with a focus on flooding. This study made use of information and data from related literature and government documents such as the Disaster Prevention and Mitigation Act of 2007 and the National DRM Plan of 2015. These were used to provide a snapshot of how the government manages risks.

Chiang Mai: Information Ecosystem Mapping

By: Amy Willox, Jennifer Joy Gadher, Sunayana Sen, Rebecca Bicksler

During the field lab, participants from the Resurgence Urban Resilience Trust (resurgence.io) led a working group around information ecosystem mapping (IEM) for floods in Chiang Mai. They developed an IEM with the Nong Hoi community in Chiang Mai city by conducting interviews with residents, meeting with local organizations and identifying the needs of disaster managers.
Posters of the ecosystems mapping process and Chiang Mai’s information network were created at the end of the field lab.

Oral Histories of Flooding in Chiang Mai

By: Feroz Khan, Robert Soden, Jeff Moynihan

Over the month, participants worked to film and conduct interviews with flood-affected residents of Chiang Mai to provide a space for them to tell their stories. These interviews ranged from hour-long oral history interviews mapping the life courses of flood-affected residents to video interviews filmed on-site with residents describing the process of recovery. In particular, the neighborhood of Nong Hoi was focused on, thanks to the partnership with the Foundation for Older Persons’ Development (FOPDEV). This in turn was translated into process maps, visual reports and participatory maps.

Serious Games: Help me Design my Game

By: Poocharoen Ora-orn, Nirattigorn Sangdee, Giuseppe Molinario, Maricar Rabonza, Jennifer Joy Chua

The School of Public Policy (SPP) of Chiang Mai university has been researching methods of communication and understanding of complex issues, such as air quality and urban flooding. These issues are multifacted and complex, not only in the physcal processes but as much, if not more, in the socio-cultural perceptions and behaviours that allow them to occur, affect them, and are elicited in response to them. Using games is one way to allow citizens to understand the issues around these subjects. Serious games are just that — games that ultimately discuss very serious issues. SPP led a day long workshop during the un-conference where they tested prototypes of serious games and brainstormed together with the lab participants  on issues revolving around flooding. Game design, or the development of goals, structure, rules of the games, was as important to making the games compelling as designing “characters” that would populate the games. Making the games fun, and humorous was also of principal importance, as through humour, complex issues are communicated and resonate for longer.

3D Topographical Projection of Chiang Mai

By: Thiegnburanathum Poon, Nirattigorn Sangdee, Giuseppe Molinario

Our friends and collaborators from the Chiang Mai University School of Public Policy created a facilitated dialogue 3d display of the city, with overlaid GIS layers of different themes projected onto it. The display allows SPP to have a dialogue with participants of workshops on contemporary urban issues – including flooding – while also providing the context of the ‘Living with Water’ exhibition to animate that conversation.

Toolkits on Flood Risk Perceptions

By: Pamela Cajilig and Sneha Malani

Our field lab participants, Pamela Cajilig and Sneha Malani, created toolkits for understanding flood risk perceptions of the elderly residents in the Nonghoi neighbourhood. The toolkits include performance and drawing to overcome language barriers between the field lab participants and Nonghoi residents.

Flood Mindmap

By: Sneha Malani, Gurusaravanan Manoharan, Xavier Asuncion

Chiang Mai’s flooding is a complex issue and only understandable when viewed from multiple perspectives. Through the interdisciplinary work, participants had the opportunity to understand this complexity and view issues and solutions from various perspectives. A mind map to communicate, deconstruct, and analyze the various inter-linkages between stakeholders, the environment and their impacts on the city’s flooding was produced by three field lab participants.

Augmented Reality (AR) sandbox

By: Giuseppe Molinario

During the UR Field Lab in Chiang Mai, in June, I had the opportunity to experiment with the creation of an “AR-sandbox”. The AR-sandbox is an open source project developed by the University of California Davis, with an online community that has created everything from how to videos and DIY step by step software installation procedures. The piece created was in the context of an art exhibit called “living with water” closing a month-long un-conference on the topic of urban flooding in Chiang Mai…

Short Documentary: Perceptions of the People

By: Olivia Jensen, Jungsuh Lim

Over the course of one week during the field lab, Dr. Olivia Jensen from the Lee Kuan Yew School of Public Policy in the National University of Singapore and artist Jungsuh Lim, worked on collecting and translating footages of peoples’ views on issues surrounding floods. These footages will be used to create a short documentary aimed to convey the thoughts of people to decision-makers.

Storymap: Living with Water

By: Giuseppe Molinario

Some of the work done at the field lab was most adequate for displaying in a story map. This work was fashioned inside a story map platform and then submitted to #vizrisk challenge by GFDRR WB and Mapbox. The ultimate goal was to showcase some of the more compelling projects done during the un-conference on an interactive map; a powerful storytelling tool.

Interactive Installation: City Interactive Dialogue

Our friends and collaborators from the Chiang Mai University School of Public Policy created a facilitated dialogue table to allow exhibition visitors to discuss contemporary urban issues – including flooding – while also providing the context of the ‘Living with Water’ exhibition to animate that conversation. 

Workshop: Personas for Understanding Perceptions of Risk

By: David Garcia, Pamela Cajilig, Robert Soden

Personas are fictional characters developed from real-world experiences to help designers identify, understand, and empathise with project stakeholders whose challenges and priorities need to be ordered and reconciled. An introductory workshop on how to create and use personas was led by Pamela Cajilig, David Garcia, and Robert Soden in week 2 of the field lab.

Blog Post: Living with Water in Chiang Mai

By: Giuseppe Molinario, Sneha Malani, Sue Lim, Pamela Cajilig and Wahaj Habib

The experience in risk communications, and particularly in visualizing risk, was written up in a blog post published on Medium. The blog post served also as an entry to the #vizRisk challenge, organized by UR with Mapbox and the Data Visualization Society. The blog focused on those projects developed during the Un-conferene that had tanngible results in terms of risk communications and particularly in data and risk visualization.

VisRisk Competition Submission

By: Giuseppe Molinario, Jungsuh Lim, Pamela Cajilig

A blog post and mapbox-based story map provided an entry to #vizrisk challenge. The projects highlighted were all developed during the month of the un-conference, with almost no previous work. Some of these were quite firmly related to human geography: particpatory mapping, oral-histories and serious games for example, all analog risk visualizations with strong anthropological and socio-economic components. Others were more “traditionally” GIS and satellite remote sensing analisyes and visualizations of risk data.