I’m zipping through traffic in Chiang Mai, Thailand, on a scooter with a front brake that sticks and intermittent lights — a cheap rental for the month. It’s 3pm on a July afternoon, the overcast skies have finally given respite from the relentless heat, and the daily routine of rain has finally returned, after a relatively dry wet season, the locals say. A stark contrast with the heavy monsoonal rains ravaging a lot of South Asia right now.
I’m on my way to an art space that has been converted masterfully and beautifully from an old ice factory. There’s a coffee shop that has some of the best coffee I’ve ever had and some pretty great pain au chocolat, there’s a florist, and tables strewn with Monocles and architecture magazines. There’s usually a barrage of more or less amateur photographers and models constantly using the space’s unusual spaces as a backdrop. I was part of an exciting art workshop here, merging art and science to communicate disaster risk, specifically from flooding.
The art exhibit was called “Living with Water” and it closed the rather atypical month-long un-conference on the topic of urban flooding. Under the umbrella of Understanding Risk, during the Chiang Mai Urban Flooding field lab we flew drones and acquired high resolution digital terrain models (rigorously shared freely on GeoNode and OpenAerialMap), we drove around in Songtaws (the typical Chiang Mai red truck taxi) acquiring 360° imagery that we then posted on Mapillary. Mapillary is a free and open repository of street level imagery, similar to google street view, which by the way just reached over 600.000.000 images uploaded to it.
That was just the first week. In the three weeks that followed the smorgasbord of creativity and intense interdisciplinarity led to a variety of projects, among which the development of a zine documenting the un-conference process, low-cost sensors for a flood monitoring project in the Chiang Rai region, and lectures and workshops with folks from Planet labs, NASA, Facebook, World Bank, WWF, the Chiang Mai Department of Disaster Prevention and the School of Public Policy of Chiang Mai University. When that was done, we rested for a few minutes, then we flew more drone missions, worked with flood modelling using machine learning methods and held trainathons to train algorithms to classify flooded areas using Synthetic Aperture Radar (SAR) data — (a trainathon is like an OSM mapathon, but digitizing features with the goal of of training an algorithm — you heard it here first). A working group also focused on planning nature-based solutions for flood prevention and mitigation.
In this blog we highlight some of the important work we did over the span of the un-conference. All the projects had in common a focus on the holistic understanding and communication of risk. That communication is often visualization and we therefore decided to submit our processes and results of data visualization to the Understanding Risk’s #VizRisk challenge. To do this we complemented this blog post with a mapbox-powered storymap, using the powerful (and open source) storymap.js tool developed by the knightlab. The storymap is embedded below (and also available full screen here).
Flooding, Chiang Mai, and The UR Field Lab
Flooding is a pervasive problem in Chiang Mai. Chiang Mai is the second largest city in Thailand, with a rich history and both religious and cultural importance. Its 723 years have been marked by water. In the northern Lanna region of Thailand, where the Lanna Kingdom ruled here in and Chiang Mai, nestled at the base of the sacred Doi Suthep mountain, was its capital. The Lanna kingdom ruled until the end of the 1800's, and translated to the “kingdom of a million rice fields”. Nowadays, this expanding metropolitan area is a million people strong instead and the rice fields have been replaced by urban and peri-urban settled areas. Agricultural burning induced haze has become the predominant issue in this city and the entire region. But haze and flooding are two faces of the same coin. They are both primarily caused by lack of effective regulation and planning of land use.
As the naturally permeable surfaces are replaced with man-made impervious ones such as concrete and asphalt the water falling in monsoonal rains rushes into flash floods as opposed to being absorbed, slowed down and directed by nature. Flooding happened here before urbanization, but now it is worse and the encroachment of housing (that is not traditionally raised anymore) on flood plains, means that vulnerable communities bear the brunt of floods.
Throughout the entire un-conference, we played along with the idea of mashing up analog and digital tools, creating our analog facebook (lower-case “f” is not a typo), enforcing a strict start time every morning at 9am where we would start with announcements and then proceed to add sticky notes of impromptu sessions and follow-ups to previous sessions, on our analog schedule.
All of this while our Slack workspace was getting blown up by a flurry of activity and communication. Sometimes from folks offsite, sometimes in other countries as they had left already, not arrived yet, or not coming at all. Sometimes across the room silent communication (“hey your ice coffee is about to fall…”). Sometimes hating it “ what we really need here is just another Slack Channel” as Robert Soden, a co-organizer would say, but all in all having hit a sweet spot of digital/analog work that gave us all the tools we needed, but also allowed for actual human interaction throughout the day.
Visualizing Urban Growth / River Dynamics during the period 1984–2018
The visualization was created to understand the Land Cover and Land Use Change (LCLUC) dynamics in Chiang Mai in the context of urban growth and river dynamics (Ping River). The GIF animation below was generated using Landsat satellite imagery. The time series is based on yearly composite images from 1984–2018. The images are rendered as false-color created by ratios of different values in the images. Specifically, settlements and built up areas are shown as yellow and red shades. The dramatic growth of the city can be clearly discerned over the years with the city expanding in all directions. This rapid and irreversible urban growth is a primary driver in changing Land Use dynamics and a cause of increased flood risk and damage to the natural ecosystem as well.
Chiang Mai Flooding Timeline
Chiang Mai floods more often than is perceived. These floods range in severity as does the damage that they cause. Preliminary literature review allowed our participants researching the history of flooding in Chiang Mai to pull together a timeline of flooding in the city since the 1950’s. Flooding was also split into thematic sections: demographic changes, Ping river dynamics & urban expansion, urban development and flood responses.
Walking as Ethnographic Research:
Contextual Understanding of Flooding in an Urban Neighbourhood
Pamela Cajilig had been wanting to test out the idea of walking as a form of ethnographic research even before arriving in Chiang Mai, interested in applying this notion to her PhD research. Working together with the local NGO Forum for Older Persons Development (FOPDEV) the neighborhood of Nong Hoi was selected as a community to focus on for a lot of the qualitative work being done at the un-conference. In particular Pamela and many other participants assisted FOPDEV in conducting participatory mapping and disaster preparedness planning workshops for older persons in the community. Nong Hoi was is a district located Southeast of the Ping River. The area’s proximity to the river made it particularly vulnerable to flooding.
“Flood Modelling” 101
The ethnographic experience in the Nong Hoi community of Chiang Mai also led several of the participating researchers to realize the importance of visual and non-verbal ways to communicate disaster risk. This is not just important in areas where there might be low literacy, but also to overcome language barriers. Seeing the Nong Hoi residents pointing and measuring flood water height with their bodies led us to understand that flood is experienced through the body, and the knowledge that flooding is in part collected through the senses.
One of the exercises developed therefore uses the image of the body as a way to anchor discussion about the flood, so that verbal data about the flood can be complemented with visual data. The facilitator points to different points of the body to indicate flood levels: ankle, shin, knee, thigh, chest, and neck. With each flood level as indicated by body part, the group is asked whether they would stay in their houses or evacuate.
Head Level. In order to safeguard their possessions, the community would stay in their homes (on the second floor) as far as possible, even if the flood reaches head level.
Shoulder Level. Residents in low lying areas within and around Nong Hoi with no second floors evacuated. Residents with second floors moved themselves upstairs with their belongings.
Hip Level. Experience starts to dramatically diverge at this point, with a participant from a low-lying area stating that this would already equate to neck level in his house. Participants who only have first floors said they prepared to evacuate.
Knee level. Participants, especially those in low lying areas, reported that they started to get concerned about the flood level and start to frequently monitor news on the radio.
Ankle level. All participants, regardless of location and house type, said they did not evacuate
Vulnerability Mapping: Putting Nong Hoi on the Map
The focus on the Nong Hoi community highlighted right away the absence of a map of critical infrastructure for this area. A first step was surveying existing data on OpenStreetMap which revealed how many objects such as buildings, roads and amenities were missing. Using the OSM Tasking Manager, participants were able to map around 90% of the Nong Hoi neighborhood. The figure below shows the before and after OSM map.
After OSM mapping using satellite imagery, the team went to the field to gather attribute information about the mapped features. The team used a digital survey created with the OSMAnd and Field Papers. The map below is the result, putting Nong Hoi on the map with much more information useful for vulnerability mapping than existed before.
The data obtained from satellite imagery and from the field was then also analyzed in QGIS to obtain several other visualizations of critical infrastructure features, for example building height, building elevation, number of floors in the buildings. These features were all summed up in a final building vulnerability map for the Nong Hoi area.
A Shelter Timeline
Crossing language barriers for understanding risk
How do you cross language barriers in understanding risk? How do you cross cultural barriers? One way is to use visualization. The elderly community in Nong Hoi drew for us their housing typology from the 1980s, and during the two major floods of 2005 and 2011. The housing types were divided into 3 types. 1) on stilts, 2) on the ground level and 3) with more than one story. They roughly marked the flood level from that year. This led to discussions on whether they felt safe in their shelters, and whether/how their homes are resilient to potential flooding.
More people lived in stilted homes in the past. Only 4–5 elders have a higher floor in their homes, and the rest of the community have ground floor homes, making them more vulnerable to flooding. Some of them added a higher floor to their homes after the 2005 flood. The image below shows some of these drawings.
Disaster Preparedness Planning
All these data, both qualitative and quantitative, help us understand risk, and in turn to communicate risk more effectively. Disaster preparedness planning was one of the outcomes of our process. Projects to assist older persons in planning for disaster preparedness were conducted from the lenses of community development, architecture, communication, art, policy research, and geography. The team thanks the Forum for Older Persons Development and the community of older persons in Nong Hoi for generously sharing their experiences and expertise.
Mind-Mapping Chiang Mai’s Flood Issue
Ultimately the simplest yet biggest lesson learned is that Chiang Mai’s flooding is a complex issue and only understandable when viewed from multiple perspectives. Through interdisciplinary work we had the opportunity to understand this complexity and view issues and solutions from various perspectives. The mind map is a tool to communicate the deconstruction and analyze the various inter-linkages between stakeholders, the environment and their impacts on the city’s flooding.
Participatory Mapping for the Mappers that Map
Usually participatory mapping is done with local communities. Mapmaker David Garcia engaged experts and decision-makers at the field lab to map collaboratively. The map was for participants to draw on, mark, write and trace their experience of Chiang Mai, but it also provided an analog way to bring it back down to basics: what’s on the map? How do things relate spatially? What should be on the map in order to capture more effectively people’s vulnerability to hazards and prepare them better, so as to become more resilient ?