NRT Flood Impact Analysis on Road Networks
A case study in the Mandalay region, Myanmar
FOR DEMONSTRATION PURPOSES ONLY
Introduction
The world is changing due to climate change, affecting uncertainty regarding flood risk. For emergency response management, this means there is a strong need to quickly analyze the impact of floods on communities and their infrastructural networks. Here, we present the results of a test-case that uses a combination of Deltares tools to analyze the near-real-time impact of floods for emergency response.
The core of this project is a combination of the HYDrologic Remote sensing Analysis for Floods (HYDRAFloods) tool , a near-real-time flood mapping and monitoring system, and the Deltares Criticality Tool , which we used to quantify the cascading impacts of hazards on infrastructure networks by analyzing the accessibility of relevant locations. Development of this method was financed by Deltares challenge "Resilient Infrastructure meets Enabling Technologies".
It's possible to apply this method to any infrastructure type. Think about transportation infrastructure (roads/rails), but also other types of infrastructure networks such as electricity, sewage, gas, or the connectivity to areas that are vital for the functioning of a country, for example, airports, ports and industrial zones with water treatment.
This method consists of two assessments:
Step 1: Network analysis - this assessment tells you your infrastructure network redundancy and indicates hotspots where congestion might occur.
Step 2: Impact assessment during flooding - this assessment analyses the impact of a flood event on the road network, like which roads fail and what health care facilities become inaccessible.
Heavy monsoon rains in Myanmar
In this case study, we examine the Mandalay region in central Myanmar. It was severely affected by flooding after heavy monsoon rains during July and August 2019. Every year many regions in the world are flooded due to monsoon rains, resulting in large scale evacuations and limited access to health care. During these events, the main transportation network is one of the most crucial parts of emergency response, as it is used for 1) the delivery of goods, 2) evacuation and 3) deployment of emergency hospitals.
On the left of the slider is the regular extent of the Ayeyarwady river around Mandalay and on the right the flood extent during the 2019 floods based on satellite imagery (derived from Copernicus Sentinel data 2019, processed by ESA). The colours indicate the number of weeks flooded, varying from yellow (1 week) to dark purple (5 weeks).
Method: On-the-fly assessment of infrastucture
Making use of the Deltares Criticality Tool, we were able to identify the impact of the floods on road network connectivity on-the-fly. To enable the delivery of goods, evacuation, and the deployment of emergency hospitals, crisis managers should know which places are still accessible by land transport for emergency relief. The method used can provide this information. The tool is scalable and applicable at any location in the world. Besides assessing near-real-time, results also can be used for longer-term studies, such as the estimation of direct damages and/or long-term disruption of the roads due to flooding.
Mandalay case
As a test case for the coupling of the two tools, we focused on the impact of floods on the community in the Mandalay region, by assessing road infrastructure criticality and connectivity between population centres and hospitals.
The population centres on the map are represented by the orange bubbles - the larger, the more people that live in that area [ data source ] The hospitals are indicated with the white heart in the red circle [ data source ].
Area of interest
The scope of the case study is decided by three bounding boxes surrounding the flooded area.
- Population: 9,847,024 people live in the area (orange-dotted extent)
- Hospitals: for every population centre, the nearest health centre is identified (red-dotted extent)
- Roads: the bounding box of the road network is created ~100 km from the centre of the population extent. This extent is the largest, to enable alternative routes outside the flooded areas for all population centres to all hospitals. All roads within this black-dotted extent are used for the analysis.
Goal
The goal of this study was two fold:
- Make use of open-source data (Open Street Map, free satellite-imagery) to calculate the impact of floods near-real-time based on network criticality.
- Analyse which population centres (black dots) are disconnected from or suffer from delays to reach health care (white heart in red dots), resulting from the floods.
To be able to find the routes between population centres and health care practices, those locations are projected to the nearest point on the road. Those are shown in the map to the left as black dots for population centres and as white hearts in red dots for health care practices. Results of this study could be used to identify locations where emergency response is most needed.
Step 1: Network analysis
The critical points in the road network were identified with a redundancy and road usage intensity analysis. A redundancy-based analysis shows the effect on the redundancy of the system when a single link of the network is disrupted at a time. For each disrupted link, a redundancy analysis identifies the best existing alternative route or, if there is no redundancy, the lack of alternative routes. This is performed sequentially, for each link of the network.
Network redundancy
The redundancy analysis identified the most critical links of the road network: those that experience the longest detour distance when disrupted. However, to calculate the impact, redundancy is used in combination with road usage (next map). For example, the links with the longest detour distance when disrupted are not necessarily the most used links and thus could have a smaller effect on society.
The centre of the road network showed the highest redundancy due to the high density of the network. Along the outer edges of our study area, redundancy decreased due to fewer detour options.
Road usage intensity
As we are interested in the accessibility of health care, we also identify the road segments that might get extraordinarily busy during the flood event. Therefore, we assess the relative road usage intensity without and with flood event. This relative intensity is determined based on the number of times a road segment is on the route from all population centres (black dot) to its three nearest health centres (white hearts in red dots). The colours indicate the relative road usage intensity, ranging from light yellow (low) to dark red (high).
Duration of disruption
Many roads were flooded for multiple days during the event. We estimated the duration of the disruption by counting the number of weeks the road segments were flooded, based on the satellite imagery. The roads can be non-disrupted (light) up five weeks disrupted (dark).
Step 2: Impact assessment
For every week of flooding, we estimated the number of people that were affected. We distinguished between those by being disconnected from their three closest hospitals and those who experience delays. This means when the shortest road is disrupted, and a detour is needed.
Road disruption due to flooding
This image shows the flood extent (based on the HydraFloods classification) during the first week of August 2019 as well as disrupted roads. Roads were assumed to be disrupted if they were partially covered by water on the flood map.
Disrupted roads are indicated with grey dotted lines. Non-disrupted roads are shown in green, disrupted roads where a detour was available from one to the other end of the road segments are orange.
Due to the higher elevation of Mandalay, many roads were still accessible, while surrounding areas are partly flooded.
Impact on communities
Based on the previous analysis it is possible to identify the number of people in population centres that do not have access to health care (red bubbles) and the number of people that could still reach health care, albeit with a detour (yellow bubbles).
In the first week of August, the most intense floods occurred. From the analysis, it was estimated that over a million people were not able to reach any health care and, 200.000 people were suffering from delays to reach health care.
Number of people affected by road disruption due to flooding.
The extra distance is calculated per route from population centre to its three closest health care location, that is impassable and requires people to take a longer route to reach this location. Over all those extra distances, the average is shown here.
Road usage intensity during flood
Finally, we could recalculate the road usage intensity of the non-flooded segments. From this, the most critical roads during these floods and areas most sensitive to congestion are identified and can be used by first responders for evacuation planning.
Dark red color scale indicates roads with an increased chance of congestion during floods.
Conclusion
This project combines two novel tools to demonstrate Deltares' ability to detect near-real-time floods and assess their impact on critical infrastructure. With this Proof-Of-Concept, we show the possibilities of earth observation data for critical infrastructure assessment. By utilizing global earth observation products and open global data, the tool is easily scalable to anywhere in the world.
Emergency managers have a strong need for tools that quickly analyze the impact of hazards on infrastructure networks. Like during flood events, medical care has to be distributed and emergency response has to be allocated to most severely impacted areas. With our tool, we demonstrate the possibilities of near-real-time assessment of critical infrastructure and provide a unique method for forecast-based emergency management.
Disclaimer
This project uses publicly available data; amongst others infrastructure networks from Open Street Map and satellite images from ESA. We did not validate this data during the study, and any inaccuracies due to overestimated numbers, inconsistencies in the data or missing values where disregarded. Flood detection algorithms are used on raw satellite products, which might contain inaccuracies due to satellite scattering mechanisms. The influence of scattering mechanism on flood map accuracy has not been taken into account.
