Brooke Lamoureux

Prof Fraiser

Data Science and Human Development

Feburary 13, 2020

Assignment 1: Annotated Bibliography

Problem Statement

Disaster management and relief in low income countries, such as Haiti, is very short of sufficient and can effect the lives of Haitians due to slow emergency responce times, leading to higher mortality rates.

Source 1

Pham, T. T. H., Apparicio, P., Gomez, C., Weber, C., & Mathon, D. (2014). Towards a rapid automatic detection of building damage using remote sensing for disaster management: The 2010 Haiti earthquake. Disaster prevention and management, 23(1), 53-66.

Annotation 1

Haiti, unfortunately, has been victim to many natural disasters in recent history. One of which being a devastating earthquake reaching a magnitude of 7.0 in January of 2010. In a low income country such as Haiti, disaster management falls very short of sufficient. At the time Haiti was the poorest country in the Wesstern Hemisphere. The earthquake had no mercy when it wreaked havoc on the capital, Port-au-Prince, and destroyed about 60% of the city’s government buildings and 80% of its schools. In this specific article, the use of automatic mapping was used to measure damage to infrastructure, and provide rapid assessments of building damage.

In the study, optical images (15cm of spatial resolution) combined with height data (LiDAR, 1m of spatial resolution) measured building damage throughout the capital. This data was then organized into three categories of destruction: intact buildings, collapsed buildings, and debris. Through the use of this automatic object-oriented technique, the amount of damage done to each building is measured immediately and proved to be very reliable. The accuracy of classification varied from 70- 79 per cent. There were various reasons for error including: limited spectral information of the optical images, resolution difference between the two data, high density of buildings but most importantly, certain types of building collapses could not be detected by vertically taken images. The automatic damage mapping strategy discussed in this article could absolutely be used in emergency situations. The entire process took approximately 15 hours, which is much faster than any semi- or fully automatic technique that has been proposed in the past. Semi automatic techniques are based on visual interpretations of images and are very time and labor intensive. Automatic damage mapping could be combined with manual visual interpretation to increase and accelerate response times of humanitarian rescues and resources.

Amartya Sen defined human development as the enhancement of freedoms that allow people to lead lives that they have reason to value. Freedoms such as transparent guarantees. Haiti in particular has a very limited amount of freedom. Haiti’s people are living in great poverty and struggle daily to survive. When disaster strikes, it is only right that their right to safety is protected and transparent. With automatic damage mapping, many lives could be saved and building restoration can be handled much quicker and efficiently. The people of Haiti rely heavily on this method of damage assessment to enhance their freedoms and improve their own capabilities, leading to a positive effect on human development.

Source 2

Yates, D., & Paquette, S. (2011). Emergency knowledge management and social media technologies: A case study of the 2010 Haitian earthquake. International journal of information management, 31(1), 6-13.

Annotation 2

In this research study, the goal was to test the idea that social media is flexible yet robust enough to understand the knowledge structures that align with how knowledge is gathered, shared, and deployed in a disaster response, in particular, the Haiti 2010 earthquake. Social media is responsible for the world’s incredible response and involvement. For example, the earthquake prompted the U.S. Government to rely heavily on social media to gain knowledge and coordinate with the U.S. Agency for International Development, the U.S. State Department, and the U.S. armed forces. If social media is effectively maintained, it can eliminate manually intensive knowledge sharing processes and provide quicker response times to victims.

One social media/ collaborative tool that was used as a response tactic was Microsoft SharePoint. “The platform could be considered ‘social’ as it provided several key functions of social media: it allowed web pages to be created ‘on the fly’ by anyone on the team… Further, all contributions were tagged with the contributor’s name and contact information. Anyone who had a comment or additional information would know exactly who to contact for more information, or how to contextualize their response if they had information to share” (Yates, & Paquette, 2011). Before Microsoft SharePoint, efforts were very linear. There are now mechanisms such as Intellink that develop a shared site that any user with access can access and share knowledge. The SharePoint system was able to cross over boundaries, translate knowledge, and convert that knowledge into understandable information across organizations.

The impact of social media can be viewed through the 3-T framework. The 3-T framework illustrates three different boundaries: syntactic, semantic, and pragmatic, each represents a different milestone of information. Traditional knowledge management systems (KMS) make it difficult to understand these boundaries since it is very manual. Social media has the ability to shortcut these outdated routines and connect with responders more directly and provide access to everyone’s knowledge.

Yates and Paquette are focusing on partnerships for the goals when attempting to enhance human development. Social media has the ability to increase public participation in disaster response, especially in a low income country such as Haiti. Strategies such as Microsoft SharePoint can increase the amount of knowledge known about a disaster and increase the amount of collaboration between responders there is. Traditional response techniques were time and labor intensive. Using the 3-T framework, we can measure the amount of information being communicated across specific boundaries, leading to a positive increase in disaster management and response.

Source 3

Bengtsson, L., Lu, X., Thorson, A., Garfield, R., & Von Schreeb, J. (2011). Improved response to disasters and outbreaks by tracking population movements with mobile phone network data: a post-earthquake geospatial study in Haiti. PLoS medicine, 8(8).

Annotation 3

Currently, there are no accurate methods in effect that track population movements after disasters. The tracking of this information could increase the amount of relief being received by low income countries. In this research study, position data of SIM cards from the largest mobile phone company in Haiti was used to estimate the magnitude of population movements after the Haiti 2010 earthquake. This study also examined the same movements following the cholera outbreak.

Researchers followed the daily positions of SIM cards 48 days before the earthquake and 158 days after the tragedy. The position of the SIM cards were determined because of their proximity to the mobile phone towers they connected to when a call goes through. There were a total of 1.9 million SIM cards being tracked. Those 1.9 million SIM cards excluded inactive ones, and they made a successful call both pre-earthquake, and the last month of the study. Each SIM card represented about 3.2 persons, this estimate was used to measure the moving SIM cards and moving people. An estimated 63,000 persons (197,484 SIM cards) were located in Port-au-Prince on the day of the Earthquake, but left 19 days after. This means that 20% of Port-au-Prince’s population contributed to the net outflow of people, due to the tragedy. The main reason for the research being conducted in this study was to prove that the use of SIM cards to track human movements can be a much faster and valid method of relief distribution. Results from SIM card tracking were compared to results gained from the Haitian National Civil Protection Agency and their post earthquake survey. The HNCPA survey included 2,500 households averaging 4.9 persons. The survey asked questions such as “Did you leave the metropolitan area after January 12 (earthquake day) even if it was for a short time?” and “to what department (province) did you go to?” This survey conducted by the HNCPA is subject to individual interpretation. For comparison, the SIM card tracking method was much more effective and efficient. It produced results within a day.

The use of this technology is changing human development, especially in low income countries. By looking at the way humans react to disaster, relief workers are able to get materials where they are needed to. By studying the movement of SIM cards, we can get the materials to people in distress as soon as possible. We are striving to complete the sustainable cities and communities human development goal and provide even the poorest of cities the opportunity to thrive after a disaster.

Source 4

Heinzelman, J., & Waters, C. (2010). Crowdsourcing crisis information in disaster-affected Haiti. Washington, DC: US Institute of Peace.

Annotation 4

“Triaging crowdsourced reports can help prioritize and direct information” (Heinzelman & Waters, 2010) Basically, social media has proven source after source that it is one of the most effective methods of disaster management and response to date. For example, the international community was able to respond in record time to a 7.0 magnitude earthquake. The extensive search and rescue missions were a reflection of how effective social media is to post disaster reaction time. Traditional disaster-response systems focused on information-sharing among teams of responders all over the world. Though this traditional method worked well for a long period of time, it lacked the ability to prioritize data that flowed in from outside sources. This made it very difficult to benefit from valuable information coming from inside the Haitian community. Ushahidi is an open-source crisis-mapping software that was first developed in Kenya. It provides a vivid way to capture, organize, and share critical information. This information is coming straight from Haiti and is the most detailed and accurate. Where is this information coming from? Social media outlets such as twitter, blogs, and facebook. Information was also gathered from messages sent from mobile phones.

Social media methods, such as Ushahidi, can provide reports of trapped people, medical emergencies, and specific needs. Specific needs that were met after the tragedy were needs such as food, water, and shelter. These needs were received and plotted on maps that got updated in real time by a group of volunteers from around the world. The information gathered was combined with geographic information and was available to anyone with internet connection. Responders on the ground were now connected to all information available and responded accordingly on how, when, and where to supply resources.

Of course, with any new information gathering system, there were some challenges. One challenge being: verifying and triaging large volumes of reports being received. The solution to this problem was to include some manual monitoring and sorting of all the information given. While this may be time consuming, it proved to be very effective and still produce better results and save more lives than traditional disaster-response systems.

In conclusion, the Ushahidi-Haiti project conducted in Haiti demonstrated the potential of the use of crowdsourcing maps and social media for disaster response and management. It provided extensive and useful information to make a foundational model for the world to use and analyze in preparation for future disasters and emergencies.

Source 5

Taskin Kaya, G., Musaoglu, N., & Ersoy, O. K. (2011). Damage assessment of 2010 Haiti earthquake with post-earthquake satellite image by support vector selection and adaptation. Photogrammetric Engineering & Remote Sensing, 77(10), 1025-1035.

Annotation 5

Remote sensing technology is a very powerful tool in post disaster management and response. There are two different approaches used to detect earthquake damage: mono-temporal and multi-temporal. The mono-temporal approach is especially used when attempting to provide an effective emergency extraction of earthquake damage. It does not depend on how available pre-earthquake imagery is. One method of a mono-temporal approach is called support vector selection and adaptation. This method detects damaged regions from a post- earthquake image. The SVSA method uses support vectors from linear support vector machines. There are two stages to this method: selection and adaptation. First, in the selection stage, the linear SVM deployed to get the support vectors from the original training data. The support vector data is then classified with the training data with the K nearest neighbor algorithm created by Cover and Hart in 1967. In the adaptation stage, the remaining reference vector data are adapted to fit other training data and make them ready for classification and the nearest neighbor rule. Adaptation of data is obtained by using the Learning Vector Quantization algorithm.

Another example of an algorithm used to detect damage using a mono- temporal technique would be that of Gamba and Casciati in 1998. Their method only used post- earthquake imagery for damage assessment. Pre- earthquake imagery was only used to analyse building infrastructure and the use of the Geographic Information System (GIS). The data collected from the pre-earthquake images was compared to the post- earthquake images were used to identify areas that had the most damage and needed the most help immediately.

In a multi- temporal approach, damage and change can be detected though the analysis of registered multispectral remote sensing images acquired from the same geographical area and two different times. For example, Rathje deployed a semi-automatic thematic algorithm that identified damage patterns based on pre- and post- earthquake high-resolution images. A disadvantage of a multi-temporal approach using change detection could be something as simple as brightness values or a certain time gap between image timestamps.

Once all damage assessment data is gathered, no matter what method was used, it is time to identify. After the tragedy, the most damage was found to be in the capital, Port-au-Prince. “In order to develop a map of the distribution of damage across Port-au-Prince, the pixels predicted as damage by the SVSA were counted for each part of blocks decomposed, and a tone of gray color was assigned to the corresponding block with respect to its number of damage pixels counted.” (Kaya, Musaoglu, Ersoy) The damage maps obtained by the SVSA and the distribution of damage maps were overlapped by satellite images. Damage assessment is now on a combined map.