top of page

Tracking Public Response and Relief Following the 2015 Nepal Earthquake

Paper Accepted by IEEE International Workshop on

Collaborative Internet Computing for Disaster Management, 2016 (link)

The massive Nepal earthquake that occurred on April 25th, 2015, has caused severe damage to Nepalese villages and housing construction, while incurring a huge human tragedy.

MOTIVATION

There was significant support from the global community as well as immediate assistance and aid from both international and local entities after the 2015 Nepal Earthquake. However, recovery has been very slow, reconstructions are still sorely incomplete.
There are many datasets available from a range of sources that have recorded various aspects of this disaster and its fallout since after the earthquake. In this project, we present a snapshot of the post-earthquake response and relief activities from various datasets.

DATASET

GDELT (Global Database of Events, Languages and Tones)

  • The goal of the GDELT project is to monitor and share global media news about events around the world to the public.

  • The dataset is machine-coded by the Textual Analysis By Augmented Replacement Instructions (TABARI) system and receives daily updates from thousands of news articles.

  • It collects and stores the type, people, countries and other 57 features of the events. Especially, it tries to capture the prevailing tone in the news reports.

  • The GDELT event database includes 250 million entries with each entry capturing two actors and the action performed by one actor to the other.

  • The attributes of the two actors (i.e. Name, Country, Type, etc.) and the category of actions (i.e. Provide aids, Appeal, etc.) enable us to analyze the interactions among the international and domestic actors.

  • For event, GDELT uses the Conflict and Mediation Event Observations (CAMEO) scheme to code. It is a hierarchical set of event descriptions. Each code represents for a specific category of event.
     

OPEN NEPAL

  • Financial Aid data for earthquake (by country) .csv

  • List of organizations and their activities (Supported by different donor agencies) .csv

  • Citizen Feedback in Earthquake Affected Areas .csv

HDX (Humanitarian Data Exchange)

Nepal - Who's Doing What Where (Housing Recovery and Reconstruction Platform) .csv

  • Where: district (14 affected areas)

  • Who: organization (Implementing / Funding)

  • What: activity type (skill development-Carpenter / Mobile Technical Support)

  • When: activity start and end date

ANALYSIS

PUBLIC RESPONSE in NEWS MEDIA

News Coverage

​The plot visualize the relationship between news volume and time period.

  • Media and organizations reacted very quickly to the Nepal earthquakes.
  • There was a sudden surge especially on April 27th.
  • The news coverage then quickly faded over the next 72 hours.
  • It suggests that people’s attention increased sharply right after the disaster but after three days, it faded away.

Event Types of Daily Actions

This figure shows the daily proportions of event types indicated by colors.

 

  • The types of events include “appeal,” “express intent to cooperate,” “engage in material cooperation,” “yield,” “fight,” etc.

  • Three of the 19 event types dominate in the bar chart: dark green, red, and light blue, corresponding to event codes 1, 4 and 7, respectively.

  • In GDELT dataset. The three types of events are further described below.

1 make public statement

2 appeal

3 express intent to cooperate

4 consult

5 engage in diplomatic cooperation 

6 engage in material cooperation 

7 provide aid

INTERNATIONAL and DOMESTIC SUPPORT

Daily Records

of International and Domestic Aid

based on the GDELT Dataset

The plot shows the change of both international and national aid in quantity over time for each category.

 

  • The biggest spike is on April 27.

  • For the military aid (72), the spike appears on May 4, which may be because of issues reported about airport management.

  • Most of the aid was provided by the international community.

  • A small spike on May 12 corresponds to the second significant earthquake of magnitude 7.3 that occurred on May 12.

We denote by non npl-npl the international aid, and by npl-npl the aid from/within Nepal.

GDELT dataset shows five types of aid:

70 general aid

71 economic aid

72 military aid

73 humanitarian aid

74 from military and peacekeeping groups

75 Grant asylum

Comparison of Donation Records

from GDELT and Open Nepal

We compare the GDELT dataset with the transaction dataset in terms of the economic aid to Nepal.

 

  • In the GDELT dataset, economic aid code is 71; we extracted these records and then grouped the records by country.

  • Here we select eight main countries. We calculated each country’s proportion of contribution.

  • The figure shows that the GDELT dataset is consistent with the transaction dataset.

  • The top donor countries also dominated the news media attention.

  • The data indicate that countries like China and Canada appeared more in news reporting but their donations were comparatively smaller.

Monthly Changes of

Targeted Sectors for Donations

We plot the 8 active sectors that received more funds in the period from April 25 to December 28, 2015.

 

  • More than 0.5 billion US dollars was collected for building shelters and non-food items.

  • Based on survey for victims in affected areas, long and short term shelters have been indicated as one of the biggest problems or concerns.

  • The Nepalese government put a significant effort in minimizing disruption in education

Donation Progress for Different Organizations

We show the 10 most active donation organizations, and plot donation amounts in pledged, disbursement and commitment for different international actors in the plot above.

 

  • Some donors have not completed their pledges.

  • Some donors seem to have delayed to donate due to corruption and political instability issues

  • These issues may cause problems for further reconstruction.

RELATIONSHIP between PEOPLE’S ATTITUDES and RECONSTRUCTION PROGRESS

We combine result from three different datasets to analyze reconstruction efforts and citizens’ feedback.

(a) Damage degree

We plot how many houses got damaged by districts to indicate the damage degree.

(b) Reconstruction stage

We count all the finished projects about housing recovery and reconstruction after the quake, grouped by districts.

(c) Satisfaction degree 

We analyze the questionnaire of the survey conducted in 14 affected districts. Based on the responses to the question “Is support provided in a fair way?”

(d) Urgent degree of housing issues in citizens’ minds

Based on the responses to the question “What is your biggest problems?”, we plot (d) that counts number of answers indicating long-term or short-term shelters.

  • Some districts got severely destroyed, and people living there needed more help with households than expected.

  • However, they somehow received less attention.

  • Support did not seem to have been provided in a fair way. This may reflect the deficiency of post-disaster reconstruction management.

  • Media rapidly increased their attention on disaster especially within the first 72 hours.

  • Their coverage quickly declined, while the international and domestic donations/support kept coming.

  • Nepal received a lot of support.

  • Some donations appear only as a pledge instead of disbursement. The survey responses show that the government may not have made adequate efforts for reconstruction activities.

  • The data suggest that the government did not distribute support in an equitable way. Citizens’ disappointment in specific earthquake affected districts can be seen in the survey data.

Found patterns of media response associated with significant events
Found patterns of national disaster management
Connected exceptions with news reports
Verified data consistency by comparing datasets from different sources with different aspects
Discovered key issues and proposed hypothesis based on datasets
Conducted research to prove hypothesis
Combined three datasets to achieve more insights about reconstruction

ACHIEVEMENT FLOW

CONCLUSION

bottom of page