Automated Detection of Hazards
H.O.P.E. Human observation for predictions of the environment
Summary
H.O.P.E. is a project that seeks to detect, predict and report on disasters based on different data sets. It is proposed that the primary variables that describe each disaster be identified and characterized by selection algorithms. Subsequently, a learning algorithm will be implemented that can help the automatic classification of hazards. The result of this process will allow scientists to study such events in detail and will also help decision makers to act early. It is known that it is of great importance to have truthful and complete information about possible disasters, to do this, the creation of a mobile app and a dissemination plan through social networks.
How We Addressed This Challenge
Problem Statement
The main challenge of this project is to be able to process the vast databases compiled by the satellites and stations of our planet, so that all the climatic changes that the earth has suffered over the years can be visualized and compiled. It is intended to characterize each natural phenomenon that may be available in the databases, and then proceed with the automatic extraction of these.
Despite the fact that in many phenomena different variables are correlated, it is thought that each phenomenon may have a particular fingerprint that allows it to be characterized with some precision. This information would allow to implement a learning algorithm, which can later be applied to carry out the extraction process automatically.
Tools such as Nasa World Viewer or AMSR Earth Environment Viewer already allow the visualization of different phenomena, but what this project intends is to compile those phenomena that stand out for the severity and impact they have caused on the population, although for this it is necessary to find the more significant in the vast databases, making an automatic classification.
Justification
Natural phenomena as well as disasters can have a significant impact on our ecosystems, economies, and security. Climate change and increasing exposure in risk areas are expected to increase the frequency of threats from meteorological events and exacerbate their impacts on public safety. To counter these effects, the global community aims to improve disaster preparedness, response and recovery. This requires, among other things, reliable and rapid means of information on disasters and their impacts.
Observations of the Earth are often used to help identify its location, extent, duration, and potential impacts. However, for researchers to carry out detailed studies and create mitigation strategies to reduce these impacts, they must be able to detect natural phenomena quickly and study them in depth. NASA's data sets are freely accessible and provide observations of the Earth that are often used to monitor these hazards and their impacts. Models based on machine learning can help researchers quickly map hazardous weather situations and assess their intensity and the extent of their impacts.
While AI methods are already important for natural hazard risk analysis, both in practice and in research, further work is needed to improve AI techniques and their implementation in risk analysis. NASA produces large volumes of satellite data that are used to detect and track natural phenomena. However, many of the phenomena are not automatically detected or tracked within large databases. This leaves experts tasked with manually searching through petabytes of data to analyze phenomena.
Assessing risk is only part of the solution, but acting on such assessments is another. Citizens may not always respond to risk warnings as authorities expect; They do this not because they act irrationally, but because they feel severely restricted in terms of the options offered to them. Understanding how people interpret risks and choose actions based on their interpretations is vital to any disaster reduction strategy. Knowing the risk and knowing how to be able to respond to the risk are not the same. Some people choose not to prepare and others may be interested, but need more guidance. Attention should be paid to how people take interpretations of risks based on their own experience, feelings, personal values, cultural beliefs, interpersonal and social dynamics. Using communication media such as social networks to share information about risk warnings can be a powerful tool to present the information, taking advantage of the reach of these means that are used on a daily basis.
Objectives
Overall objective
Generate an automated hazard detection application based on machine learning that shows and predicts natural phenomena that have had a high impact and reports on their main consequences on the population.
Specific objectives:
- Characterize the natural phenomena that have impacted the population.
- Use a classification or learning algorithm that allows analysis of large data sets and automatically classifies some phenomenon.
- Identify the scope and main consequences of the phenomenon to be studied.
- Develop a mobile application that alerts and provides information on natural phenomena.
- Implement an outreach plan through social networks that reaches the general public.
How We Developed This Project
SPECIFIC METHODOLOGY APPLIED TO HEATSTROKES
INTRODUCTION AND PROJECT JUSTIFICATION
In March 2020, the World Meteorological Organization published the Declaration of the State of the World Climate 2019, which it shares proven and accurate information for decision makers regarding the need to adopt preventions against climate change.
In this Declaration, it is mentioned that extreme heat conditions are taking an enhancement in losses in human health. Even, greater consequences are registered in places where extreme heat occurs in the context of an aging population, urbanization and inequalities in access to health.
In 2018, vulnerable people around the age of 65, hit the record of 220 million plus exposure to heat waves, leaving the previous record set in 2015 for 11 million.
Last year, the high temperatures were a record in Australia, India, Japan and Europe, generating adversities in health. In the same way, the biggest heat wave hit Japan in late July and early August 2019, causing more than 100 deaths and a heavy burden on the health system with 18,000 additional hospital admissions. Meanwhile, Europe experienced two major heat waves in the summer of 2019; in the Netherlands, this event was associated with 2,964 deaths, almost 400 more deaths than during an average summer week. In England, there were 572 deaths more than those observed above the baseline for all causes of mortality in people over 65 years of age.
On the other hand, in the United States, a study made by NOAA (National Oceanic and Atmospheric Administration) affirms that the highest rates of premature babies, that is, when a child is born before completing 37 weeks of pregnancy, are located in the southeast of the country, relating the extreme heat as a contributing factor to these types of births.
These effects can be accentuated by the heat island, understood as the phenomenon that occurs when a city experiences much warmer temperatures than nearby rural areas and is related to human activity.
Currently, the coronavirus is a disease that has been the main global concern, however, it is temporary and with temporary impacts, however, anthropogenic global warming has been growing for decades and requires priority actions.
In addition, there are many countries committed to the goals of the 2030 Agenda for Sustainable Development, among themes, there is Goal 13 project, referring to Climate Action, with specific goals such as strengthening resilience and capacity to adapt to climate-related risks and natural disasters in all countries; also to improve education, awareness-raising and human and institutional capacity regarding climate change mitigation, adaptation to it, reduction of its effects and early warning.
For all the above, it is important to have a system that detects thermal sensations that can be dangerous for people, directly impacting their health through heat stroke, especially the most vulnerable such as pregnant women, children and the elderly.
PROBLEM STATEMENT
Dealing with this, H.O.P.E has decided to confront this problem for scientists, politicians -decision makers- and the whole population via a divulgative program.
SCIENTIFIC SIDE
Particular objectives:
- Investigate the primary variables that describe heat waves.
- Search available NASA data for such selected variables.
- Calculate a standard heat index.
- Identify the most significant heat waves.
- Implement a recognition algorithm on the heat waves found.
- Search for heat waves automatically using the previously implemented recognition algorithm.
POLITICIANS AND DECISION MAKERS SIDE
- Create an interactive map to show the most important heat waves in history.
- Make a bibliographic review of the damage caused by such heat waves and report it.
DIVULGATIVE SIDE
- Give a protocol of recommendations on what to do before a heat wave.
- Show infographics on social media.
- Develop an app to have specific information about hazards and give alerts.
Website:
Social networks:
- INSTAGRAM: https://www.instagram.com/hope.nasa.space.app/
- FACEBOOK: https://www.facebook.com/HOPEM%C3%A9xico-113409793860368/
- TWITTER: https://twitter.com/HOPEMXICO1
How We Used Space Agency Data in This Project
NASA World View:
- Surface Air Temperature
- Surface Relativity Humidity
- Land Surface Temperature
Data from this sources was employed as images in grayscale. Such images were processed by a pixel by pixel comparative analysis. This allowed to find patterns in both sensors, which is required for heat wave detection
Data & Resources
- Tapete, D. (2020). Key Topics and Future Perspectives in Natural Hazards Research.
- Guikema, S. (2020). Artificial intelligence for natural hazards risk analysis: Potential, challenges, and research needs. Risk Analysis.
- https://2020.spaceappschallenge.org/challenges/inform/automated-detection-hazards/details
- Guikema, S. (2020). Artificial intelligence for natural hazards risk analysis: Potential, challenges, and research needs. Risk Analysis.
- Eiser, J. R., Bostrom, A., Burton, I., Johnston, D. M., McClure, J., Paton, D., ... & White, M. P. (2012). Risk interpretation and action: A conceptual framework for responses to natural hazards. International Journal of Disaster Risk Reduction, 1, 5-16.
- de Bruijn, J. A. (2020). Natural Hazards in a Digital World Algorithms for Using Social Media in Disaster Management.