Slide Another Day

https://docs.google.com/presentation/d/1cvhUckV_f5i1Hzc0sxTZqIVtAPZAa0Qp/edit?usp=sharing&ouid=108969284255834283662&rtpof=true&sd=true

https://github.com/hpslavov7/nasaapps2021

General concept

Landslides can result in massive loss of life and enormous economic damages, destroying roads, homes and critical infrastructure. In terms of risk-management we identify two types of measures – long-term engineering and constructive efforts to stabilize an unstable slope or an active landslide and immediate measures to minimize the possible damages. The latter might include for example evacuating people from endangered areas, diverting traffic away from roads, that lie on the path of a potential landslide, and improving the readiness (stocking with provisions, fuel etc.) of both people and institutions in case a certain town might potentially be isolated from the rest of the world for a longer period of time due to damaged roads or might be without power for long. This type of measures however can’t be sustainably kept in place permanently, because that would require a lot of resources. This is why our concept is that the better solution is to use a prediction model, that can serve as an early warning system, providing a reliable risk assessment, in accordance with which the measures should be implemented. This way, even with the uncertainties and limitations of the model, the impact can be more global and can reach even countries and areas, where there’s no resources for a national landslide database and an early-warning system.

Purpose and achievements

The main purpose of our application GeoSpace is to provide an easy to use and understand, inexpensive way to inform the target groups of the local landslide hazard level. The application relies on the massive amounts of relevant data, captured and provided by NASA, but at the same times encourages and consecutively improves its results, when local institutions enrich the databases with their own data, which in some cases is more accurate, because it’s locally measured, by methods that are more precise, but can’t be implemented globally. This achieves four positive results:

• The institutions, especially in regions with low human, technical and financial resources, can benefit from an information source, that provides a real-time analysis, thus improving the decision-making process.
• The general public can be alerted and can take appropriate precautions before the danger becomes imminent. This improves significantly the process, since in many locations around the world usually the information flow starts after the landslide is activated and is focused mostly on damage control.
• The chance for interaction and contribution to app’s data not only improves the quality of the prediction results, but also boosts awareness and involvement and promotes active, preemptive measures to limit the impact of landslides.
• In many countries most of the data, concerning landslides, area characteristics etc., is not in an easy to use format, often stored on hard copies or local machines, where it can’t be quickly and easily accessed by all the necessary experts, when needed. GeoSpace allows local and national institutions to use it not only as a risk-assessment tool, but also as a database.

The concept of GeoSpace

In our application we’ve chosen to create a landslide susceptibility map, based on different factors, specific to each location, and to combine them with two dynamic factors – rainfall and earthquakes, that are triggering events for landslides, with the risk, naturally, being greater in areas with high susceptibility. For the purposes of the hackathon we’ve limited the scope of our application to a database of existing landslides, since even partially contained/stabilized landslides are prone to landslide re-activation due to both static and dynamic factors, contributing to susceptibility.

The visitors see a very neat, user-friendly UI, where they can view the map with pin-pointed landslide locations.

Fig. 1 Home page of GeoSpace

We’ve chosen the design to be simple, intuitive and without unnecessary details, so everyone can easily take advantage of it. By choosing to browse they can select a specific landslide location and will receive the exact coordinates and level of risk, calculated for it.

Fig. 2 Map with pin-pointed landslides and data for a specific landslide

From the start page they can also choose if they want to contribute landslide data, based on whether they are individuals or government officials (Simple and Government Report), and can upload precipitation data. This is a very important aspect of our model, since ground-based data, collected by means of classical measurement methods, is very accurate and complements perfectly the satellite data from NASA, which on the other hand allows to have global coverage.

Fig. 3 Tab Report Landslide

Our concept involves on the next stage of development to send warning notifications to registered users nearby a landslide, when the dynamic factors raise the level of risk.

How we built our model

An important part of this process is to choose the factors, that are being used for the risk assessment, and the method for determining the relative importance of these factors. As a base for our methodology we’ve used a paper, that best reflected our approach and the available time and data we had – Assessment of landslide susceptibility at a local spatial scale applying the multi-criteria analysis and GIS: a case study from Slovakia. The technique for determining the relative importance is the Analytic Hierarchy Process (AHP) for organizing and analyzing complex decisions, developed by Thomas L. Saaty. We relied on their judgment, based on previous studies, for assigning rating and relative importance to each factor. The data, that we found available in the limited time we had, that has a global coverage, doesn’t include all the necessary categories, so the plan for developing the application on later stages is to i) obtain more global data sources and then ii) to create two matrices – one, where the number of factors has been lowered and the matrix is re-calculated, and a secondary one, which is to be deployed when local authorities provide additional data. The seven contributing factors for assessing landslide susceptibility, which were confirmed in various papers, with their respective relative importance are slope angle, geology (permeability level), slope aspect, elevation, distance from rivers, distance from faults and land use. The susceptibility of the area in question is divided accordingly in one of five stages, ranging from very low to very high.

The second part of our risk assessment is to take into account the impact of two landslide triggering events – earthquakes and intensive rainfall. Due to the limited time and the fact that at this stage we are analyzing areas with existing landslide, i.e. areas, that are very prone to being influenced by such events, we chose a simpler approach to include them – considering them as thresholds. We’ve used three dynamic factors – distance from epicenter and local magnitude (on the Richter scale) of the earthquake, and rainfall.

The algorithm postulates that the first check is to see if the area in question is in the 10 km radius from the epicenter (as it is established statistically as the zone, most affected from an earthquake, in Keefer, 2000). If so, the second threshold check is if the local magnitude of the earthquake is above 4 (as postulated by Keefer as the minimum magnitude, that can lead to landslides). If both conditions are fulfilled, the application raises the risk to very high, suggesting that there’s need for additional local assessment of the situation and/or preventative measures. For the precipitation rate we use near-real time data from NASA’s EarthData and this is another opportunity for local authorities to improve the results by providing real-time, accurate local data for the precipitation rate and duration. These are then used in the following expressions, as formulated by Hong et al.:

Here I is the intensity in millimeters per hour and D is the rainfall duration in hours. Choosing the expression based on the whether the 24-hour-threshold is exceeded or not, our application calculates if the threshold for triggering a landslide is reached or not, and then updates the risk accordingly to very high.

GeoSpace offers two separate options for the general public and for official bodies to contribute and improve the data, used in the analysis. For individuals there’s a simple report form, where they can report active landslides, that they’ve encountered. For government officials there’s an advanced report, which allows them to expand both the landslide repository and the separate contributing factors. The option to browse landslides allows the user to see the data, that is of interest to them.

Used programming languages:

Backend - C#

DataBase - Azure SQL-server

Front-end - Vanilla JavaScript, HTML ,CSS

3-rd party libraries - jQuery, jQuery UI

https://earthdata.nasa.gov/

We chose this challenge in light of the increasing number of catastrophic events, which cause a lot of loss of life and damage even in well-prepared countries. We decided that an application like ours could directly impact a vast amount of people all over the globe, giving them a better fighting chance against the consequences of climate change, regardless of economical capabilities.

Our experience was very positive, even though (or in some sense because) we encountered many obstacles to what seemed naively like a very clear-cut process. Since we didn’t have the time and the specific expertise for a traditional approach in determining the relative importance of the factors, we were forced to look and found the AHP methodology, which not only solve this issue for us, but also gave us the chance to be more flexible with the number and type of contributing factors we use, depending on what is available for a specific are. Another issue was how to work with the HDF5 files, available in EARTHDATA, which took a great deal of time. We found a way to properly use the files for the precipitation, but decided to use mock data for the other factors, so that we can focus on the other aspects of our project.

We had the chance to get out of our comfort zone and to dive into a field, where none of us had major knowledge, so it turned into a fun weekend.

https://earthdata.nasa.gov/

Vojtekova J., Vojtek M., 2020, Assessment of landslide susceptibility at a local spatial scale applying the multi-criteria analysis and GIS: a case study from Slovakia https://www.tandfonline.com/doi/full/10.1080/19475705.2020.1713233

Keefer D., 2000, Statistical analysis of an earthquake-induced landslide

distribution — the 1989 Loma Prieta, California event https://homeweb.csulb.edu/~rodrigue/quake/keefer2000.pdf

Keefer D., 1984, Landslides caused by earthquakes https://pubs.geoscienceworld.org/gsa/gsabulletin/article/95/4/406/202914/landslides-caused-by-earthquakes

Hong Y, Adler R., Huffman G., 2006, Evaluation of the potential of NASA multi-satellite precipitation analysis in global landslide hazard assessment - https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2006GL028010

#landslides #EarthData #riskassessment #webapplication

This project has been submitted for consideration during the Judging process.