Cloudy with a Chance of Landslides

Category: Land Cover Change & Disturbances
Project Team: Cloudy with a Chance of Landslides
Team Location: NASA Marshall Space Flight Center at National Space Science Technology Center – Huntsville, Alabama

This map shows areas in Malawi that are susceptible to landslides using the MaxEnt Model. Image Credit: East Africa Disasters II Team

This map shows areas in Malawi that are susceptible to landslides using the MaxEnt Model. Image Credit: East Africa Disasters II Team

Maggi Klug
Max Grossman
Jordan Cissell

Dr. Jeffrey Luvall (NASA at National Space Science Technology Center)
Dr. Robert Griffin (University of Alabama in Huntsville)}
Eric Anderson (University of Alabama in Huntsville)

Past/Other Contributors:
Leigh Sinclair (Center Lead)
Tyler Finley
Padraic Conner
Jeanné le Roux


Malawi has become increasingly prone to landslides in the past few decades. This can be attributed to the terrain, types of soil and vegetation, increased human interference, and heavy flooding after long periods of drought.  In addition to the floods and droughts, landslides cause extra stress to farmlands, exacerbating the current food security crisis in the country. It can be difficult to pinpoint just how many people are affected by landslides in Malawi because landslides often occur in rural areas or are grouped with other disasters, such as floods or earthquakes. This project created a Landslide Susceptibility Map to assess landslide-prone areas in Malawi using variables such as slope, distance to roads, soil type, land cover type, and precipitation. These variables were derived using imagery from Landsat 8 Operational Land Imager (OLI), Shuttle Radar Topography Mission Version 3 (SRTM-v3), Global Precipitation Measurement (GPM), and Tropical Rainfall Measuring Mission (TRMM) satellites. Furthermore, this project created Landslide Exposure Maps to estimate how much of the local population lives in susceptible areas by intersecting population data with the Landslide Susceptibility Maps. Additionally, an assessment of GPM and TRMM precipitation measurements was generated to better understand the reliability of both measurements for landslide monitoring. Finally, this project updated NASA SERVIR’s Global Landslide Catalog (GLC) for Malawi by using WorldView data from Google Earth and Landsat 8 OLI. These end products were used by NASA SERVIR and the Regional Centre for Mapping of Resources for Development (RCMRD) for aiding in disaster management throughout Malawi.

Previous story | Main Page



Daryl Ann Winstead (Mekong River Basin Agriculture) 18-08-2016, 17:21

Wow! Great project! How many landslide points were provided for the Global Landslide Catalog? Thanks in advance for your response!

Jordan Cissell (East Africa Disasters II) 18-08-2016, 17:28

Hello, Daryl Ann, and thank you very much for your kind words! Thank you, too, for your question. Ten landslide points were added to the Global Landslide Catalog as a result of this project. While over 300 landslides were identified and used as inputs to the Princeton University Maximum Entropy Model, not all of these points could be updated to the Global Landslide Catalog, as the Global Landslide Catalog only includes landslides with an occurrence date that can be narrowed down to one month or less.

Kimberly Berry 18-08-2016, 14:30

Great project! What are the drivers of the landslides in the region?

Max Grossman 19-08-2016, 11:57

Thanks for the question Kimberly.

For Malawi the main ‘drivers’ for landslides are the country’s climate, earthquakes, and geology.

Malawi’s climate is such that the country experiences a very intense rainy season and dry season (from November to April and May to October respectively). During that rainy season, consistent heavy rainfall can incite a slide.

Earthquakes act as a potential driver because a quake can break and shift rock underlying a slope, causing a slide.

Geology matters because in certain parts of Malawi, deep soils underlie the earth’s surface. These areas more susceptible to earthquake and rain induced landslide triggers than solid rock given they are loosely and not tightly held together.

If we combine these three factors, slides can be attributed to something called ‘soil liquefaction’. This is the process in which soils are saturated with water (perhaps because of rain) and the shakes from a quake give soil a liquid characteristic. With its momentary liquid characteristic, fast debris flow, landslides, occur.

amber jones 22-08-2016, 14:21

That’s a great question and a very thorough answer.
To build on that, can landslides have greater or lesser intensity? Would gravity play a role and could the GRACE satellites help add that as a variable?

Sean McCartney 12-08-2016, 13:03

Terrific job! I’ve never seen Maxent applied to landslide modeling, only species distribution modeling. 13 million people sounds like a lot of people to be potentially impacted by landslides in Malawi. I hope SERVIR and your local partners are able to make better decision based on this work.

Jordan Cissell (East Africa Disasters II) 18-08-2016, 17:36

Hello and thank you, Sean! We also hope that this project’s finding can help inform ongoing disaster management and prevention practices in Malawi. While landslide susceptibility is certainly not the only factor in a business or government entity’s deliberations regarding where to construct a building, the project’s Landslide Susceptibility will enhance the information available for this decision-making process. Additionally, the Landslide Exposure Map could potentially help aid organizations prioritize the construction of distribution centers or medical stations in areas showing a dense, exposed population.

Vishal Arya 12-08-2016, 12:02

This is a really cool project! At the end, you mention something about near-real time monitoring. Will the team be looking into creating such a tool?

Jordan Cissell (East Africa Disasters II) 18-08-2016, 17:46

Vishal, hello and thank you for your comment and question! The team set out at the beginning of the term to develop a script for near-real time monitoring of landslide susceptibility in Malawi. Unfortunately, while the team was able to identify more than 300 landslides from which model susceptibility, only 10 of these landslides could be narrowed down to an occurrence date of one month or less. Therefore, the scope of the project’s data lacked the temporal precision to effectively model landslide susceptibility in near real-time. However, the team hopes that this project will serve as a baseline investigation, one which future researchers will be able to build off in pursuit of a near real-time model! To this end, the project assessed the accuracy of remotely sensed TRMM-PR and GPM IMERG precipitation estimates against CHIRPS 2.0 ground measurements in order to inform future projects’ usage of these products.

Molly Spater 12-08-2016, 11:49

Wow! Great work! I’m really impressed by your methodology and am interested in learning more about the model your team used from Princeton University and how it can be utilized to predict landscape vulnerability.

Margie dannenbaum 11-08-2016, 19:54

I was wondering why landslides matter in this region of the world? What significance does it have for life in the northeastern United States where I live?

Max Grossman 18-08-2016, 13:54

Great question Margie. Countries like Malawi do not always allocate disaster management resources or have the emergency response agencies to properly response to disasters. So, landslides matter because without institutions or resources, landslide costs can be very burdensome for governments and people affected.

The ‘affected people’ component is particularly true in Malawi – our study found large populations to be exposed to landslides.

A good corollary to Malawi’s situation would be the hypothetical of a post Hurricane Sandy response where Federal Emergency Management Agency (FEMA), state emergency responders, and the Coast Guard were not working at full capacity.

As for the Northeastern United States, landslides really matter in the Appalachian Mountains (places in Pennsylvania, New York, and the New England). Research is more thorough for Southern Appalachian states like NC and WV, however slides do occur throughout the mountain chain.

So to answer why slides are significant to the Northeastern US, the answer is they do, but institutions give assurance that the effects will be dealt with. That is, the potential for landslide burden similar to that which Malawi faces exists here, but federal, state, and local governments fund emergency response institutions that dampen disasters’ effects.

Teresa Fenn 11-08-2016, 16:20

Great project! I know landslides are generally more likely in regions without vegetation. Did you consider correlating landslide events to vegetation cover in the pre-landslide image?

Jordan Cissell (East Africa Disasters II) 18-08-2016, 17:54

Hello and thank you, Teresa! One of the project’s inputs to the Princeton Maximum Entropy Model was a raster layer provided by the Regional Centre for Mapping of Resources for Development depicting landcover types in Malawi, such as grasslands, shrublands, agricultural lands, and forested land. This landcover layer was intersected with Global Forest Change data created by Hansen et. al. at the University of Maryland to consider the role deforestation played in landslide formation. The results of the Maximum Entropy Model showed that landcover was a significant explanatory variable in landslide formation, especially deforested areas, shrublands, and sparse forests.


Leave a Reply