People hold photos of relatives killed in Nepal's April earthquake during a memorial rally.Navesh Chitrakar / Reuters

Each day around the world, far more events take place than the international media can possibly report on. It’s largely up to news organizations to decide what is “newsworthy,” and thus what they will—and won’t—tell the public.

As a result, even the most devastating international crises receive only a short burst of attention before they are replaced in the headlines by the next breaking story. This phenomenon is known in communication studies as “compassion fatigue” or “media fatigue.” From the standpoint of disaster response and international aid, this is an especially acute problem for crises that result in large populations being displaced or needing assistance, since media attention quickly fades away even if the needs of the affected population persist for years.

In collaboration with Ben Parker of IRIN News, I recently explored how media fatigue manifested itself in global coverage of the April 25, 2015 earthquake in Nepal. The graph below shows the news articles published worldwide (primarily online, but also including a selection of print and broadcast material), in the 65 languages monitored by the GDELT Project, that mentioned Nepal at least twice in the body of the article from February 19 to July 24, 2015. (The GDELT Project is a nonprofit research initiative, supported by Google Ideas, that tries to understand global society through open information like local news media.) In total, more than 300,000 articles are represented in the graph.


Coverage of Nepal, Before, During, and After the 2015 Earthquake

Total number of worldwide news articles in 65 languages monitored by GDELT that mention Nepal at least twice, from February 19 through July 24, 2015 (Click here to expand the chart.) (Kalev Leetaru)

Overall, the media examined had very little to say about Nepal until the April 25 earthquake, with a surge of coverage on the day of the earthquake, followed by even more coverage on the 26th and 27th. After this initial 72-hour period, coverage immediately began to ramp down exponentially through May 9. In other words, the period of focus on Nepal lasted approximately two weeks.

On May 12, a magnitude-7.3 aftershock, coupled with the first potential major loss of life of first responders (a military helicopter that vanished and was presumed crashed) caused another significant increase in coverage. That renewed attention quickly receded over the subsequent 72 hours, with a small uptick on May 15 corresponding to the discovery of the wreckage of the missing helicopter and confirmation that all aboard had perished.

While still paying more attention to Nepal than it did before the earthquake, the world’s media examined here appears to have largely moved on from the disaster since mid-May. Even the subsequent crash of a Doctors Without Borders helicopter on June 2 does not register on the graph.

Remarkably, this timeline of dissipating attention appears to be consistent across monitored media in different countries. The graph below shows the normalized number of articles per day about Nepal from news outlets published in Australia (blue), China (red), Russia (green), and the United States (purple), representing four major media markets and three different languages. Since the total daily news output of each country is different, the timeline normalizes the raw article counts using what’s called a “z-score,” which essentially reports how far above or below average each day’s Nepal-related article count is for that country. All four countries show nearly identical rates of increased and decreased news attention devoted to Nepal, with the period of elevated attention lasting around two weeks.


Coverage of Nepal in Australia, China, Russia, and the United States

Z-scores of news articles monitored by GDELT in Australia, China, Russia, and the United States that mention Nepal at least twice, from February 19 through July 24, 2015. (Click here to expand the chart.) (Kalev Leetaru)

The graph below applies the same process to the March 13, 2015 landfall of Cyclone Pam on Vanuatu. Unlike the Nepal earthquake, which struck without warning, Cyclone Pam’s landfall was predicted to be highly likely days in advance, which the graph of 28,000 articles below, across the same 65 languages, reflects in a gradual ramp-up of coverage about Vanuatu starting on March 9. Yet, other than this slight early buildup of coverage, the graph looks nearly identical to that seen for the Nepal earthquake.


Coverage of Vanuatu, Before, During, and After the 2015 Cyclone

Total number of worldwide news articles in 65 languages monitored by GDELT that mention Vanuatu at least twice, from February 19 through July 24, 2015 (Click here to expand the chart.) (Kalev Leetaru)

Overlaying the Nepal and Vanuatu graphs—using z-scores again to normalize, given the difference in the total number of articles about Nepal versus those about Vanuatu—the graph below shows changes in the volume of monitored media about both disasters for the 30 days following their onset (April 25 for Nepal and March 13 for Vanuatu). Both disasters produced an initial surge to six times the global average number of articles about each place, followed by a rapid decrease in coverage over the following two weeks, and then that coverage leveling off. (The second peak for Nepal reflects the second earthquake that occurred 17 days after the first.)

The two graphs are nearly identical, indicating a possible pattern that coverage of a natural disaster fades away within two weeks (and has a half-life of just three and a half days). This is the case despite vastly different death tolls for each disaster (more than 8,600 in Nepal compared with just 11 in Vanuatu), suggesting that duration of attention to a disaster is not dependent on the number killed.


Worldwide Coverage of Nepal and Vanuatu Within 30 Days of Disasters

Z-scores of worldwide news articles monitored by GDELT mentioning Nepal (blue) or Vanuatu (red) at least twice in the 30 days after their respective disasters (Click here to expand the chart.) (Kalev Leetaru)

It is important to note that the earthquake in Nepal and cyclone in Vanuatu represent just two of the numerous natural disasters that have occurred in 2015, and that the media examined were drawn primarily from online and selected print and broadcast coverage. Yet, even with these limitations, the similarities in volume of coverage between the two very different disasters and across very different countries suggests that there is only a brief window onto the impact of a natural disaster before the news cycle moves on. Even during this brief window, coverage might not focus on the population most affected by the destruction. IRIN News found that in the first 24 hours, nearly a quarter of coverage about the Nepal earthquake focused on the fate of foreign tourists trapped on Mount Everest.

Such a short news attention span means that policymakers and NGOs have just days, or up to two weeks, to raise awareness about an emerging crisis before it vanishes from international view, which in turn affects their ability to respond or raise funds. On the three-month anniversary of Nepal’s quake in July, with the global media having seemingly moved on, the United Nations issued a reminder that could well apply to many other forgotten disasters: “The emergency is not over yet.”

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.