A man walks past a dead cow in Farado Kebele, a region of Ethiopia stricken during the country's drought and famine in 2016.Tiksa Negeri / Reuters

It’s easy to talk about how climate change will alter Earth’s surface in the century to come. It will raise sea levels, flood cities, and set off droughts. As this month’s dire United Nations report shows, decades of climate science have made the worldwide dangers of human-caused warming unambiguous.

Yet it’s far harder to talk about how these changes will play out locally. No two places will experience climate change identically: The coasts of Borneo and the shores of Great Britain, for example, will see the land and weather transform in vastly different ways. But people living in the United Kingdom have a far better idea of what those changes will look like.

Call it climate science’s data gap. When studying the Earth’s climate, researchers must understand the past before they can understand the future. But across huge swaths of the world, scientists simply don’t have the data they need—especially the kind of in-depth, long-term observations that can place current weather in context—to understand that past.

These problems are particularly acute for countries in the global South. In 2015, the UN’s Intergovernmental Panel on Climate Change (IPCC) projected how climate change would affect temperature and precipitation values on each continent. In Africa, Asia, and South America, it rated more than 40 percent of its own predictions to be “low confidence.”

In Europe and Australia, by contrast, only about 12 percent of its projections were of low confidence.

“There’s essentially an order-of-magnitude difference in the amount of science available across regions of the world,” says Katharine Mach, a senior climate researcher at Stanford University and one of the authors of the IPCC report. She pointed out that the IPCC spent as much time and energy writing a chapter on Australia and New Zealand as it did for the entirety of Asia.

“If you were to take the 8,000 studies relevant to climate change in Asia, and the 3,000 for [Australia and New Zealand], and if you were to do publications per person [living in those places], obviously the number would look starkly different,” she told me.

The most recent IPCC report, released earlier this month, only focused on how the Earth overall would be affected by 1.5 degrees Celsius of temperature rise and did not produce continent-specific chapters.

At least 60 years’ worth of data might be needed to draw meaningful observations about temperature or precipitation trends, says Edvin Aldrian, a climate scientist for the Indonesian government who is currently on the IPCC. For parts of his home country, Aldrian says data is scattered at best. In some cases, it might not exist.

Aldrian argues that the IPCC’s methods required a good base of historical data about a country. Without a solid grasp of the baseline science, it becomes difficult to propose effective strategies for communities facing the consequences of climate change, like crop failures due to shifting rain patterns.

It also makes it harder to understand what’s happening now. In much of the global North, researchers are already documenting changes that seem influenced by climate change. Most regions in North America and Europe have experienced a documented increase in hot days and a decrease in cooler days since the 1950s, for example. Based on that data, climate models can predict with high confidence that countries in northern Europe will “very likely” see an increase in hot days in the future, along with more intense precipitation in the winter months, and perhaps won’t experience major changes in drought conditions.

But across broad swaths of Africa, Asia, and South America, that kind of accounting isn’t possible. In East Africa, for instance, there isn’t even enough published scientific literature to say whether temperatures have shifted since the 1950s.

The same goes for precipitation changes. When the IPCC mapped how rainfall amounts had changed from 1951 to 2010, North America and Europe were clearly shaded in a color gradient showing increases or decreases in rainfall. But almost all of central Africa, as well as patches of Asia and South America, had to be left blank due to insufficient data:

Can’t Ask a Weatherman Which Way the Wind Blows

This map from the 2014 IPCC report shows how annual precipitation amounts have changed since 1951. There was not enough data to establish a meaningful trend for whole regions of Asia, Africa, and South America. (United Nations / 2014 IPCC Technical Report)

There’s a reason for this data gap: The majority of the studies analyzed or evaluated by the IPCC concerned North America and Europe. The report summarized more than 10,000 studies about Europe, but only 2,982 about Africa.

This effect isn’t just dependent on southernness, but on historical affluence. Though Australia is roughly half the size of South America, the former contributed about 1,000 more studies to the IPCC report.

“If you’ve got 10,000 available studies on the issues versus 3,000, you’re actually able to say much more rigorously and specifically that tons of impacts are happening in Europe, whereas the evidence is a lot scarcer in [other parts of the world],” Mach told me.

The IPCC’s models do extrapolate and fill in all areas of the globe with predictions for changes in precipitation and temperature, she added. But those conclusions might not be as accurate or nuanced for regions with less historical data to begin with.

This effect is particularly troublesome when dealing with a complex ecosystem like the Sahara, which expands and contracts seasonally due to natural variations in ocean currents. It can become difficult, if not impossible, to understand that unique region by tracing back rainfall and weather records more than a few decades, and it can pose a problem to isolating short-term patterns from long-term climate trends.

The catalog of places that suffer worst from the climate data gap is not entirely arbitrary. Despite sharing a history of colonization by Europeans, for example, the Indian subcontinent tends to have much better records than Africa.

“It mattered to [the British empire] that they had good data” from India, says Amir Jina, an environmental and developmental economist at the University of Chicago. “Because agricultural products were such a big component of the income from [the colony], weather data was very important. So the British put in place this data-collection infrastructure in their colonies.”

Of course, this did not make the British any better stewards of the continent. Colonial officials may have implemented the scientific apparatus to collect weather data and crop records, but they still sought environmental and social control over India. Their bounty of data did not stop them from acting to the deliberate detriment of locals, exporting millions of pounds of rice out of the colony during periods of famine.

“A lot of places in sub-Saharan Africa, it’s the opposite story,” Jina told me. “[Some of the continent] was colonized during the scramble for Africa—and so it mattered that colonial officials knew where gold was or the mines were to extract it,” but not so much that they monitored fluctuations in the weather.

Collecting good data is also expensive, particularly if the country in question lacks financial and institutional resources, Jina added.

In some cases, the historical gaps in climate data and the infrastructure to collect it can be made up for by technological advances and international scientific projects. Satellites, for instance, can produce high-resolution images and temperature and precipitation records for most places on Earth.

But just because this data exists, climate scientists don’t automatically have access to it. Take the situation in the world-famous, ecologically diverse, and climatically vital Himalayas.

In 2012, Kamal Bawa, a conservation biologist at the University of Massachusetts at Boston, used satellite imagery to track changes in temperature, rainfall, and vegetation patterns in the Himalayas. The study was one of the first large-scale climatic analyses for the region, and it states that the Himalayas have already experienced at least 1.5 degrees Celsius of warming on average over the past 25 years.

Bawa told me that it can be difficult for researchers to navigate the politics of the mountains, which straddle Pakistan, India, Nepal, and China. There’s also a lack of funding, institutional resources, and international attention to a region that is particularly vulnerable to climate change both ecologically and socially.

In some cases, the best data just come from asking around. In a 2011 study, Bawa and his colleagues assessed how local villagers understood the changes in the landscape’s weather patterns and vegetation.

“When you go to the villages, people are talking about weather all the time,” he said. “Their daily activity is defined by what’s going to happen in the next two hours. ‘When is the rain going to end so we can do our chores?’ That’s a constant discussion.”

In other cases, they’ve just had to estimate. Jina and a team of researchers at the University of Chicago recently released a paper that models heat-related deaths in future climate scenarios. First, the team compiled one of the most comprehensive data sets to date on mortality: It covers 56 percent of the global population.

But this still left the other 44 percent to be filled in with a supercomputer model. “If we had data on 1 percent of the world, we could extrapolate out and say we had information on 100 percent of the world,” Jina said. “But the accuracy of that is open to question.”

“There’s this quote that I really love from Galileo that really describes science in a nutshell: ‘Measure what you can measure and make measurable what you can’t,’” he told me. “There are things which are very easy to measure for social and historical and other reasons, but as researchers, we’re not doing as well at measuring new things and new places. That’s an embedded problem in science and history—but usually it’s the most vulnerable people being written out of all the knowledge we’re generating. And I think that’s an enormous problem.”

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