雅思阅读 191: When the Mountain Lets Go（当山体松开时）

改编由 Natural Hazards and Earth System Sciences / AGU。雅思阅读 Section 3 难度，约 1050 词。 素材来源：https://nhess.copernicus.org/articles/26/611/2026/nhess-26-611-2026.html

Reading Passage

A. A landslide is, in geological terms, a sudden and violent event: seconds separate a stable hillside from a torrent of rock and mud that can bury a village. But the instability that precedes it builds slowly, invisibly, over weeks of rain. Understanding when a slope will fail — not whether it will, but on which hour — is one of the oldest and most difficult problems in engineering geology. The stakes are enormous. In mountainous regions from the Chinese Loess Plateau to the Pacific Northwest, rainfall-triggered landslides kill thousands of people every year and destroy roads, reservoirs and whole settlements. As extreme rainfall becomes more frequent under a changing climate, the need for accurate, localised warnings has moved from academic curiosity to a matter of public safety. A slope that has stood for a thousand years can fail in an afternoon if the rain arrives in the wrong pattern. Population growth in mountainous regions has also put more people, homes and infrastructure in the paths of slides that would once have fallen on uninhabited ground.

B. The physical mechanism is deceptively simple. A dry soil slope is held together by matric suction — the tension that exists between water and air in unsaturated soil pores, gluing particles together like damp sand in a child's bucket. When rain falls, it infiltrates downward, raises the water content and collapses that suction. Pore water pressure rises, particles lubricate, and the frictional strength holding the slope against gravity falls. Engineers express this as a factor of safety, a ratio of resisting forces to driving forces. When the ratio stays above 1.25, the slope is considered safe; below 1.0, failure is inevitable. Experiments on unsaturated loess slopes in China have shown how nonlinearly this works. Under a heavy storm of 120 millimetres per day on a 30-metre slope, the factor of safety drops below the engineering threshold in a single day and below 1.0 after three days. Under gentler rain at 5 millimetres per hour, the same slope can remain stable or metastable for far longer. The same total rainfall therefore produces very different outcomes depending on whether it arrives in hours or days.

C. The practical question is which slopes, at which moments, cross that threshold. Two complementary approaches have emerged. The first is physically based modelling, which uses real measurements of soil depth, slope angle, bedrock topography and rainfall infiltration to compute the factor of safety for every hillside in a region. A widely used framework called the Fast Shallow Landslide Assessment Model runs these calculations across entire counties, combining a susceptibility map — which slopes are inherently dangerous — with rainfall thresholds — how much rain, over how long, tips them over. The second approach is statistical and machine-learning based, which trains algorithms on historical landslide inventories and weather records to predict where and when failures are likely. A 2026 study in Zixing, China, found that combining machine-learning susceptibility maps with typhoon-specific dynamic rainfall thresholds outperformed older, fixed-threshold systems, particularly by distinguishing between short, intense downpours and longer, cumulative soil saturation. The two approaches are increasingly run together rather than chosen between.

D. Typhoons present a particular challenge. A typhoon brings not one steady rainstorm but a sequence of bands — outer rainbands, eyewall, inner bands — each delivering a different intensity and duration. A slope may survive the first band because the soil has not yet reached a critical saturation, only to fail hours later when a second band arrives on already-wet ground. Traditional warning systems that use a single total-rainfall threshold for an entire event miss this distinction. The newest models, by contrast, track both antecedent rainfall — how wet the slope was before the storm began — and short-term forecast rainfall during the event itself. A three-dimensional warning framework developed in China combines background susceptibility, antecedent effective rainfall and now-cast rainfall, using a genetic algorithm to optimise the threshold. Early results suggest this approach improves prediction success by roughly ten per cent over conventional statistical models. It also allows warnings to be issued hours earlier, when evacuation is still possible. A warning that arrives after the slide has moved is, however well phrased, no warning at all. The goal is the opposite: a few hours of lead time in which roofs can be reinforced, roads closed and families moved to safer ground. Those few hours are the difference between a disaster and a manageable event. In mountain villages, they are also the difference between life and death. No warning system is perfect, but every improvement saves lives, and every false alarm that is taken seriously builds the trust that the next real alarm will be heeded.

E. Even the best model cannot remove uncertainty. Soil depth varies across a single slope by metres, bedrock topography is hidden beneath the surface, and a single root system from a tree can reinforce a patch of hillside that sensors cannot see. Ecological studies have shown that grass roots — for example, those of tall fescue on loess slopes — measurably increase the factor of safety by binding surface soil. Warnings therefore err on the side of caution, which produces both missed events and false alarms that exhaust public trust. The frontier of the field lies in combining the three approaches — physical models, machine learning and real-time rainfall nowcasting — so that warnings are local, dynamic and specific enough to act on. The mountain will always let go without warning in the end; the aim is to give the people below as much warning as the mountain will permit, and to ensure that when the alarm sounds, the village takes it seriously. Good warnings, issued accurately and acted upon, are the cheapest landslide protection available.

Questions 1-4

Choose the correct heading for paragraphs B, C, D and E from the list of headings below.

List of Headings i. The physics of how rain destabilises a slope ii. The history of landslide disasters in Europe iii. Two approaches to predicting where and when slopes fail iv. Why typhoons are especially difficult to warn for v. The remaining uncertainty and the path forward vi. How earthquakes trigger tsunamis vii. Why all landslides are easy to predict

1. Paragraph B: ____
2. Paragraph C: ____
3. Paragraph D: ____
4. Paragraph E: ____

Questions 5-8

Choose the correct letter, A, B, C or D.

5. What holds a dry soil slope together? A. Concrete retaining walls. B. Matric suction between water and air in soil pores. C. Tree roots alone. D. Gravity pulling downward.

6. What did the loess-slope experiment find about a 120 mm/day storm on a 30 m slope? A. The slope remained completely stable. B. The factor of safety dropped below 1.0 after three days. C. The slope failed instantly within minutes. D. No pore water pressure was generated.

7. What is the advantage of typhoon-specific dynamic thresholds? A. They use a single total-rainfall number for the whole storm. B. They distinguish short intense downpours from cumulative soil saturation. C. They work only in desert regions. D. They replace the need for rain gauges.

8. How do grass roots help stabilise a slope? A. They absorb all the rainwater. B. They bind surface soil and raise the factor of safety. C. They make the soil more slippery. D. They have no measurable effect.

Questions 9-13

Do the following statements agree with the claims of the writer?

Write:

• TRUE if the statement agrees with the information
• FALSE if the statement contradicts the information
• NOT GIVEN if there is no information on this

9. Under gentler rain at 5 mm/h, a 30-metre slope can remain stable for longer than under a 120 mm/day storm.
10. The factor of safety is considered safe when it falls below 1.0.
11. The Zixing study found that combining machine learning with dynamic thresholds outperformed fixed-threshold systems.
12. A three-dimensional warning framework has been developed using a genetic algorithm.
13. Landslides cause more deaths annually than earthquakes worldwide.

Questions 14-15

Complete the summary below using NO MORE THAN TWO WORDS from the passage.

As rain infiltrates, it collapses matric suction and raises pore water (14) __________, causing the factor of safety to fall (15) __________.

答案与解析

← 上一篇 | 返回雅思焦点 | 下一篇 →