Land subsidence is a geohazard resulting from sudden or gradual (years to decades) collapse of the earth’s surface due to the removal of underground material. This can occur due to a variety of factors, both natural (such as earthquakes, volcanic activity, and compaction of fine-grained, unconsolidated sediments) and anthropogenic (such as mining and groundwater extraction). It poses a serious problem in urban areas as it can cause buildings to collapse, damage infrastructure, threaten life and create resource management problems.

As land subsidence increases in densely populated areas, one of the main drivers of subsidence is groundwater withdrawal; that is, the removal of water stored in the underground pore space and its transport for human consumption and agricultural irrigation results in soil compaction. underground deposits.

This is the focus of a new study published. Geophysical Research LettersScientists found a significant positive correlation between groundwater withdrawal and subsidence rate, implying that these areas should be the main focus for water management to reduce this geohazard.

Tsymur Davydenko, Ph.D. A researcher from the Colorado School of Mines (USA) and colleagues Dr. Pejman Tahmasebi and Professor Nima Shokri turned to deep learning AI to predict land subsidence on a global scale. Explaining the importance of their research, Tzimur and Dr. Tahmasebi said: “Land subsidence is a devastating phenomenon that damages infrastructure and groundwater and endangers people’s lives. It is stated that population growth plays an undeniable role in the extraction of resources, which leads to collapse.

“Based on our research, we realized that there are many local studies investigating land subsidence in different countries, but not enough research on a global scale. Such maps either do not provide information about the amount of subsidence, which is more important than just being able to observe it, or they are limited to regional scales. Using state-of-the-art data-driven modeling techniques, we have presented the first global map of subsidence rates that can be used to inform groundwater management and mitigation policies.”

The researchers used existing data from land subsidence studies and remote sensing to create a training dataset of 46,000 subsidence scenarios. These, along with a selection of 23 climate, geographic and topographic conditions (including rainfall, soil composition, sediment thickness and slope), were used to train a machine learning model that could then predict the total land area and population at risk of subsidence. in these areas.

They determined that it was more than 6.3 million km ² The Earth’s surface (~5% of total land area) is subject to subsidence rates considered significant enough to cause damage and require mitigation strategies; these exceed 5 mm per year. This follows from previous work that suggested 12 million km. ² There has been a subsidence of 430 mm per year on the Earth’s surface. More than 6.3 million km of these ² 231,000 km ² Population density has been identified in urban areas indicating that approximately 2 billion people (25% of the world’s population) are located in these high-risk areas.

Global population density in areas affected by rapid subsidence (>5 mm/yr; a) and increased for North America (b), South America (c), Europe and North Africa (d), the Middle East (e), and the South, East and Southeast Asia (f)

The machine learning model identified groundwater withdrawal as the main determinant of subsidence, followed by seismic activity from earthquakes, then environmental conditions affecting groundwater recharge (i.e. precipitation), thickness of sedimentary units (larger units have more space for final compaction), average temperature. of the hottest months (important for arid and semi-arid regions susceptible to subsidence), clay content of the soil and population density.

Given that groundwater extraction is a major problem, Tsimour and Dr. Tahmasebi suggests ways in which the planet’s population’s dependence on it could be changed in the future. “Strategic measures to minimize dependence on groundwater include increasing water efficiency, implementing a strict regulatory framework and promoting agricultural practices that optimize water use. Additionally, investments in water recycling and reclamation technologies can increase water availability without over-reliance on groundwater.

“Phasing out groundwater extraction can be complemented by the use of alternative sources such as treated wastewater, rainwater harvesting and rainwater management. However, the transition must also take into account the spatial limitations of the reservoirs and the economic feasibility of seawater desalination. By integrating conservation practices, using technological innovations, and diversifying water sources, a sustainable water supply ecosystem can be created that will reduce environmental and socio-economic problems associated with groundwater overdraft.”

The sediments most affected by the collapse are 3.8 million km² A large portion of unconsolidated sediments (10% of the global area) was identified as the main risk, with the highest subsidence rate being 320.6 mm/yr. Cultivated land represented the largest area at risk, at 2.1 million km2² Worldwide (12.2% of the world’s cropland) subtropical mountainous regions and temperate oceanic climates have experienced increasing subsidence rates of more than 50 mm per year.

Overall, South Asia is modeled as having the largest land area at risk of subsidence (2.2% of total land area experiencing subsidence rates of more than 50 mm per year) and the largest number of people affected (20 million). Other countries where subsidence rates exceed 50 mm per year include the Philippines, Iran, Costa Rica, Indonesia and Uzbekistan.

While this study provides an important global soil subsidence map that will help businesses, farmers and local governments in high-risk areas plan for future challenges, the model needs further refinement. Therefore, Tsimour and Dr. Future models need to have high enough resolution that local governments can use data for mitigation strategies, Tahmasebi said. “definitely possible” He says it is.

For example, given the importance of groundwater extraction in the dataset, which includes more detailed information on extraction depth, aquifer type, delay between extraction and groundwater withdrawal, and the interaction of the oil and gas industry (which contributes to 4.36% of the current sedimentation record), These are necessary steps to strengthen this important work. With population growth increasing our dependence on groundwater and droughts exacerbating climate change, the impact of groundwater depletion on land subsidence will continue to be a pressing issue in the coming decades.