Where does northwest China's increasing moisture come from? New study points to local sources
image:
Schematic illustration of the mechanisms responsible for the humidification in Northwest China.
view moreCredit: Haipeng Yu
For the millions living in Northwest China's arid expanse, where water scarcity has shaped life for centuries, a quiet question has emerged: why is it getting wetter?
Now, scientists who have spent their careers studying this region offer a surprising answer. The recent increase in precipitation is not arriving from afar. It is, instead, mostly rising up from the land itself.
Published in Advances in Atmospheric Sciences, the study was conducted by researchers from the Northwest Institute of Eco-Environment and Resources at the Chinese Academy of Sciences, Lanzhou University, and the Lanzhou Institute of Arid Meteorology of China Meteorological Administration. The team shares deep roots in the region they study.
Lead author Professor Haipeng Yu first arrived in Lanzhou as a student in 2005, drawn to study China's dryland climate. He never left. Senior authors Professors Jianping Huang and Qiang Zhang were both born in Northwest China and have devoted their entire careers to understanding its arid and semi-arid regional climate.
Their combined decades of observation have tracked a fundamental shift.
A Turning Point in the 1990s
Traditionally, precipitation in Northwest China was thought to depend heavily on moisture carried in from outside. The new research confirms that while external moisture still dominates the long-term average, the region's trend toward humidification since the late 20th century is being driven largely by local sources: enhanced evaporation from soil, plants, and water bodies—a process intensified by warming temperatures and ecological changes.
The research identifies the late 1990s as a critical turning point. Around that time, summer precipitation shifted from a long-term decline to a sustained increase. Spatially, the changes are uneven—western areas around the Tianshan and Altun Mountains have seen substantial increases, while some eastern parts have experienced drying trends.
Tracing the Source
Using a Dynamic Recycling Model, the team quantified the contributions of different moisture sources. Comparing 1961–1997 with 1998–2020, they found that annual precipitation increased by 10.62 mm (9.18%), while local evapotranspiration rose by 10.42 mm (9.12%). Crucially, nearly 78% of the increase in precipitation came from locally recycled moisture, with only about 22% from enhanced external transport.
This marks a fundamental shift in understanding. More than half of the region's average precipitation still comes from outside, but the increase since the late 1990s is overwhelmingly local in origin.
Land–Atmosphere Coupling
Warmer temperatures, increased meltwater from glaciers and snowpack, and vegetation recovery have all contributed to rising evapotranspiration, creating a feedback loop that fuels additional precipitation.
"For decades, the textbook answer was that Northwest China's rain comes from somewhere else," says Yu. "Our findings show that since the late 1990s, the dominant contribution to precipitation growth has shifted to local moisture recycling. After spending my entire adult life here, it's remarkable to see the data confirm that something fundamental is shifting."
Implications and Uncertainties
The study notes that large-scale oceanic variability, such as the Atlantic Multidecadal Oscillation, may add complexity to future projections. The authors also caution that the current trend may not be sustainable. As glaciers and snow reserves decline under continued warming, the meltwater that supports enhanced evapotranspiration could diminish, potentially slowing or reversing the humidification.
"This work provides quantitative evidence that local hydrological feedbacks have become the dominant mechanism behind recent precipitation increases," says Professor Zhang. "That has important implications for drought monitoring, prediction, and water-resource management."
"The warm–wet shift reflects an integrated response of the regional water cycle to warming, cryospheric changes, and ecosystem recovery.” Professor Huang adds: Having grown up here, I know these aren't abstract questions—they affect real communities, real farms, real lives."
Journal
Advances in Atmospheric Sciences
Article Title
Local Evapotranspiration and Atlantic Decadal Variability Dominate the Humidification of Northwest China
Finer-scale simulations show promise for forecasting dangerous valley storms
Institute of Atmospheric Physics, Chinese Academy of Sciences
image:
The Tethering Horse Post on Laji Mountain stands as a dramatic sentinel over the complex terrain of Eastern Qinghai. This towering limestone pillar, rising abruptly from the ridge, exemplifies the kind of rugged topography that makes weather forecasting in the region so challenging.
view moreCredit: Qinghai Meteorological Observatory
As climate change intensifies the water cycle, communities in mountainous regions face growing threats from flash floods and landslides triggered by sudden, violent rainstorms. An international research team has shown that increasing the resolution of weather forecasting models to the kilometre scale could improve the ability to predict these events—not just in China's Qinghai Province, but in complex terrain worldwide.
The study, published in Advances in Atmospheric Sciences, focused on a catastrophic rainstorm that struck the Hongshui River valley in eastern Qinghai on 13 August 2022. The event caused widespread flooding, damaged crops, and affected nearly 6,000 households. Using the Weather Research and Forecasting (WRF) model, scientists compared simulations at different resolutions: 9 kilometres, 3 kilometres (matching China's operational forecasts), and 1 kilometre.
The results showed clear differences. Only the 1-kilometre simulation accurately captured the storm's intensity, timing, and location.
"The 1-kilometre grid spacing allowed the model to capture the subtle wind patterns within the valley that actually triggered the storm," said Yongling Su, lead author of the study and a forecaster at the Qinghai Meteorological Observatory. "As the sun heated the valley slopes during the day, predictable upslope winds developed. But in the evening, these collided with cooler air draining down the mountainsides, creating narrow lines of forced rising air that ignited the thunderstorm cells. At coarser resolutions, these critical details were simply smoothed out."
The research revealed that the thermodynamic conditions for storms—instability and moisture—were similar across all simulations. The critical difference lay in how well the models represented the low-level valley winds that provide the final trigger for convection.
"For forecasting extreme precipitation in complex mountainous terrain, increasing resolution from 3 kilometres to 1 kilometre can yield noticeable forecast improvements," said Robert Plant, Professor of Meteorology at the University of Reading and corresponding author of the study. "The 1-kilometre grid enables the model to better simulate the intricate flow structures within valleys that govern where and when the most dangerous storms develop. This is relevant not just for Qinghai but for mountain valleys in many parts of the world."
The findings have implications for operational forecasting. While running ultra-high-resolution models across entire continents remains computationally demanding, the researchers propose a more targeted approach.
"Our goal is practical," Su explained. "We want to provide forecasters in Qinghai and similar mountainous regions with more precise tools. By running higher-resolution 'on-demand' forecasts when and where dangerous storms are anticipated—essentially zooming in on high-risk areas within broader operational models—we may be able to issue heavy precipitation warnings earlier and more accurately."
The study also highlighted a limitation of traditional "convective parameterization" schemes—mathematical formulas that approximate storm development. In simulations where these schemes were active, weak rainfall began too early, followed by a delay in the main storm, effectively disrupting the model's timing.
"Using a convection parameterization scheme led to premature removal of early atmospheric instability," Plant noted. "This delayed the real storm and reduced its intensity in the simulation. When we let the model represent convection directly at high resolution, the timing and magnitude aligned more closely with observations."
While the study analysed one event in depth with supporting evidence from a second, the researchers suggest the underlying mechanisms are likely applicable more broadly.
"This is about understanding how valley circulations develop—how air moves up slopes during the day and drains down at night—and how these flows can contribute to storm triggers," Plant added. "Better representation of these wind patterns in models supports better predictions."
This approach, the researchers suggest, could strengthen disaster prevention efforts in mountainous regions globally, from the Andes to the Alps, the Himalayas to the Rockies.
Journal
Advances in Atmospheric Sciences
Article Title
The Benefits of Kilometre-scale Simulations for Extreme Summertime Precipitation in the Eastern Valleys of Qinghai
Article Publication Date
7-Mar-2026
