A groundbreaking study has unveiled compelling evidence of decadal shifts in Earth’s atmospheric circulation, achieved through an innovative analysis of cloud motion vectors captured by NASA’s Multi-angle Imaging SpectroRadiometer (MISR). Spanning two decades, from 2000 to 2020, this research leverages high-resolution stereoscopic imagery to characterize how winds traverse the globe at varying altitudes, revealing subtle yet significant changes that may reshape our understanding of climate dynamics.
The MISR instrument aboard NASA’s Terra satellite offers a unique vantage point, capturing cloud motion at an unprecedented spatial resolution of approximately 17.6 kilometers by 17.6 kilometers. Unlike common gridded monthly mean products typically used in climate studies, the MISR Cloud Motion Vector (CMV) product provides a comprehensive list of individual cloud track retrievals tagged with precise latitude, longitude, altitude, and time information. These height-resolved vectors are derived through a sophisticated stereoscopic method, comparing cloud features across multiple viewing angles within short intervals of 3.5 minutes, allowing researchers to directly observe the atmospheric flow conditioned by cloud top dynamics.
Utilizing only the daytime descending node of the Terra satellite’s orbit to maintain consistency in local solar time across high-latitude regions, the analysis ensures robust temporal sampling. The reliability of MISR products is well established, with extensive validations against independent datasets showing minimal biases and measurement precisions on the order of a few meters per second in wind speed and several hundred meters in cloud-top height. This validation underpins the confidence in their suitability for long-term trend analysis of atmospheric motion.
In parallel, the study incorporates the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 reanalysis dataset, which provides high-resolution global atmospheric data across 37 pressure levels. These hourly reanalysis fields, gridded at 0.25 degrees latitude and longitude, offer a comprehensive picture of the atmospheric state, crucial for benchmarking and corroborating trends observed in the MISR data. A novel nearest-neighbour matching technique synchronizes ERA5 wind components with the precise time, location, and altitude of each MISR observation, creating the ERA5_MS dataset for direct comparison.
Central to the research is a meticulous aggregation of observational data into a three-dimensional grid composed of latitude, longitude, and altitude layers arranged in 1-kilometer vertical bins spanning from the surface to 20 kilometers in height. Monthly mean velocity components (zonal U and meridional V) and wind speed are computed, facilitating detailed examination of spatial and temporal trends. To enhance statistical robustness, only bins with at least 5,000 cloud motion vectors over the 20-year period contribute to the zonal average analyses, primarily focusing on the troposphere where clouds predominantly reside.
Trend analyses employ rigorous statistical methods, including the nonparametric Mann–Kendall test for monotonic trends and Sen’s slope estimator to quantify trend magnitudes. To mitigate false positives arising from multiple testing across thousands of grid points, an advanced False Discovery Rate (FDR) correction is applied, ensuring that only statistically robust trends are highlighted. This stringent framework enables detection of subtle changes in wind patterns with a high degree of confidence.
The study examines two key metrics derived from the zonal mean zonal wind component to characterize shifts in large-scale circulation: Lat_U0, the latitude where the low-level wind near the surface changes direction from easterly to westerly, marking the tropical circulation edge; and Lat_Umax, the latitude where the jet stream attains its maximum zonal wind speed in the lower troposphere. These measures, calculated using the Tropical-width Diagnostics (TropD) software package, allow tracking of tropical expansion or contraction and poleward shifts of jet streams, critical indicators of changing atmospheric dynamics.
A thorough investigation into potential confounding factors confirms the integrity of the trend signals. Analyses reveal negligible changes in the sample sizes of cloud motion vectors within each bin and minimal biases introduced by shifts in cloud top height distributions or longitudinal sampling over time. Correlations between wind speeds and cloud heights within altitude bins are weak, and spatial patterns of confounding variables do not align with observed wind trends. This reinforces the conclusion that observed changes predominantly reflect genuine atmospheric circulation variations rather than artefacts of sampling or observational inconsistencies.
Furthermore, assessments of cloud-conditioning biases in both MISR and ERA5 datasets indicate that while MISR CMVs are inherently cloud-top conditioned, ERA5 sampling reveals only a modest correspondence between cloud fraction and sampled winds. Notably, restrictive criteria for identifying cloud tops in ERA5 samples demonstrate that the wind field analyses remain consistent across various cloud conditions, affirming the robustness of the comparative results and underscoring known limitations in modelled cloud physics within reanalyses.
Comparative evaluations with independent wind measurements and prior studies bolster confidence in the findings. Validation against satellite scatterometers and altimeters reveals close agreement between ERA5 surface winds and MISR-derived wind speeds, with latitudinally dependent biases generally below 1 meter per second. Aeolus Doppler lidar data further corroborate these findings in the free troposphere, highlighting both the strengths and shortcomings of numerical weather prediction models embedded within ERA5, especially near the tropopause where Kelvin wave dynamics and associated wind shear are notoriously difficult to simulate.
Seasonal analyses of multi-decadal trends reveal largely consistent behavior across all seasons, indicating that the drivers of wind trend patterns operate independently of seasonal variability. However, two exceptions emerge: the southern hemisphere polar jet’s strengthening diminishes during boreal winter, potentially linked to stratospheric ozone depletion’s influence on polar circulation; and the subtropical jet’s strengthening seen in ERA5 is less pronounced or absent in MISR data during certain seasons, suggesting differences in representation of moist convection and atmospheric dynamics between observations and models.
Together, these nuanced insights paint a compelling picture of an evolving atmospheric circulation system, shaped by complex interactions among climate drivers, clouds, and large-scale dynamics. The ability to detect such decadal changes using MISR’s cloud-top-conditioned winds emphasizes the transformative potential of satellite stereoscopic observations in climate monitoring. These findings not only deepen understanding of atmospheric responses to climate forcing but also highlight areas where reanalysis models can be further refined to better represent the coupling of clouds and circulation.
The implications of this work extend beyond atmospheric science, offering vital information for forecasting extreme weather, understanding shifts in storm tracks, and anticipating changes in precipitation patterns tied to jet stream variability. As Earth’s climate continues to evolve, leveraging high-precision satellite observations alongside sophisticated reanalysis is proving indispensable in unraveling the complex tapestry of climate change impacts.
This study sets a new benchmark for longitudinal atmospheric observations, combining advanced remote sensing techniques with cutting-edge data assimilation. It exemplifies how merging diverse datasets can yield transformative insights into planetary-scale processes, underscoring the critical role of sustained satellite missions and open data in advancing climate science.
Looking ahead, the integration of these cloud motion vector datasets with emerging instruments like Doppler lidars and next-generation climate models promises enhanced capability to track and predict atmospheric circulation shifts. As computational and observational technologies advance, studies like this pave the way for increasingly detailed, accurate depictions of our dynamic atmosphere and its future trajectory under changing climatic conditions.
Subject of Research: Decadal changes in atmospheric circulation analyzed through cloud motion vectors and reanalysis data.
Article Title: Decadal changes in atmospheric circulation detected in cloud motion vectors.
Article References:
Di Girolamo, L., Zhao, G., Zhang, G. et al. Decadal changes in atmospheric circulation detected in cloud motion vectors. Nature (2025). https://doi.org/10.1038/s41586-025-09242-1
Image Credits: AI Generated
