The Tibetan Plateau (TP), often called the "Asian water tower," is a vital source of freshwater for twelve major Asian river systems, supporting nearly two billion people. Its unique alpine ecosystems are rich in biodiversity but highly sensitive to climate change. Anthropogenic greenhouse gas (GHG) forcing and elevation-dependent warming have caused the TP to warm significantly faster than the global average over the past 60 years, leading to environmental changes including glacial retreat, lake expansion, altered streamflow, permafrost degradation, and vegetation changes. Interestingly, contrasting precipitation patterns are observed: drying in the southeast and northwest Himalayas, and wetting in the inner TP. Previous studies have suggested the weakening Indian monsoon and uneven aerosol emissions as contributing factors. However, the mechanisms behind these contrasting precipitation changes, particularly the inner TP wetting, remain unclear. This study investigates the interplay between large-scale circulation changes, local moisture recycling (specifically precipitation recycling), and the observed hydrological shifts on the TP. The research aims to quantify the role of warming-induced changes on evapotranspiration and the feedback to atmospheric conditions and precipitation, while also considering the impact of both local and non-local moisture sources on the current and future climate sensitivity of the Asian water tower. Understanding these processes is crucial for developing effective water resource management strategies and ensuring climate resilience for downstream societies.

Existing literature highlights the TP's sensitivity to climate change, with studies examining the influence of the weakening Indian monsoon and anthropogenic aerosols on precipitation patterns. Some research has explored the role of local moisture recycling using isotopic analysis and budget methods, but estimates of the precipitation recycling ratio vary significantly across studies, ranging from 10% to 80%. The lack of a comprehensive understanding of the relative contributions of local vs. non-local moisture sources, particularly concerning the observed contrasting wetting and drying patterns across the TP, necessitates a more detailed investigation using advanced techniques.

This study employs a three-dimensional (3D) Lagrangian approach for moisture tracking, coupled with the WaterSip moisture-source diagnostics, to analyze the TP's hydrological changes over 40 years (1971-2010). The Lagrangian model is driven by the CERA-20C reanalysis product, providing high-resolution data for accurate tracking of atmospheric moisture parcels. The WaterSip diagnostic allows for the quantification of both local (precipitation recycling) and non-local moisture contributions to precipitation, accounting for the cascading effect of moisture loss during transport. The study classifies global circulation regimes to determine the relative importance of the Indian monsoon and westerlies in transporting moisture to the TP. A multi-product ensemble of evapotranspiration and precipitation data from multiple observational products and reanalyses is constructed to evaluate the biases in the CERA-20C data and improve the robustness of the analysis. The study also leverages single-forcing CMIP6 simulations from the Detection and Attribution Model Intercomparison Project (DAMIP) to investigate the role of anthropogenic GHG and aerosol forcing on TP precipitation and moisture recycling. Specifically, the five best-performing CMIP6 models, based on their ability to reproduce the observed inner wetting-Himalayan drying pattern, are used for ensemble analysis. Moisture budget decomposition is used to separate the contributions of various physical processes to precipitation changes, including evapotranspiration, thermodynamic and dynamic effects, transient eddies, non-linear effects, and moisture storage. Statistical analyses, including the Mann-Kendall test and Student's t-test, are used to assess the significance of trends.

The study's key findings include: 1. **Dominant Role of Precipitation Recycling:** Precipitation recycling accounts for approximately half of the TP's summer precipitation and 46.5% annually, significantly higher than previously estimated using 2D Eulerian tracking methods. This underscores the TP's increasing dependence on its internal water cycle. 2. **Inner Wetting Driven by Recycling and Convergence:** The observed inner TP wetting is primarily attributed to a surge in evapotranspiration and enhanced moisture convergence, counteracting the decline in moisture transport from the Indian monsoon and westerlies. 3. **Himalayan Drying due to Moisture Divergence:** In contrast, Himalayan drying results from moisture divergence that outweighs increased evapotranspiration in the region. 4. **Weakening of Monsoon and Westerlies:** The Lagrangian tracking reveals a significant decline in the fractional contributions of both the Indian monsoon and the westerlies to TP precipitation, indicating a shift in dependence from external to internal moisture sources. 5. **Dual Effect of GHG Forcing:** CMIP6 simulations confirm a dual effect of GHG forcing: enhanced wetting via increased evapotranspiration and precipitation recycling, while concurrently weakening the southerly monsoons via atmospheric stabilization. Anthropogenic aerosol forcing temporarily countered this effect, causing drying, but this impact is anticipated to diminish in the future. 6. **Faster Growth of Evapotranspiration:** Evapotranspiration increases faster than precipitation in percentage terms, both presently and in future projections under various climate scenarios, emphasizing the growing vulnerability of the TP’s hydrological cycle to local land surface changes. 7. **Increased Sensitivity to Local Processes:** The TP's water cycle is becoming increasingly self-constrained, with greater sensitivity to local evapotranspiration and land-atmosphere interactions, and reduced dependence on external moisture sources.

The findings highlight the crucial role of precipitation recycling in shaping the TP's recent hydroclimatic changes, a factor that has been underappreciated in previous studies. The dual effect of GHG forcing, increasing both local evapotranspiration and concurrently weakening monsoon-driven moisture transport, further amplifies this vulnerability. This shift toward a more self-constrained water cycle indicates increased sensitivity to local land-atmosphere interactions, making the TP's hydrological system increasingly vulnerable to disruptions. These findings support recent research highlighting the importance of reduced aerosol emissions in moistening High Mountain Asia and contradict those implying that oceanic climate change is the main driver for terrestrial water storage deficit in the TP. The results emphasize the need for improved climate models that accurately simulate land-atmosphere interactions and regional moisture recycling dynamics to better predict future hydrological changes.

This study demonstrates the accelerating impact of human-induced warming on the TP's water cycle through amplified precipitation recycling and a decline in external moisture sources. The increasingly self-constrained nature of the TP's hydrological cycle underscores its growing vulnerability to local land surface changes, particularly in the context of continued GHG emissions. Future research should focus on improving climate model resolution, parameterizations, and data quality to enhance the accuracy of projections. This enhanced understanding is crucial for effective water resource management and climate change mitigation strategies in High Mountain Asia.

While this study employs a comprehensive methodology, several limitations need acknowledgement. The CMIP6 models used in the study, though selected for their superior performance, still exhibit uncertainties in simulating TP precipitation. The Lagrangian tracking approach, although advanced, is subject to inherent uncertainties associated with numerical diffusion and atmospheric dynamics. The analysis relies on reanalysis data, which may introduce uncertainties due to inherent biases and limitations in atmospheric measurements. Future research should consider incorporating higher resolution simulations and more comprehensive datasets to refine the estimates.