The southwestern United States is no stranger to drought, experiencing dramatic fluctuations in aridity throughout history. However, the current drought, which began around 1998/99, is exceptionally severe due to the confluence of several factors: a rapidly growing population, intensive agriculture dependent on the region's water resources, and a warming climate exacerbated by rising greenhouse gases. This warming increases atmospheric evaporative demand, potentially reducing soil moisture and streamflows, and contributing to increased wildfire activity. The combination of reduced precipitation, warming temperatures, population growth, and intensive water use is placing unprecedented stress on water resources and agriculture, despite efforts to reduce water consumption. This research aims to determine the causes of the ongoing drought, investigating the roles of natural variability and forced climate change, and projecting future conditions. Climate models suggest a reduction in winter precipitation in the far southwestern US and robust spring drying across the West Coast, driven by changes in moisture advection. However, natural variability, such as the Pacific Decadal Oscillation (PDO) and Atlantic Multidecadal Oscillation (AMO), also plays a significant role. The wet period of the late 20th century coincided with a warm tropical Pacific phase of the PDO, while the current megadrought is associated with a cool phase. The AMO has been in a warm phase since the mid-1990s, also influencing drought variability. Therefore, understanding the interplay between radiatively-forced change and the evolution of the PDO and AMO is crucial for predicting future precipitation in the region. Coupled climate models have limitations in accurately simulating the interactions between the tropical Pacific and North American precipitation. Some models struggle to reproduce the observed Pacific and Atlantic decadal variability, while others may misrepresent the tropical Pacific's response to rising greenhouse gases. This study utilizes an atmosphere model forced by empirically derived SST variations from observations to overcome these limitations, ensuring more realistic projections and addressing several key questions regarding future drought scenarios in the Southwest.

Existing research highlights the long history of drought in the southwestern US, with studies documenting dramatic fluctuations in aridity over both instrumental and millennial timescales (Cook et al., 2004; Cook et al., 2007; Seager et al., 2005). The relationship between climate change and drought has also been investigated, with studies demonstrating the contribution of anthropogenic warming to increased drought risk (Diffenbaugh et al., 2015; Williams et al., 2020). The impact of warming on reduced river flows (McCabe et al., 2017; Xiao et al., 2018) and increased wildfire activity (Abatzoglou & Williams, 2016) has been established. The effects of natural climate variability, specifically the PDO and AMO, on southwestern US droughts have also been explored (McCabe et al., 2004; McCabe et al., 2008; Meehl & Hu, 2006), emphasizing the influence of tropical Pacific SST on precipitation patterns (Huang et al., 2005; Seager et al., 2007). Studies using coupled climate models have projected future changes in precipitation and soil moisture (Gao et al., 2014; Maloney et al., 2014; Ting et al., 2018; Cook et al., 2020), while others have focused on attributing observed drying trends to specific climate drivers (Lehner et al., 2018; Kumar et al., 2023). However, limitations in the ability of coupled models to accurately simulate decadal variability and the response of the tropical Pacific to rising GHGs necessitate the innovative approach adopted in this study.

This study employs an atmosphere-land model (CAM6-LR) forced by imposed sea surface temperatures (SSTs) to investigate the drivers of the southwestern North American megadrought. The model, a low-resolution version of the NCAR Community Atmosphere Model 6, is coupled to the Community Land Model (CLM) to simulate soil moisture. For the historical period (January 1979 – August 2021), the model was forced by blended SST data from NCAR, combining Hadley Center's HadISST1.1 and NOAA Optimal Interpolation data, along with standard CMIP6 forcings (excluding tropospheric aerosols). Sixteen ensemble members were generated with perturbed initial conditions. Future projections (to 2040) were generated using a large ensemble approach. The model was forced with different sequences of PDO and ENSO variability (generated using a cyclostationary linear inverse model, CSLIM), and AMO variability, coupled with two distinct scenarios of radiatively-forced SST trends: one based on an extrapolation of observed trends using regression to GHG forcing, and the other using the CMIP6 multimodel mean. A total of 80 simulations (grand ensemble) were generated for each SST trend. The southwestern US region was defined as 25°N to 40°N, 125°W to 100°W, land only. The CSLIM was used to generate ensembles of plausible SST scenarios that capture the characteristics of historical observations (1958-2017). This involved computing stationary and cyclostationary linear inverse models (LIM and CSLIM) on the nine leading principal components of global SST from HadISST data, separating the secular trend from natural variability. CSLIM was then integrated to generate synthetic SST data representing alternative histories of natural SST variability. These data were used to select various combinations of PDO and AMO states for the 2020-2041 period, ensuring consistency with historical patterns and preventing interference from the imposed anthropogenic SST trends. The selected PDO and AMO states were combined with the forced SST trends and a 1979-2018 HadISST climatology to create the total SST forcing fields for the model. Finally, analyses were performed to identify the separate effects of the PDO and AMO and to compare the future projections with the late 20th-century pluvial period.

The study's key findings include: 1. The 21st-century megadrought in the southwestern US is characterized by a striking cooling in the central to eastern tropical Pacific and a reduction in cool-season precipitation across the southwest, which the model accurately simulates as an SST-driven phenomenon. 2. The model successfully reproduces the observed decadal shift in cool-season precipitation and soil moisture, with a correlation coefficient of 0.66 between observed and modeled cool-season precipitation. 3. The model indicates that cool-season precipitation is a significant driver of summer soil moisture on interannual to decadal timescales, with little evidence of additional evaporative demand-driven drying beyond the effects of reduced precipitation. 4. Projections to 2040 reveal a strong dependence of future cool-season precipitation on the PDO and AMO. A cool tropical Pacific and warm tropical Atlantic (worst-case scenario) results in significantly reduced precipitation, while a warm tropical Pacific and cool tropical Atlantic (best-case scenario) leads to improved conditions. However, even in the best-case scenario, precipitation levels may not return to late 20th-century amounts. 5. The radiatively-forced response is stronger if the equatorial Pacific cold tongue shows no warming, similar to observed trends. 6. The projected future drying is driven by both reduced cool-season precipitation and warming that increases evaporative demand. 7. Despite the variations in scenarios, there is a high probability that soil moisture will not return to late 20th-century levels, emphasizing the lasting impacts of human-induced climate change.

This study demonstrates the dominant role of ocean-driven cool-season precipitation variability in shaping the ongoing and future drought conditions in the southwestern US. The findings highlight the critical importance of understanding natural decadal variability, specifically the PDO and AMO, in addition to radiatively-forced changes, for predicting future hydroclimate. The substantial impact of cool-season precipitation on summer soil moisture underscores the need for a seasonal perspective in drought analyses. The study’s projections reveal a considerable uncertainty in future precipitation and soil moisture, depending on the evolving states of the PDO and AMO. While a best-case scenario is possible, it is unlikely to restore conditions to those of the late 20th century. This emphasizes the lasting legacy of human-induced climate change on the region's hydroclimate, even if future natural variability leads to wetter conditions. These findings have significant implications for water resource management and drought adaptation strategies in the region.

This research provides compelling evidence for the dominant role of ocean-driven cool-season precipitation changes in the ongoing and future southwestern US megadrought. Natural decadal variability, particularly PDO and AMO interactions, will continue to influence near-term (next 2 decades) drought conditions. While a best-case scenario is possible, a return to late 20th-century moisture levels is unlikely. Future research should focus on improving our understanding and prediction of tropical Pacific decadal variability and its interaction with other climate modes, particularly the response of the cold tongue to rising greenhouse gases. Enhanced efforts to improve model simulations of these processes are critical for more robust predictions.

The study’s projections are limited by the uncertainties inherent in predicting natural decadal variability of the Pacific and Atlantic Oceans. While the CSLIM approach generates plausible scenarios, it does not offer precise forecasts of future ocean states. The model’s relatively low resolution may also affect the accuracy of regional climate projections, although the focus on broader spatial patterns reduces this concern. Additionally, other factors, such as land-use changes and water management practices, can impact water resources and are not explicitly accounted for in the modeling. However, these limitations do not detract from the study's central finding: the significant influence of ocean-driven precipitation variability on the ongoing and projected drought in the southwestern US.