Author: Sam Bartlett

CW3E Publication Notice

Wet Antecedent Soil Moisture Increases Atmospheric River Streamflow Magnitudes Non-Linearly

May 28, 2025

A new study published in the Journal of Hydrometeorology shows that wet soils significantly increase flood risk during atmospheric rivers (ARs). The paper, titled “Wet Antecedent Soil Moisture Increases Atmospheric River Streamflow Magnitudes Non-Linearly,” was led by Desert Research Institute Ph.D. candidate Mariana Webb with co-authors Christine Albano (DRI), Adrian Harpold (UNR), Daniel Wagner (USGS), and Anna Wilson (CW3E).

By analyzing more than 43,000 AR events across 122 watersheds on the U.S. West Coast between 1980 and 2023, the researchers found that peak streamflow is, on average, 2-4.5 times higher when soils are already wet compared to when they are dry. This response is non-linear, with AR-driven streamflow sharply increasing when antecedent soil moisture exceeds a watershed-specific threshold (Figure 1). Together, these findings help explain why some ARs cause catastrophic flooding while others of comparable intensity do not. Weaker ARs (AR1-2), normally considered low risk, can generate major floods if they coincide with wet antecedent conditions, while many stronger ARs (AR4-5) may not lead to flooding when conditions are dry.

The study also found that antecedent soil moisture plays a larger role in determining flood magnitude in specific physiographic and climatic settings. In regions like California and southwestern Oregon, AR-driven streamflow is highly sensitive to antecedent soil moisture. These watersheds typically have shallow, clay-rich soils, lower winter precipitation, and higher evaporation, leading to limited hydrologic storage capacity and more variable soil moisture between storms. In contrast, in Washington and the interior Cascades and Sierra Nevada, watersheds tend to have deeper soils, snowpack, and consistently wet winter conditions, which increase storage capacity and reduce soil moisture variability. While antecedent soil moisture still can play a role, it provides less added value for flood prediction in these regions because soils are often consistently wet or insulated by snow.

Tailoring flood risk evaluations to a specific watershed’s physiography and climate characteristics could improve flood-risk predictions. Because AR-driven streamflow varies not just with storm intensity but also with antecedent soil moisture, the study highlights the value of integrating land surface conditions into AR impact assessments. The strong alignment between each watershed’s antecedent soil moisture threshold and its average winter soil moisture opens the door to using seasonal soil moisture averages as a practical proxy for identifying heightened flood potential—simplifying implementation in flood forecasting models and early warning systems. The study further highlights the importance of prioritizing watersheds where antecedent soil moisture has a greater influence on flood response, such as those with limited storage and high variability in soil moisture. Increased monitoring in these high-risk catchments, including real-time soil moisture observations, could significantly enhance early warning systems and flood management as AR frequency and intensity continue to evolve with climate change.

To read more about this study access the entire publication here.

Figure 1. Map showing the antecedent soil moisture threshold, above which event maximum streamflows are larger, at each watershed. Watersheds with no statistically significant (p-value <0.05) threshold value (n=14) are shown in grey. Inset scatterplots show individual maximum streamflows for atmospheric river (AR) events as a function of the associated ASM for four watersheds. The vertical red line indicates the location of the calculated ASM threshold. (Figure 4 from Webb et al. (2025))

Webb, M. J., Albano, C. M., Harpold, A. A., Wagner, D. M., & Wilson, A. M. (2025). Wet Antecedent Soil Moisture Increases Atmospheric River Streamflow Magnitudes Non-Linearly. Journal of Hydrometeorology., https://doi.org/10.1175/JHM-D-24-0078.1.

CW3E Publication Notice

Characteristics of Precipitation Patterns in Moisture-dominated versus Wind-dominated Atmospheric Rivers over Western North America

May 16, 2025

A paper titled “Characteristics of Precipitation Patterns in Moisture-dominated versus Wind-dominated Atmospheric Rivers over Western North America” was recently published in the AGU’s Journal of Geographical Research: Atmosphere. This study was led by Wen-Shu Lin (CW3E) and co-authored by scientists in CW3E: Joel Norris, Mike DeFlorio, Jonathan Rutz, Jason Cordeira, and Marty Ralph. This work aligns with CW3E’s strategic goal to advance understanding of ARs, extreme precipitation, and their associated impacts, and was sponsored by the California Department of Water Resources Atmospheric River Program and US Army Corp of Engineers.

This paper discussed the mechanisms for precipitation associated with two AR flavors over the western North America: the moist-dominated (moist-ARs) and wind-dominated ARs (windy-ARs). The moist-ARs are categorized as ARs with stronger moisture but weaker wind, while the windy-ARs are categorized as ARs with stronger wind but weaker moisture. Interestingly, the two AR flavors have distinct precipitation patterns (Fig. 1), although they share similar duration and integrated water vapor transport (IVT). The moist-ARs are associated with stronger precipitation in the inland and interior areas, while windy-ARs are associated with stronger precipitation at the coastal area (Figs. 1d-f).

The possible mechanisms leading to the difference in precipitation patterns between moist-ARs and windy-ARs are investigated based on the synoptic patterns of several variables. Moist-ARs are associated with a zonally-oriented and weaker trough at 500 hPa (Fig. 2 left column), leading to an enhanced westerly IVT, and thus favoring inland penetration of moisture. Moist-ARs are also associated with stronger moisture in the upper than lower troposphere (Fig. 3), that more moisture might be transported into inland and interior areas. In contrast, windy-ARs are associated with an enhanced trough at 500hPa and enhanced cyclone at sea level (Fig. 2 middle column) that provide greater pressure gradient force to maintain the stronger wind speed of windy-ARs. These lead to an enhanced southerly and southwesterly IVT, especially in the lower troposphere (Figs. 3a-c). At the coastal area, windy-ARs also accompany with stronger IVT convergence and synoptic scale ascent forcing than moist-ARs, which are beneficial for windy-ARs producing precipitation at the coastal areas.

Figure 1. (a-c) Mean precipitation accumulated over AR duration [mm AR^-1] in NCA, ORWA, and BC, respectively. (d-f) Precipitation pattern differences [mm AR^-1] between moist-ARs and windy-ARs in NCA, ORWA, and BC, respectively. Black dots on (d-f) indicate that the differences between moist-ARs and windy-ARs are significant at the 95% level based on a Student’s t-test. The pink dots mark the grid points of coastal region, the red triangles mark the grid points of inland transects, and the purple triangles mark the grid points of interior transects. (Figs. 4a-f from Lin et al. 2025)

Figure 2. Spatial patterns of H500 anomaly (shading in m), SLP anomaly (green contours every 2hPa), T850 anomaly (brown contours every 1 K) for moist-ARs, windy-ARs, and the differences between moist-ARs and windy-ARs landfalling in (a-c) NCA, (d-f) ORWA, (g-i) BC averaged over the AR total duration. For all contours, the zero lines are omitted, solid lines are positive, and dashed lines are negative. The black dots on (c, f, i) indicate that the difference in H500 anomaly between moist-ARs and windy-ARs is significant on a 95% level based on a Student’s t-test. (Fig. 5 from Lin et al. 2025)

Figure 3. The differences of IVT anomaly (vectors and shading in kg m^-1 s^-1), IWV anomaly (blue contours), and WIND anomaly (brown contours every 1.5 m s^-1) between moist-ARs and windy-ARs and for (a-c) 1000 – 700hPa (blue contours every 1 kg m^-2) and (d-f) 700 – 200hPa (blue contours every 0.2 kg m^-2). The black dots indicate that the difference in IVT anomaly between moist-ARs and windy-ARs is significant on a 95% level based on a Student’s t-test. (Fig. 7 from Lin et al. 2025)

Lin, W.-S., Norris, J. R., DeFlorio, M. J., Rutz, J. J., Cordeira, J. M., & Ralph, F. M. (2025). Characteristics of precipitation patterns in moisture‐dominated versus wind‐dominated atmospheric rivers over western North America. Journal of Geophysical Research: Atmospheres, 130(9), e2024JD041966.. https://doi.org/10.1029/2024jd041966

CW3E Publication Notice

Beyond Expectations: Investigating Anomalous 2022-2023 Winter Weather Conditions and Water Resources Impacts in California

May 6, 2025

A new paper titled “Beyond Expectations: Investigating Anomalous 2022-2023 Winter Weather Conditions and Water Resources Impacts in California” authored by researchers from the Center for Hydrometeorology and Remote Sensing (CHRS) at UC Irvine, in collaboration with CW3E researcher Vesta Afzali Gorooh and Eric Shearer from the U.S. Army Engineer Research and Development Center, was recently published in the Bulletin of the American Meteorological Society (BAMS). The study, led by UC Irvine PhD Candidate Claudia Jimenez Arellano, comprehensively analyzes the anomalously wet and cold winter of 2022-2023 in California, characterized by substantial “weather whiplash.” Using data spanning 2002 to 2023, the research team examined key hydrometeorological variables such as precipitation, temperature, snowpack, reservoir storage, and atmospheric river (AR) events.

Findings revealed that wintertime precipitation for 2022-2023 reached the highest levels in over two decades in Southern California, driven primarily by 11 moderate to strong AR events (Figure 1). Interestingly, despite a lower number of intense AR events than in the 2016-2017 winter season (Figure 2), Southern California experienced higher cumulative precipitation in 2022-2023. This study highlights that approximately 360 mm of cumulative precipitation fell in Southern California, while about 700 mm accumulated statewide. Notably, a sequence of AR events in March, categorized as Category 4, 2, and 1 on March 9th, 19th, and 28th, significantly boosted statewide precipitation (Figure 3). The increase in precipitation during March considerably improved the state’s total water storage, reaching levels comparable to other notably wet years despite starting from significantly lower reservoir levels.

Overall, the anomalous winter conditions of 2022-2023 significantly alleviated drought across California, leaving 94% of the state drought-free by the conclusion of the water year. Lower-than-average temperatures in the subsequent spring and summer months also slowed snowpack melting, effectively mitigating flood risks despite record-high snow accumulation. The study emphasizes AR events’ critical role in California’s water resource management and the increasing importance of forecasting tools like Forecast-Informed Reservoir Operations (FIRO) in adapting to future climatic extremes.

Figure 1. AR landfall location, date, category, and coverage over California during winter 2022-2023. (Figure 5 in Arellano et al. 2025)

Figure 2. AR landfall location, date, category, and coverage over California during the winter of 2016-2017. (Figure 6 in Arellano et al. 2025)

Figure 3. Daily cumulative precipitation and number of AR events across all categories over California and Southern California during a) 2010-2011, b) 2016-2017, c) 2018-2019, and d) 2022-2023 winter. (Figure 4 in Arellano et al. 2025)

Arellano, C. J., Rouzegari, N., Dao, V., Zadeh, M. B., Shearer, E., Gorooh, V. A., Nguyen, P., Hsu, K., & Sorooshian, S. (2025). Beyond Expectations: Investigating Anomalous 2022-2023 Winter Weather Conditions and Water Resources Impacts in California. Bulletin of the American Meteorological Society (published online ahead of print 2025), BAMS-D-23-0336.1. https://doi.org/10.1175/BAMS-D-23-0336.1.

CW3E Publication Notice

Toward Calibrated Ensembles of Neural Weather Model Forecasts

April 23, 2025

Scientists from the CW3E machine learning team recently published an article titled “Toward calibrated ensembles of neural weather model forecasts” in the Journal of Advances in Modeling Earth Systems. This study was led by Jorge Baño-Medina (CW3E) and co-authored by Agniv Sengupta (CW3E), Duncan Watson-Parris (SIO), Weiming Hu (James Madison University), and Luca Delle Monache (CW3E). This work aligns with CW3E’s strategic goal to develop artificial intelligence (AI)-based predictive capabilities for extreme weather associated with atmospheric rivers (ARs). The work was supported by the California Department of Water Resources AR Program and the U.S. Army Corps of Engineers’ Forecast Informed Reservoir Operations.

The central question addressed in this study is whether AI data-driven models, that have revolutionized the domain of weather prediction in the past couple of years, can be used to generate very large, calibrated ensemble forecasts. Answering this question is crucial for advancing our ability to create sharp weather predictions and reliably quantify their uncertainty. This, in turn, provides avenues to improve probabilistic forecasting and the prediction of extreme events, which are critical for informed and effective decision-making.

AI-based weather models offer a transformative opportunity to produce significantly larger ensembles with far lower computational costs and in much shorter time frames than traditional dynamical systems. By overcoming these computational barriers, very large ensembles based on AI models can better represent the tail of the predictive distribution, which is associated with extreme events, when compared to traditional methods. Physics-based models still play a crucial role in the process of generating a prediction, because they are a key component (along with data assimilation and a wealth of observations) to generating the training data sets needed by AI models.

This study introduces an innovative methodology to generate ensemble forecasts using AI-driven global weather models. Our methodology accounts for two critical sources of uncertainty: model uncertainty and initial condition uncertainty (Figure 1). For model uncertainty, we use a novel technique of model checkpointing—sampling the model weights at various epochs during training—resulting in a diverse ensemble of 90 distinct models. With this approach the AI model needs to be trained only once. For initial condition uncertainty, we apply the breeding of growing modes technique, traditionally used in numerical weather prediction, to generate six bred vectors. This combination of 90 models and six bred vectors results in an unprecedented ensemble of 540 members, enabling three-dimensional global weather forecasts of several parameters up to 5 days in advance. Remarkably, this ensemble is generated in only 5 minutes on a single graphics processing unit (GPU), with even shorter times achievable with a larger number of GPUs.

Figure 1. Schematic of the proposed strategy to generate a 540‐member ensemble. An initial condition field, defined by the state of a set of atmospheric variables, is perturbed with 6 bred vectors (initial condition perturbations). Then the perturbed initial conditions feed each of them 90 distinct NWMs (model perturbations), to generate a 540‐member ensemble. Figure 1 from Baño-Medina et al. (2025).

Our approach exhibits significantly lower errors and better calibration—where the ensemble spread closely matches the error of the ensemble mean—compared to benchmark AI-based systems, and it rivals the world-leading physics-based/stochastic probabilistic systems of the European Centre for Medium-Range Weather Forecasts, for key atmospheric variables: air surface temperature, surface zonal wind velocity, total column of water vapor and geopotential at 500 hPa. To visualize the performance of the entire ensemble, the total column water vapor was averaged over the Feather River basin for an impactful atmospheric river over the West coast of North America from April 2018. Figure 2 displays the evolution of the 7‐day forecast for each of the 540 members from EnAFNO540 (turquoise), the ensemble mean of EnAFNO (blue) and ERA5 (black). The ensemble is able to accurately capture the evolution of the atmospheric river up to 7‐days of lead time.

Figure 2. Temporal series of the total column water vapor averaged over the Feather River Basin for an impactful AR of April 2018 for the 540 EnAFNO members (turquoise), the ensemble mean (blue) and ERA5 (black). Figure 4 from Baño-Medina et al. (2025).

In conclusion, this paper offers promising insights into the transformative potential of AI weather models for probabilistic forecasting, particularly in the context of extreme weather events.

Baño‐Medina, J., Sengupta, A., Watson‐Parris, D., Hu, W., & Delle Monache, L. (2025). Toward calibrated ensembles of neural weather model forecasts. Journal of Advances in Modeling Earth Systems, 17(4), e2024MS004734, https://doi.org/10.1029/2024MS004734.