cw3e-web - Center for Western Weather and Water Extremes

CW3E Publication Notice: Probabilistic Predictions from Deterministic Atmospheric River Forecasts with Deep Learning

CW3E Publication Notice

Probabilistic Predictions from Deterministic Atmospheric River Forecasts with Deep Learning

October 29, 2021

William E. Chapman, PhD candidate with CW3E, recently published a paper (Chapman et al., 2021) in Monthly Weather Review, along with co-authors F. Martin Ralph, Negin Hayatbini, and Luca Delle Monache (CW3E/Scripps Institution of Oceanography), Shang-Ping Xie (Scripps Institution of Oceanography), Stefano Alessandrini (National Center for Atmospheric Research), Aneesh C. Subramanian (University of Colorado Boulder), and Sebastian Lerch (Karlsruhe Institute of Technology). The paper, titled “Probabilistic Predictions from Deterministic Atmospheric River Forecasts with Deep Learning”, contributes to the goals of CW3E’s 2019-2024 Strategic Plan to support Atmospheric River (AR) Research and Applications and Emerging Technologies by examining Deep Learning (DL) methods for generating reliable and accurate probabilistic forecasts of integrated vapor transport (IVT) under AR conditions.

This study uses a 34-year reforecast, based on CW3E West-WRF mesoscale model of North American West Coast IVT, to test the dynamically/statistically derived 0-120 hour probabilistic forecasts for IVT under AR conditions. These predictions are compared to the Global Ensemble Forecast System (GEFS) dynamic model and the GEFS calibrated with a neural network. The DL methods are also tested against an established, but more rigid, statistical-dynamical ensemble method (the Analog Ensemble). Using continuous ranked probability skill score and Brier skill score as verification metrics, the study’s findings show that the DL methods compete with or outperform the calibrated GEFS system at lead times from 0-48 hours and again from 72-120 hours for AR vapor transport events (Figure 1). The study examines the implications of varying the length of the training dataset, with results showing that the DL methods learn relatively quickly and ~10 years of hindcast data are required to compete with the GEFS ensemble. The results from this study indicate that DL methods, which are flexible and easy to implement with modern DL toolboxes, generate reliable and skillful probabilistic forecasts, learn relatively quickly, and can compete with or outperform the GEFS system for AR vapor transport events. While there is still work to be done to optimize these methods, this study shows that they are already capable of adding value to state of the art weather prediction.

Figure 1: Brier skill score at thresholds of 250 (a), 350 (b), and 500+ (c) [kg m^-1 s^-1] forecasted units of IVT against the forecast lead time for the ensemble mean or predicted mean from each forecast system. The GEFS ensemble is used as the reference forecast, with positive values showing percent improvement. Shown predictions include GEFSnn (dark red), GEFS (light red), CNN (white), NN (light blue), AnEn (blue), FCN (dark blue). The error bars indicated the 95% bootstrap confidence intervals (where n=1000).

Chapman, W. E., Delle Monache, L., Alessandrini, S., Subramanian, A. C., Ralph, F. M., Xie, S., Lerch, S., & Hayatbini, N. (2021). Probabilistic Predictions from Deterministic Atmospheric River Forecasts with Deep Learning, Monthly Weather Review, https://doi.org/10.1175/MWR-D-21-0106.1.

CW3E Event Summary: 19-26 October 2021

October 28, 2021

Click here for a pdf of this information.

Strong Atmospheric Rivers Bring Heavy Precipitation and Flooding in California

Multiple atmospheric rivers (ARs) made landfall along and impacted the US West Coast between 19 Oct and 26 Oct
The first two ARs produced AR 4 conditions in southwestern Oregon and AR 2/AR 3 conditions were observed elsewhere along the coast from the San Francisco Bay Area to the Olympic Peninsula.
The third AR reached AR 5 conditions over California near Point Reyes due to the combination of maximum IVT values (> 1000 kg m−1 s−1) and AR duration (> 48 hours) (based on the Ralph et al. 2019 AR Scale).
The third AR was the strongest October AR to make landfall in the Bay Area in the last 40 years.
Portions of Northern California received more than 15 inches of total precipitation from the three storms.
Intense rainfall on 24 Oct caused flooding in the Bay Area and triggered multiple slides in Northern California.
The combination of heavy rain and high winds downed trees and caused power outages throughout the Bay Area.
Although there was heavy rain, reservoirs throughout the western Sierra Mountains and else where only saw small gains in storage.
The third and strongest AR was well forecasted out to 3-days lead time, with the presence of the AR well forecasted out to 6 days.
Subseasonal forecasts showed higher AR activity 3 weeks out.

Click images to see loops of GFS IVT/IWV analyses Valid 0000 UTC 17 October – 0000 UTC 26 October 2021

NOAA/NESDIS/STAR GOES-West GEOCOLOR

Valid 1200 UTC 23 October – 1200 UTC 26 October 2021

Summary provided by CW3E personnel; 28 October 2021

To sign up for email alerts when CW3E post new AR updates click here.

*Outlook products are considered experimental

CW3E AR Update: 22 October 2021 Outlook

October 22, 2021

Click here for a pdf of this information.

As atmospheric river conditions begin to dissipate from the current AR, focus turns to extreme AR over Weekend

AR conditions associated with the current AR bringing precipitation to California are forecast to end in the late morning and early afternoon
A much stronger AR is forecast to make landfall over California Saturday night
Models continue to forecast AR 5 conditions over the San Francisco Bay Area with AR 3 to 4 over a majority of the North and Central CA Coast
The California-Nevada River Forecast Center (CNRFC) is forecasting as much as 13 inches of precipitation to fall across the high-elevations in the Northern Sierra
While soils are currently dry due to extensive drought, the National Weather Service has issued numerous flash flood and flood watches due to the potential for high rain rates and post-fire debris flows

Click images to see loops of GFS IVT & IWV forecasts Valid 1200 UTC 22 October – 1200 UTC 29 October 2021

Summary provided by C. Hecht, C. Castellano, B. Kawzenuk, J. Kalansky, L. Odell, and F. M. Ralph; 22 October 2021

To sign up for email alerts when CW3E post new AR updates click here.

*Outlook products are considered experimental

CW3E AR Update: 20 October 2021 Outlook

October 20, 2021

Click here for a pdf of this information.

Multiple Atmospheric Rivers to Bring Heavy Precipitation to Northern California

A series of landfalling atmospheric rivers (ARs) will impact the western US this week into early next week
AR 4/AR 5 conditions (based on the Ralph et al. 2019 AR Scale) are expected in coastal southern Oregon in association with the first and second ARs today through Friday
The strongest AR is forecasted to make landfall across Central and Northern California on Sunday, potentially bringing AR 4/AR 5 conditions to the San Francisco Bay Area
Inland penetration of this AR may bring AR 2/AR 3 conditions to portions of the interior western US
Yet another landfalling AR is forecasted to impact the US West Coast on 26–27 Oct
The first two ARs are forecasted to produce 2–5 inches of rainfall in portions of Northern California and southern Oregon
The fourth AR is forecasted to bring widespread precipitation to much of the western US, with the heaviest precipitation amounts in Northern California
Significant snowfall accumulations are also possible in the Sierra Nevada in association with the fourth AR
Portions of Northern California may receive more than 10 inches of total precipitation over the next 7 days

Click images to see loops of GFS IVT & IWV forecasts Valid 1200 UTC 20 October – 1200 UTC 28 October 2021

Summary provided by C. Castellano, C. Hecht, B. Kawzenuk, J. Kalansky, and F. M. Ralph; 20 October 2021

To sign up for email alerts when CW3E post new AR updates click here.

*Outlook products are considered experimental

CW3E Welcomes Shawn Roj

October 19, 2021

Shawn Roj joined CW3E as a research data analyst in October 2021. He received a BS in Atmospheric Sciences from the University of Nevada, Reno (UNR) in 2019 and a MS in Earth Sciences, with a focus on climate science, from Scripps Institution of Oceanography (SIO) in 2021 under the advisement of Professor Joel Norris.

Shawn was a civil engineering land development design draftsman for a firm in Mammoth Lakes, CA for 10 years before returning to school. During his studies at UNR he was a part time student worker at the Western Regional Climate Center (WRCC) with Nina Oakley. Not only did he gain insight into the realm of climatology, but he was also awarded funding for two undergraduate research projects. The first project, under the mentorship of Dr. Benjamin Hatchett at the WRCC, was to investigate snow level rise in the Feather River basin by identifying hourly precipitation fractions as rain or snow. The second project, under the mentorship of Dr. Dan McEvoy at the WRCC, was to validate and bias correct the National Weather Service forecast reference evapotranspiration in Nevada using the Nevada Integrated Climate and Evapotranspiration network. This project became his senior thesis and developed into a journal article that is currently in revision. For his MS thesis at SIO he began the arduous task of investigating the importance of condensational heating of a snowpack during atmospheric rivers (ARs) in the Northern Sierra Nevada Mountains of California. Data for this study was collected at the Mammoth Mountain energy balance site called CUES. He found that the process was not an important driver of snowmelt during ARs at the site. Because CUES is at a high elevation and colder than many weather stations within the Feather and Yuba River watersheds during ARs, a sensitivity study with warmer temperatures showed that the process may be more important in these locations.

At CW3E, Shawn’s initial tasks will be working with the Yuba/Feather FIRO team on freezing level forecast skill metrics, helping out with AR outlooks and storm event summaries, and continuing work on his MS research in order to publish a journal article.

CW3E AR Update: 18 October 2021 Outlook

October 18, 2021

Click here for a pdf of this information.

Active weather is forecast to continue, bringing multiple landfalling ARs to the U.S. West Coast

The first AR is forecast to make landfall over Northern California on Tuesday evening, bringing moderate to strong AR conditions to the region
The second AR is forecast to make landfall on Thursday and is forecast to be stronger and last longer than the first AR
Current forecasts suggest the active pattern to continue with a third AR potentially making landfall between 23 and 25 October, but forecast uncertainty is high due to the long lead times
The Weather Prediction Center is forecasting >10 inches of precipitation over the higher elevations of the Pacific Northwest and Northern California due to these successive and potentially strong storms

Click images to see loops of GFS IVT & IWV forecasts Valid 1200 UTC 18 October – 1200 UTC 28 October 2021

Summary provided by C. Hecht, Shawn Roj, J. Kalansky, and F. M. Ralph; 18 October 2021

To sign up for email alerts when CW3E post new AR updates click here.

*Outlook products are considered experimental

2021 North American Monsoon Recap

October 15, 2021

Click here for a pdf of this information.

2021 North American Monsoon marked by wetter-than-normal conditions in southwestern US

The North American Monsoon (NAM) refers to a shift in the synoptic-scale wind pattern that transports low-to-midlevel moisture from the Eastern Pacific and Gulf of Mexico into the southwestern US during summer
The NAM is an important source of annual precipitation for parts of the southwestern US
Unlike the stronger Indian Monsoon, the NAM is characterized by episodic bursts of moisture transport and rainfall
The 2021 monsoon season was characterized by an abundance of moisture and frequent precipitation episodes, particularly in Arizona
Arizona experienced its 9th wettest July–September period since 1895
Anomalously wet conditions during July–September brought much-needed drought relief to portions of the Four Corners Region

Summary provided by C. Castellano, J. Kalansky, and F. M. Ralph; 15 October 2021

CW3E AR Update: 14 October 2021 Outlook

October 14, 2021

Click here for a pdf of this information.

Prolonged Period of Active Weather Expected in the Pacific Northwest

An atmospheric river (AR) will make landfall across Washington early Friday morning
AR 3 conditions (based on the Ralph et al. 2019 AR Scale) are forecasted in coastal Washington
Limited moisture and weak storm dynamics are expected to constrain precipitation totals outside of the Olympic Peninsula and North Cascades
After a brief lull, a period of very active weather is forecasted to impact to the U.S. West Coast, with a series of landfalling ARs becoming increasingly likely during 20–24 October
The highest forecast confidence in landfalling AR activity during this period is in Northern California, Oregon, and Washington, but there is considerable uncertainty in the evolution of these ARs and their associated low-pressure systems

Click images to see loops of GFS IVT & IWV forecasts Valid 1200 UTC 14 October – 0000 UTC 18 October 2021

Summary provided by C. Castellano, C. Hecht, B. Kawzenuk, J. Kalansky, and F. M. Ralph; 14 October 2021

To sign up for email alerts when CW3E post new AR updates click here.

*Outlook products are considered experimental

CW3E Welcomes Weiming Hu

October 13, 2021

Weiming Hu joined CW3E as a Machine Learning Postdoctoral Researcher in September 2021. He received his Ph.D. in Geography in 2021 from the Pennsylvania State University under the supervision of Prof. Guido Cervone.

Weiming’s research interest lies at the intersection of Atmospheric Science, Geography, and Computer Science. His doctoral dissertation is titled Uncertainty Quantification for Photovoltaic Energy Production Using Analog Ensemble. He mainly investigated the predictability of photovoltaic solar energy production over the Continental US and proposed a Machine Learning based similarity metric for weather analogs. This research is computation-intensive because a large historical repository is needed to find good-quality weather analogs.

He also conducted research on rare and extreme event forecasting like heatwaves in the US and Europe. Urban areas are particularly vulnerable to temperature increases due to not only the heat island effect but also the complex demographics and different levels of facility accessibility. Weiming and his team showed that regional models can be downscaled to the street level using crowdsourced data. The downscaled forecasts can provide valuable information and timely alert for the local population and government.

During his Ph.D. training, Weiming served as a Departmental Graduate Representative, and he mainly supported the mentor-mentee program where new students are paired with senior students to help them better adapt to the new environment. He has been an active member of the American Geophysical Union since 2016 and a fervent supporter of the Free Opensource Software campaign since 2019.

At CW3E, Weiming will continue his research under the supervision of Dr. Luca Delle Monache on weather forecasting, mainly on short-term forecasts and subseasonal-to-seasonal forecasts using Machine Learning and Deep Learning. The research outcome will significantly benefit water governance and operations of water reservoirs in California and even the Western states in the future.

CW3E Participates in a Dialogue on Drought in the Upper Colorado’s Yampa River Basin after the First-Ever Shortage Declaration on Lake Mead

October 13, 2021

The annual Yampa Basin Rendezvous (YBR) is part of a larger CW3E effort to connect graduate students, post-doctoral scholars, researchers, staff, and faculty from CW3E to the local communities of river basins throughout the West, along with local, state, and federal experts in policy and relevant science. This year, the conversation focused on drought and its impacts on the environment, people, and the economy was incredibly timely. The Yampa River is one of the wildest remaining major tributaries of the Colorado River, and provides crucial water supplies to local stakeholders and to locations as far removed as Arizona and Southern California. The American West, as a whole, is in the throes of a serious, multi-year drought that resulted in extremely low runoff in the Yampa this year. In August 2021, this important Colorado River source river was put on call for the third time in history (with all three calls having happened since 2018) and the first ever shortage was declared on Lake Mead.

CW3E and our partners at Colorado Mountain College, Friends of the Yampa, Yampa Valley Sustainability Council, Steamboat Ski and Resort Corporation, Yampa River Fund, and the Upper Yampa Water Conservancy District, among others, virtually came together on September 9th and 10th, 2021, for the fourth annual YBR. YBR 2021 was a two-day series of four interactive panels examining the Yampa River Basin through the lens of drought, water availability, and conservation. The panels included speakers and moderators with expertise on the regional and national scale and featured lively, engaging discussions including nearly 150 participants. The agenda can be found here.

During our four panels, key voices on drought and conservation were represented. Statewide experts and makers of Colorado water policy included the Colorado Commissioner of Agriculture, the State Engineer and Director of the Colorado Department of Water Resources, as well as representatives from the Colorado Water Conservation Board and the Maybell Irrigation District.

All spoke about the dire situation and the challenges facing our communities now, and emphasized the need for collaborative solutions between agencies and institutions that rely on the Colorado river for a variety of different reasons. The science panel on the first day highlighted the challenges in predicting reduced runoff from the snowpack this year. In several presentations the soil moisture deficit and thirstier atmosphere was discussed in detail, along with innovative efforts to better forecast future years’ water supply. Audience members were engaged and asked a great number of questions, enhancing the discussions guided by expert moderators. All panel talks and discussed were recorded and can be viewed here.

This year, the steering committee introduced two exciting new features to YBR, which will be part of the annual event moving forward:

Enhancing student involvement: Colorado Mountain College and CW3E held a session highlighting relevant undergraduate student research. Each student presented a “poster” in the form of a 3-slide lightning presentation.
Art: While some of the art integration had to be put off until next year with the switch to virtual, Jill Bergman, local artist and steering committee member, worked with a local art gallery to host an exhibit to correspond with YBR. The exhibit Routt County Icons: Sandhill Cranes and the Yampa River is hanging at Pine Moon Fine Art Gallery during the month of September. Virtual displays of the art pieces were shared during breaks in the morning events (like the image below!).

CW3E is committed to continuing to grow engagement, learning, and beneficial outcomes to enhance science-and-community-based understanding and ideas to deal with a changing climate, economy and environment. The steering committee is grateful for the opportunity to partner with so many Yampa-based organizations and institutions on impactful events like the YBR, and to serve as a training ground for people across disciplines to advance their goals and careers in support of the incredible needs of the future. These needs are not only relevant for the Upper Colorado communities and environment but also of the millions of people who depend upon the precious resource it provides to people downstream.