Field Studies

---

1. Introduction

Santa Ana wind events vary widely in duration and intensity, directly influencing destructive wildfire behavior across Southern California’s urban-wildland interface. This study analyzes four major cold-flavor Santa Ana events from 2020 to 2025 and proposes a synoptic classification framework to improve the prediction and public communication of windstorm duration and severity.

Using ERA5 reanalysis in conjunction with 10-minute sustained and peak gust observations from Eaton Canyon, we identify three potential synoptic variants based on 500 mb relative vorticity patterns that cause strong mountain wave windstorms in Southern California: single-trough, double-trough, and rotor-trough events. To test the hypothesis that major cold-flavor windstorms fall into one of these three categories, we retrospectively applied the framework to 50 years of historical windstorms, identifying candidates using ERA5 data. Historical windstorms were cross-verified using newspaper archives. Results indicate that 13 of the 17 flagged windstorm events between 1970 and 2019 were confirmed through media archives, with the four unverified cases all occurring prior to 1987.

Findings also reveal that double-trough events, while not always producing the highest gusts, sustain damaging winds for significantly longer durations than other variants and therefore pose the greatest risk for wildfire spread into urban areas, particularly when they occur before meaningful rainfall. In contrast, rotor-trough events are associated with the most powerful windstorms observed to date. Improved communication of the approaching synoptic variant may enhance resource allocation and preparedness among emergency personnel.

---

2. Methodology

We selected all four of the major Santa Ana windstorms to have struck Altadena, where the historic Eaton Fire occurred, from the past 5 years. These events occurred on:

• February 3-4, 2020
• January 21-22, 2022
• March 14, 2024
• January 7-8, 2025

Each event was identified via eyewitness recount and available weather station data at SCE Eaton Canyon (SE215) and SCE Loma Alta / McNally (SE609), both of which located in the Altadena area. These stations were selected due to their unobstructed exposure, with minimal nearby trees or structures that could otherwise affect wind speed or direction measurements. Wind events were included in this analysis only if peak gusts reached at least 50 MPH at either station. For each qualifying event, the corresponding upper-level trough was examined using ERA5 reanalysis data accessed through the PG&E 2-km WRF Model Visualization Project, maintained by the Wildfire Interdisciplinary Research Center at San José State University. Troughs were identified and tracked using 3-hour 500 mb relative vorticity fields.

Wind speed data from mesonet sensors built and operated by Southern California Edison (SCE) at Eaton Canyon near Altadena were compiled into histograms showing the frequency distributions of 10-minute sustained wind speeds and 10-minute maximum observed gusts at 5 MPH intervals.

Each event began when either wind gusts reached 30 MPH or 10-minute sustained winds reached 15 MPH, and ended when those thresholds were no longer met for at least 12 hours. These thresholds are general considered consequential for wildfire spread according to National Weather Service Red Flag Warning criteria. These specifications allow an event separated into two halves with a lull in-between to still be considered one single windstorm. Once the histograms were created, each event was then sorted by both strength and duration.

Events were classified based on surface observations to look for a correlation. Once it was established how many variants there could be, a full investigation into 50 years of ERA5 reanalysis was conducted to determine whether past wind events between 1970 and 2019 could also be place into one of the three variants. These additional wind events, which will not be examined further due to a lack of weather station data, were then searched in media databases for possible stories published about them.

The February 2020 event lasted a cumulative 17.0 hours during which wind gusts exceeded 30 MPH. Despite its duration, the storm produced only 1.5 hours of damaging gusts greater than 50 MPH. Two peaks in intensity were observed: one near the onset and a more significant one at the tail end. Of the 1.5 hours of damaging winds, 1.3 hours occurred during this final surge. The event featured three notable lulls—after sunrise, shortly after solar noon, and again following sunset. Between lulls, winds were persistent, with long stretches of gusts ranging from 25 to 40 MPH.

The January 2022 event persisted for 8.5 hours—half the duration of the 2020 storm—but generated 2.7 cumulative hours of damaging wind gusts ≥50 MPH. Wind activity was concentrated into three distinct peaks: a weak pre-sunset surge, a stronger episode just before midnight, and a slightly stronger peak before sunrise. Damaging gusts were evenly distributed across the two overnight surges. The peaks were separated by a prolonged lull after sunset and a brief lull just after midnight. Wind intensity fell sharply following each peak without forming plateaus.

The March 2024 windstorm lasted 8.5 hours and produced 3.0 hours of damaging wind gusts ≥50 MPH. The event developed gradually, initiating a few hours before sunrise and intensifying steadily through mid-morning. Damaging gusts began shortly after sunrise and peaked during mid-morning hours. Wind speeds gradually declined by solar noon and diminished further in the hour following. Unlike the 2020 and 2022 events, the wind speed profile for this storm resembled a semi-circular curve—characterized by a slow rise and fall with no sharp peaks or lulls.

The January 2025 windstorm was exceptionally long and intense, lasting 23.3 hours and producing 12.2 cumulative hours of damaging wind gusts ≥50 MPH—nearly double the total of the other three events combined or 5x any other event analyzed. The event began rapidly a few hours before sunrise on January 7, with damaging gusts recorded within the first hour. A plateau in wind intensity followed, extending through the mid-morning hours, with the strongest gusts occurring near sunrise. A sharp collapse in wind speed occurred mid-morning, leading to a lull that lasted until shortly after solar noon.

A second phase began abruptly in the early afternoon, returning to damaging gusts within one hour. A new plateau in intensity then extended through midnight and into the early morning of January 8. During this period, the Eaton Canyon SCE weather station was burned over as the Eaton Fire passed through, causing a slight bump in windspeed. The maximum gust of the event occurred a few hours after burn over. A brief lull followed midnight, succeeded by the most powerful surge of the event in the pre-dawn hours. This surge was the only occurrence in the dataset of tropical-storm-force sustained wind speeds. The event weakened rapidly before sunrise, followed by a final qualifying surge mid-morning, and ended two hours before solar noon on the second day.

Analysis of the four recent wind events alongside their synoptic environments revealed that all could be categorized into one of three distinct upper-level patterns: single-trough, double-trough, or rotor-trough events. Each variant exhibited consistent characteristics in terms of vorticity structure, jet placement, windstorm duration, and intensity profile. A subsequent scan of ERA5 data from 1970 to 2019 identified 17 additional historical periods with similar synoptic setups. While surface wind observations were not available for these older events, their upper-level characteristics suggest they are strong candidates for past major cold-flavor Santa Ana windstorms. Of these 17 cases, 13 events produced articles in local media outlets. The four that did not all occurred before 1987.

Rotor Events: Rotor-trough events are characterized by large, cold-core upper-level troughs drifting south-southwest from the Great Basin. As these troughs intensify over the Mojave Desert, they often generate shortwave ejections to their south and west. These ejections result in pulses of cyclonic relative vorticity ranging from 25 × 10⁻⁵ s⁻¹ to 60 × 10⁻⁵ s⁻¹ crossing the Transverse Ranges. This dynamic interaction produces brief but exceptionally intense mountain wave windstorms. Wind speed profiles during these events are typically arc-shaped, lacking sharp peaks or distinct lulls, and often last fewer than 12 hours, with peak gusts ≥50 MPH occurring for under 6 hours.

Examples likely include events on January 20, 1980; January 27, 1987; November 30, 1991; November 14, 1993; October 27, 1996; December 1, 2011; February 17, 2012; and March 14, 2024. The March 2024 event, for which observational data are available, showed a clear arc-like wind profile consistent with a rotor-trough structure. A similar feature was observed in the first phase of the January 7, 2025 event, suggesting a hybrid structure with both rotor and double-trough dynamics. In both cases, observed surface wind gusts patterns coincided with the passage of a shortwave ejection aloft.

Single-Trough Events:
Single-trough events are the most common variant. They typically involve a single, moderately sized trough diving southeastward from the Great Basin into the lower Colorado River Valley. As the trough progresses, the western flank of its jet streak often becomes positioned over Los Angeles County, initiating mountain wave formation across the Transverse Ranges. These events are associated with continuous relative vorticity advection (as opposed to discrete pulses), resulting in prolonged but less intense wind episodes.

Duration often exceeds 12 hours, with damaging wind gusts ≥50 MPH sustained for at least 3 hours. Wind speed profiles tend to show a singular peak followed by gradual decline, although multiple surges may occur depending on smaller-scale embedded features. Historical examples include January 1, 1973; January 27, 1984; February 21, 1985; November 2, 1986; February 19, 1988; December 8, 1988; January 16, 1991; January 6, 1997; January 6, 2003; and January 22, 2022. The 2022 event, supported by observational data, exhibited multiple well-defined surges without extended plateaus, likely associated with localized vorticity maxima moving along the jet.

Double-Trough Events: Double-trough events are the rarest but most hazardous variant due to their prolonged duration and potential for multiple episodes of damaging winds. These setups begin similarly to single-trough events, with a trough diving through the Great Basin and initiating a mountain wave. However, a second shortwave trough, often originating in the Northern Rockies, rapidly follows the first and plunges toward Southern California along the same upper-level jet axis.

This two-stage process leads to distinct windstorm phases separated by a lull, often producing two overnight periods of damaging winds. Total event duration may exceed 24 hours, with damaging gusts ≥50 MPH sustained for 6 – 12 hours. Wind profiles often contain long plateaus and strong surge onsets. Confirmed double-trough cases include December 11, 1984, February 4, 2020, and January 7–8, 2025. Both the 2020 and 2025 events exhibited significantly longer windstorm durations than other variants. Notably, the first phase of the 2025 event likely included rotor dynamics, while the second phase maintained high vorticity and strong jet support for more than 9 hours. The inclusion of rotor motion, typically associated with stronger events, aided 2025 in becoming an unusually powerful and long-lasting variant.

---

The classification framework developed in this study demonstrates that cold-flavor Santa Ana windstorms can be meaningfully grouped into three synoptic variants—single-trough, double-trough, and rotor-trough—based on 500 mb vorticity structure and jet streak alignment. These variants differ not only in meteorological setup but also in their surface expression, duration, and potential to exacerbate wildfire spread. Analysis of the four modern events using mesonet data shows clear alignment between observed wind profiles and the upper-level dynamics captured by ERA5 reanalysis.

Among these, double-trough events emerge as the most dangerous. Both the February 2020 and January 2025 windstorms were double-trough events, and both exhibited total durations at least twice that of other variants. These storms sustained sub-peak damaging winds for extended periods, with maximum gusts occurring intermittently rather than during a single burst. This behavior suggests a sustained synoptic support for mountain wave activity, which is consistent with the prolonged vorticity advection and dual shortwave trough interactions observed in ERA5. Critically, the January 2025 event occurred before meaningful rainfall. The rare combination of this vigorous and long-lasting Santa Ana windstorm enabled the Eaton and Palisades wildfires to simultaneously evolve into one of the largest urban conflagrations in modern history.

In contrast, single-trough events, such as the January 2022 storm, were shorter in duration and featured wind profiles with sharp, well-defined peaks followed by rapid declines. These were associated with transient jet streak alignments, where the most favorable dynamics for mountain wave generation passed quickly across Southern California. Despite producing locally intense gusts, the brief duration limits their capacity to drive world-record challenging conflagration expansion on their own.

Rotor-trough events, exemplified by March 2024, showed a gradual rise and fall in intensity, forming an arc-shaped wind profile. These storms appear to be triggered by the passage of a horizontal shortwave ejection, creating a wave-like response in mountain wind speeds. Though shorter in duration, rotor events can produce extreme instantaneous gusts and may represent the upper bound for windstorm intensity in the region. Their timing, waveform signature, and vorticity structure are distinct enough to merit their own category.

This study also underscores a critical gap in how fire weather threats are communicated. Current operational practices often emphasize peak wind gusts, yet our results show that event duration and sustained sub-peak winds may be more closely tied to structural fire risk and wildfire propagation. A wildfire occurring during a 20-hour windstorm with 40–50 MPH gusts may be more damaging than wildfire occurring during a a 3-hour storm peaking at 65 MPH. Incorporating variant-based synoptic classifications into fire weather forecasting and public messaging could enable better decision-making, including earlier evacuations and resource deployment.

Finally, the successful retrospective application of this classification framework to 50 years of historical ERA5 data—supported by independent media verification—suggests that this system has both predictive and diagnostic utility. It offers a foundation for more nuanced forecasting and post-event analysis, particularly in data-sparse regions or historical cases where surface observations are unavailable.

---

Future work should apply this classification to additional cold-flavor Santa Ana events and evaluate whether these synoptic signatures can be reliably forecast several days in advance using ensemble models or high-resolution WRF simulations. Future work could also search for additional variants that could be derived from a windstorm not flagged in the ERA5, or one of the given variants in this study that should be separated further.

Abstract

Over one hundred large Bigcone Douglas-Fir (Pseudotsuga macrocarpa) logs are scattered among the desert-like fields of sand, cacti, and boulders in the wash portion of Eaton Canyon on the south face of the San Gabriel Mountains. As no specimens of this species grow within a 2-mile radius, these logs are quite conspicuous. In this study, we will attempt to reconstruct historical flash floods in Eaton using aerial imagery. Then, we will back up that method by comparing results with Bigcone Douglas-Fir log locations. The first step, however, is to prove that these logs were indeed delivered to the wash from higher elevations by major flash floods.

This study finds that major debris flows are in fact responsible for placing these logs in the wash portion of the canyon. However, because of differences in magnitude of each flood as well as changes in topography between floods, this study also finds that not all log locations are the result of the same flood. This study has been able to determine which logs have moved as recently as 2021 and which logs have sat stationary for over half a century. In addition, this study has been able to determine how many acres of Eaton Canyons’ wash were destroyed during each of the most recent large floods.

Introduction

Eaton Canyon experiences rapidly changing climatological and ecological changes between its wide elevation range. Elevations between 4,500′ and 6,100′ in the canyon consist of conifer forests full of Bigcone Douglas-Firs, Canyon Oaks, Coulter Pines, and even the occasional California Incense Cedar. Meanwhile, elevations between 900′ and 3,000′ in Eaton Canyon can be very hot and dry, with well draining soil, chaparral ecosystems, and medium sized trees huddled in canyons. Eaton Canyon is also an exceptionally steep mountain face, with elevations rising several thousand feet in a matter of 3 miles. The 6-square mile canyon averages over 100″ of snow each winter at one end and can experience 6 straight months with an average high >90°F at the other end. Owing to their size, shape, and length, the vast majority of logs in the wash should be the Bigcone Douglas Fir, a resilient conifer native to the region that dominates the conifer forest above. At lower elevations, Eaton Canyon’s only primary native trees, the Western Sycamore, the Coast Live Oak, and the White Alder, do not have any of the characteristics that these logs show.

Methodology

To demonstrate that these logs were transported by flash floods, we designed three specific tests that can be conclusively attributed to such events. These tests are as follows:

1. Log Distribution: If flash flooding was indeed responsible, there should be a higher concentration of logs in wider sections of the wash, with a noticeable scarcity in narrower sections where deeper and faster-moving waters would carry logs away.
2. Log Condition: All logs should be completely devoid of branches, consistent with the forceful removal of limbs during transit via high flow events. 
3. Species-Specific Evidence: Bigcone Douglas-Fir logs should be completely absent in areas not affected by major flooding in past decades, indicating that their presence in historical flood zones can only be due to the floods themselves.

Results and Discussion

This study has successfully reconstructed past flooding events using both aerial imagery and Bigcone Douglas-Fir log locations. We find that the 1969, 1980, and 2005 floods destroyed 43.0, 48.5, and 31.6 acres of Eaton Canyon’s wash respectively. This shows that although the 1969 flood gets lots of attention due to its destruction of the El Dorado Inn, the 1980 flood was actually larger and more impactful for the local ecosystem. The mapping of these debris flows also shows that just because one particular area was left untouched by a large flood does not mean it is safe from a less severe one. The team found that while the logs are fairly well distributed along the 2-mile long riverbed, there is a downward trend the further one gets away from the canyon mouth. Additionally, these maps show that the lower section of Eaton Wash beside the Eaton Canyon Nature Center has been drifting westward toward the parking lot with each successive flow. Most recently in 2005, the north end of main parking lot was nearly destroyed after fast moving flood water begin quickly eroding the embankment separating the parking lot from the wash, something the 1969 and 1980 floods couldn’t do. Local claims that the upper parking lot was destroying during one of the floods could not be confirmed using satellite imagery. Readers are encouraged to submit evidence should these claims be correct.

Future Work

Field biologists may want to investigate why these logs appear to show little to no signs of decay while sitting out in the desert. It is plausible that the necessary microbiology, including fungi and bacteria, needed to break down the logs does not exist in the hot, dry ecosystem in which these logs have been transported to. As a consequence, even the three logs identified to have not moved since 1969 (55 years since the writing of this study) show little sign of decay. (Figure 10)

An analysis of future flood potential using 3D-imagery, climate data, and historical risk should be conducted, especially in regards to the danger posed to Los Angeles County Parks and Recreation amenities such as the Nature Center by a quickly-changing creek route during extreme flow events.

Keywords: Bigcone Douglas Fir, flash flooding, Eaton Canyon, geomorphology, hydrology, sediment transport, arid environment, log preservation, heavy rainfall, debris flow, compartmentalization, desert, alluvial fan, scarp.

Charts and Figures

Figure 3: A comparison of the canyon wash between the summers of 2004 and 2005 shows a heavy loss of vegetation as well as a change in the creek route. Note that the wash encroached on the upper parking lot, along with the loss of a large oak that once served to reduce erosion.