For about 45 days from Christmas Eve 1861, storm after storm hit California. San Francisco got 24.36 in of rain in January 1862, a month never matched since, and the Central Valley went under water for 250 to 300 miles. Below, the 1862 reports are pinned to today's state, with the flood an 1862-size storm would cause now and what a warmer ARkStorm adds.
A scenario explorer for history and planning, not a forecast or an official flood map. For a property's flood risk, see FEMA's Flood Map Service Center and Cal OES MyHazards.
Most of these are estimates made at the time, amid cut telegraph lines and impassable roads. Each one names its source.
No reliable death toll exists. California had about 380,000 people in the 1860 census and has about 39 million now.
California's 28 largest flood rivers, with what each did in 1862 and how much land the model floods along it. Land is counted for river and lake flooding within about 7 miles of each river. Select Map to see a river.
| River | In 1862 | Land flooded along it, sq mi | ARkStorm 2.0: late century vs today | |
|---|---|---|---|---|
| Working out each river's flooding… | ||||
Lines: USGS NHDPlus main stems, via the NLDI service. Runoff changes: Huang and Swain 2022, peak runoff by subregion. Flooded land: this site's model.
Every California levee system in the U.S. Army Corps of Engineers' National Levee Database, with the people and property behind it and the flood it is rated to hold back. On the map, levees are colored by protection level; zoom in for the figures, or click a protected area.
| Levee system | Protection | People | Property | FEMA status | Model flooding inside, sq mi 1862 · late century | |
|---|---|---|---|---|---|---|
| Loading… | ||||||
Source: USACE National Levee Database (public leveed areas, levee alignments and system records), downloaded October 2026. The 30 systems with the most people are listed; every system is on the map.
ARkStorm 1.0 expected California's dams to hold, with "minor spillway damage" at worst. Then Oroville's spillways failed in a far smaller storm in 2017. These are the big flood-control reservoirs and the dams the Army Corps and state inspectors flag.
| Dam | Risk or condition | Height, ft | Storage, acre-ft | Built | Why it matters |
|---|
Sources: USACE National Inventory of Dams (records and Corps risk summaries, October 2026); Masters, Yale Climate Connections, Jan 2023; Porter et al. 2011. Corps risk classes (DSAC) run from 1, very high urgency, to 5, normal; state-regulated dams carry a condition rating instead. Federal Bureau of Reclamation dams publish neither.
From the first rain in November 1861 to the raised streets of the 1870s. Select Map on any line to fly to the place.
Sediment in the ocean floor, marshes, lakes and canyons keeps a record of floods long before anyone wrote them down. Six California records from the last 2,000 years.
San Francisco has the only continuous gauge record that spans the flood: Thomas Tennent's readings, carried on in today's downtown series. Everywhere else the numbers come from letters, newspapers and later estimates, graded below.
| Season | Rain, in | January, in |
|---|
| Place | Rain | Period | Grade | Notes | Source |
|---|
From Brewer's letters to sediment cores and climate-model storms. Each entry says what that study adds.
The Great Flood is history. ARkStorm 1.0 is the USGS planning scenario built to rival it. ARkStorm 2.0 runs the same kind of month-long storm sequence through a climate model, once in today's climate and once in a much warmer one.
Working out the scenarios on this map…
| Great Flood, 1861–62 | ARkStorm 1.0 (USGS, 2010–11) | ARkStorm 2.0: ARkHist, today's climate | ARkStorm 2.0: ARkFuture, 2070s | |
|---|---|---|---|---|
| What it is | A real winter, known from letters, newspapers, one city gauge and sediment | A planning scenario worked through by 117 scientists, engineers and planners | A storm sequence from a climate model run for about 1996–2005, re-run at 3 km | The same, for 2071–80 under high emissions (RCP8.5) |
| Length | About 45 days of storms | 23 days | 30 days | 30 days |
| Built from | — | The January 1969 and February 1986 storms, run back to back | The wettest 30-day spells in a 40-member climate-model ensemble (CESM1-LENS), downscaled with the WRF weather model | |
| Rain | 24.36 in in January at San Francisco. Over 2,500 mm (100 in) reported at Sierra sites in about 40 days | Up to 10 ft in places. Many places past their 500- to 1,000-year totals | Over 500 mm (20 in) widely and 1,000 mm (40 in) in the Sierra. Peak about 2,150 mm (85 in) | About 45% more statewide, up to 80% more locally. Peak about 3,200 mm (126 in) |
| Snow line | Warm rain melted snow up to about 8,000 ft | — | Freezing level about 1,940 m (6,360 ft) | About 2,230 m (7,320 ft). Rain instead of snow from 1,200 to 2,000 m (3,900–6,600 ft) |
| Runoff | Santa Ana River about 318,000 cfs, three times its 1938 peak | Overwhelms a flood system built for 100- to 200-year floods | Baseline. Runoff is 19% of the rain | Peak runoff more than doubles statewide, up 200–400% from the Sierra. 29% of the rain runs off |
| Pacific backdrop | Not an El Niño winter | — | Weak El Niño (+0.56 °C). All eight of the model's biggest month-long storms came in El Niño years | Strong El Niño (+1.48 °C) |
| Wind | Gales blew down barns | 60 mph widely, up to 125 mph in places | — | |
| Losses | No death count. A quarter of the cattle. A third of the tax roll | About $725 billion: nearly $400 billion of property, $325 billion of business interruption. Nearly $1 trillion in 2022 dollars. 1.5 million people evacuated | Not yet estimated. Huang and Swain's paper is the first part of ARkStorm 2.0 | |
| How often | Floods this size: 5 to 7 times a millennium | About as likely in a year as a severe Southern California earthquake | About a 90- to 100-year event in the 1995–2005 climate, already about twice as likely as a century ago | Longer than 400 years in that climate. By about 2060 on high emissions, a formerly 200-year event about 3 times a century |
| On this map | The 1862, as reported preset (150%), sized to Brewer's 5,000–6,000 sq mi of valley under water | The ARkStorm 1.0 preset (100%) | The same 100% | The late century preset (145%) |
Sources: Porter et al. 2011 (USGS Open-File Report 2010-1312); Huang and Swain 2022; Dettinger and Ingram 2013; Dettinger et al. 2012; Brewer; NWS.
For the scenario set on the map. Change the storm or the tide and this table follows.
| County | Flooded, sq mi | Share of county land | Mostly from |
|---|
The flood model is the same one behind the earlier California Megaflood Map. Only the basemap and the reports are new.