Data. NCEP/NCAR Reanalysis 1 sea-level pressure, 6-hourly on a 2.5° grid. Lows are tracked every 6 hours, separately in each basin. A storm belongs to the storm year in which its lowest pressure occurs. Storm IDs (such as NP-1997-012) are the script's, numbered within the calendar year of the lowest pressure.
Near Antarctica. The Southern Ocean has a belt of low pressure around 60–65°S that averages about 985 hPa, so deep lows are routine there. Pressures close to Antarctica are also less reliable, and the count of lows there roughly doubles from the 1950s to today, which is at least partly a data artifact. For these reasons the default area stops at 60°S. Set the poleward latitude to 70 to include them.
Wave energy on the coast. When this is on, every 6-hour wave fetch of the storms that match your criteria is treated as a wave source. Its sea is estimated from the fetch length and wind speed with the standard fetch-limited growth formulas (JONSWAP, capped at a fully developed sea), which give wave height and period and so the wave energy sent downwind in 6 hours. The periods are 12% shorter than the standard formula gives, because that matches the swell periods and arrival times measured at 18 NOAA buoys (without it the model's periods ran about 2 seconds long and its swell arrived a few hours early). That energy spreads out from the fetch along great circles, mostly in the direction the wind blows (a cos² spread up to 90° either side). On the way it thins out with distance (less so within about 1,000 km, or 540 nautical miles, because a real fetch is hundreds of kilometers wide), fades only slowly (about 63% over 16,000 km, or 8,600 nautical miles, as long-period swell does), and is blocked if land lies in its path (checked every 25 km, about 13 nautical miles, so island chains such as the Aleutians and Kurils block swell except through their wider passes). Each coast segment (about 32 nautical miles long) collects the energy that reaches it, reduced by the angle it hits the coast at, and the map shows the total for the storm year per foot (or meter) of coastline. The color scale runs from the 25th to the 75th percentile of all coast segments in all years with the current criteria, so colors mean the same thing in every year. Colors are spaced evenly on a log scale across that range, because it spans roughly a factor of 40. The model's overall strength, its near-storm spreading and its swell decay were calibrated against 11 NOAA buoys (San Francisco, Bodega Bay, Half Moon Bay, Oregon, Gulf of Alaska, three around Hawaii, two off Southern California and the Stratus mooring off northern Chile and Peru) using their measured long-period swell from 2010 to 2024. Refraction, shallow water and islands too small for the map are still not modeled, so treat coast values as estimates for comparing places, years and storms. In the Whole Pacific map the same calculation covers every Pacific coast, so you can see South Pacific storms reaching North Pacific coasts and the other way round (energy is followed up to 12,000 km, about 6,500 nautical miles). Selecting a storm turns on Wave energy on the coast and shows only the energy from that storm (clearing the selection puts the setting back as it was); with the same-scale setting its colors use the 25th to 75th percentile of all single storms, so storms can be compared with each other. The other scale setting stretches the colors linearly from the lowest to the highest value currently on the coast, so the hardest-hit stretch of coast stands out within one year or one storm. When you press Play or drag the time slider, the coast starts at zero and fills in as the energy arrives. Each 6-hour pulse is split into seven period bands following a standard wave spectrum, and each band travels at its own speed (about 1,000 km, or 540 nautical miles, a day for 16-second waves, faster for longer periods). So a swell arrives the way real swell does: long-period forerunners first, then the bulk of the energy, with the period falling as it goes on, and distant coasts light up days after the storm passes. The peak swell in the coast tooltip and the Swell finder is the 6 hours with the most energy; the period shown is the energy-weighted period at that time. Energy still on its way when the slider reaches its end appears in the whole-year view. The swell angle in the coast tooltip is the direction the swell comes from at its peak (energy-weighted) and how far that is from hitting the coast straight on.
Swell field. The Swell field setting (on by default, showing swell height) fills the ocean with the same wave model while the year plays or you drag the time slider: Swell height shows how big the swell passing each point is (deep water, before it reaches a coast; smoothed over about 12 hours either side, so separate 6-hour pulses blend together), and Swell period draws the swell instead as fronts coloured by period (purple for long-period forerunners, yellow for short-period swell), with no heat map: each 6-hour fetch sends out arcs that grow at the speed of their swell, span the directions its energy travels and stop where land is in the way; thicker, stronger arcs carry bigger swell, and arcs under about 1 ft are not drawn. With one storm selected each fetch draws seven arcs, one per period band, so you can watch the long periods run ahead and the short periods fall behind; with no storm selected each fetch draws only its peak-period arc (every 12 hours). The height map is calculated on a 1° grid for a selected storm and a 2° grid for all the storms shown, lightly smoothed, and blank behind land (islands and coasts cast swell shadows) and where the swell is under about 1 ft. Each point shows the swell from catalogued storms only: no local wind sea, and no refraction or shallow-water effects. The field near a storm is the swell leaving its fetch, not the much bigger seas inside it.
Swell finder. The Swell finder tab answers "which storms were sending swell here on this date?" for any Pacific coast. It uses every catalogued storm in both hemispheres (lowest pressure 1000 hPa or below), not just the ones matching the explorer's criteria, and looks back up to three weeks for swell still on its way. Each storm's share is its part of the swell energy arriving in the chosen window; the heights are the same calibrated estimates as on the map, and From is the direction the swell comes from (hover it for the angle to the coast). After each search the finder map zooms to show the chosen coast and every storm found (untick Zoom to storms found to keep your own view). The finder map has the same display options as the explorer; with fetch arrows on, solid arrows are the fetches whose swell reached the chosen coast in the window and faint ones did not. Below the map, the day-by-day chart shows every storm whose swell reaches the chosen coast in the days around the chosen date (± 3 days by default, centred on the middle of the search window, which is shaded): each storm's swell height and period every 6 hours, the combined swell of all of them (dashed), and the morning wind for each day (when coastal wind data are loaded; 1979 on). Storms keep the same color as in the table; past eight colors, smaller storms are drawn in grey. Each new swell typically starts long-period and drops by several seconds over a day or two, as its slower, shorter-period energy catches up. Click a storm in the key or the table to pick out its lines; hover the chart to read every storm at that time. Periods are drawn only where a storm's swell is at least about 0.5 ft (0.15 m).
Swell height and period (estimated). Hovering on the coast also gives a rough estimate of the biggest swell that reached it: from the selected storm, or from all matching storms in the year. Each 6-hour pulse of wave energy is spread over its arrival window, because its longer waves travel faster and arrive first, and the strongest 6 hours are turned back into a significant wave height and peak period. These are offshore, deep-water values, before the waves feel the bottom or bend around the coast. Checked against the buoys above, the biggest swells come out right on average; at individual buoys the model runs between about 25% high and 25% low, a single event is typically within about 20%, periods are usually within 1 to 2 seconds and arrival times within about a day. Only storms that match your criteria are counted, and local wind waves are not included.
Tracks. By default each storm is drawn along its wave fetch: the centre of the largest patch of gale-force wind blowing one way over open water, which is where its swell is made, every 6 hours. That is often several hundred nautical miles from the low-pressure centre, usually on the side where the winds blow toward the coast in question. Points are colored by the sea height that fetch can build (estimated from its wind speed and length with the standard fetch-limited formulas, capped at a fully developed sea), and the red ring marks where the seas peaked (labelled for the 15 storms with the biggest seas in the year shown, 6 on a phone). The fetch track breaks where a storm has no fetch, or where the fetch jumps more than about 650 nautical miles in 6 hours because a different gale patch took over. Choose Low-pressure centre (or Both) under Track to follow the low itself. By default the criteria judge storms the same way: a storm counts when its fetch builds seas of at least the height set under Storm strength (40 ft by default, as the storm script's significant count), with the fetch where its seas peak inside the region. Choose Lowest central pressure under Storm strength to judge storms by how deep the low got instead (970 hPa by default, with its lowest point inside the region); the region menu also offers any of a storm's fetch, or any part of the low's track. The Months filter uses the month of the biggest seas (or of the lowest pressure), and storms are still assigned to the storm year of their lowest pressure, as in the script.
Coastal morning wind. The Swell historian splits each year's swell days by the wind at that coast on the morning of the day, shows below them the share of each kind of wind year by year (with a dashed line for the share of clean, offshore or light, mornings over all years), and can leave out days with a kind of wind; in the Storm explorer, Coastal wind barbs show that morning's wind along the coast while the year plays. The wind comes from the same NCEP/NCAR Reanalysis 1 model as the storms (winds about 40 m above the sea, which match buoy wind speeds closely), on its 2.5° grid (about 150 nautical miles): one reading a day for each ocean grid cell next to land, the 6-hourly time nearest 7 am local solar time, from 1979 on. Each coast piece uses the nearest of these cells. Offshore means the wind blows from the land within 45° of straight out to sea, onshore from the sea within 45° of straight in, cross-shore anything between, and light means under 5 kt from any direction. Speeds are kept in 3-knot steps and directions in 16 compass points. Checked against 22 NOAA buoys from Washington to San Diego and around Hawaii (2010–2024), the model's wind direction is typically within 10–12°, and its morning call (offshore, cross-shore, onshore or light) is right about 70–75% of the time on the West Coast and about 80% around Hawaii, against 55–60% for always guessing the commonest wind. Individual offshore or light mornings are caught only about half the time, because the grid is far too coarse for the land breezes and terrain-shaped winds at a beach; in the Southern California Bight, where islands and mountains shape the wind, the model is barely better than guessing. Treat it as the wind pattern of the day (a northwest onshore pattern, a front's south wind, an offshore event), not the conditions at a particular break. Wind data only covers the two storm boxes (15–75° north and south), so tropical coasts such as southern Mexico, Central America and Indonesia have none, and it ends when the model data last downloaded ends.
Swell angle accuracy. The swell directions in the app (the From angle in the Swell finder and Swell historian, the swell angle in the coast tooltip and the historian's direction histogram) were checked against 18 NOAA buoys that measure wave direction, from 2010 to 2024, at each buoy's biggest swells arriving from one clear direction (about 4,600 events). At deep-water buoys well offshore the model's direction is typically within about 11° of the measured one: within 15° for about 60% of events and within 30° for about 85%, with no consistent offset. Close to the storms, in the Gulf of Alaska, errors are larger (typically 14–16°), because seas there are mixed and the nearby fetches change direction quickly. Where islands or the coastline shelter a buoy, the model is consistently off by 10–17°: south of Hawaii the measured swell comes from 13–17° further clockwise than the model says, and at San Clemente Basin, San Francisco and off northern Chile 10–12° further counter-clockwise, because the model does not bend waves around islands and headlands. A swell's arrival angle depends mostly on where its fetch is, so it is less sensitive to errors in the fetch's own direction than the amount of swell that reaches each coast.
Swell historian. For the chosen coast, every catalogued storm in both hemispheres is run through the same calibrated wave model, and each storm's swell is followed separately, every 6 hours, from 1950 to the end of the data: its height, peak period, the direction it comes from (energy-weighted) and how far it has traveled. A day counts when at least one storm's swell matches all of your settings at some time that day (UTC). In the default setting a swell matches when it carries at least as much wave energy as the height and period you enter (wave power grows with height squared times period), so a smaller swell with a longer period can count, and a bigger one with a shorter period. Tick Only periods in a range to also require a period range (for example 12 s and up, to leave out shorter-period swell however much energy it carries), or tick Fixed height and period limits to set height and period ranges directly instead of energy. The period is the swell's energy-weighted period at the coast in that 6 hours. Swells from two storms arriving at once are judged separately, not added together. Distance traveled at least drops swell that came from closer than that (the energy-weighted distance from the storm's fetches to the coast), which rules out nearby, disorganized weather systems. The chosen direction range is drawn on the map as a shaded slice around the chosen coast (a faint circle when any direction counts); it keeps the same size on screen as you zoom. The histograms describe each matching day by its strongest matching swell. Distance traveled is a rough guide to how organized a swell is: the farther it comes, the more its waves sort themselves by period, so it arrives as cleaner, more regular sets, while swell from a nearby storm is more mixed and often comes with local wind. Heights are offshore, deep-water estimates (typically within about 20%), before shoaling and refraction. Local wind waves and swell from storms outside the catalog (weaker than 1000 hPa, tropical storms that never got that deep, or lows outside the tracked areas) are not included, so the counts are lower than a buoy would see, especially for short-period swell and in summer. Years before about 1979, and the South Pacific in particular, are less reliable. Only years fully covered by the data are shown.
Average fetch direction. Each storm's average fetch direction is the compass direction its fetch winds blow toward (the way its waves travel), averaged over all of its 6-hour fetches and weighted by the wave energy each one sends out. With Filter by average fetch direction on, only storms whose average falls in the chosen range count; the range runs clockwise from the first angle to the second, so 300° to 60° means northwest through north to northeast. The range is drawn as a shaded slice at the centre of the region box.
Measures. Closed-isobar area is the area inside the outermost 2-hPa isobar that surrounds only this low. Peak winds, wind areas and the wave fetch use the NCEP/NCAR reanalysis model's own winds near the sea surface (its lowest level, about 40 m up, on the same 2.5° grid). Checked against satellite sea-surface winds for 23 storms, fetches from these winds came out closer to the satellite ones (in the North Pacific typically about 170 nautical miles apart and within a few degrees of the same direction) than fetches from winds estimated from the pressure pattern. On a 2.5° grid the strongest winds near tight storm centers are still smoothed out. Wave fetch is the largest connected patch of gale-force wind blowing within one 45° direction sector. Gale winds belong to the nearest low within 1,500 km (810 nautical miles), so a storm's fetch is usually on its flank, often several hundred nautical miles from the centre; with Circles show set to wave fetch area, the circle is drawn on the fetch, and a dotted line links it to the centre for the selected storm. The fetch only uses gale winds over open water: grid cells that are more than half land (Natural Earth outlines) are left out, since wind over land makes no waves. Cells that are half or more covered by sea ice that month are left out too (NOAA OISST satellite sea ice from September 1981; earlier months use its 1991–2020 average for the same month, so ice in the 1950s–70s is approximate). In US units distances are nautical miles (1 nmi = 1.852 km), areas square nautical miles and speeds knots. Fetch arrows point the way the wind blows, which is the way its waves travel. Deepening rate is in bergerons: the 24-hour pressure fall scaled to 60° latitude, where 1 or more makes a storm a bomb. Wind categories follow the Ocean Prediction Center warning levels: gale (34–47 kt), storm (48–63 kt) and hurricane force (64 kt or more).
Caveat. Before about 1979 there were few satellite observations over the ocean, so counts and intensities in early years are less reliable. That period is shaded on the bar chart. This is much worse in the Southern Hemisphere, which had very few ship or station reports before satellites.