Weather Stations
—
Where the stations are
My two stations, the two planned sites, and the public stations and tide gauges used alongside them. Pick a region on the world map, or click a station to show its data.
—
Daily temperature
Shaded band runs from each day's low to its high; the line is the daily average.
Rain by month
Total of the station's daily rain. Months at either end of the record are partial.
Last 14 days, hour by hour
Hourly average temperature with hourly rain below.
—
Average daily high, year by year
Each point is one year's average of the daily highs. Dashed line: ERA5 straight-line trend. Shading: El Niño and La Niña periods (NOAA).
Average nightly low, year by year
Each point is one year's average of the daily lows. Dashed line: ERA5 straight-line trend. Shading: El Niño and La Niña periods (NOAA).
Rain per year
Bars: ERA5 for this location. Lines: nearby rain gauges where they have complete years.
The year then and now
Average daily high and low for each month, 1991–2000 compared with 2016–2025 (ERA5).
Every source for this site
Trends are straight lines through complete years. The last-10-years column is included for interest, but ten years is short: one El Niño year at either end moves it a lot, so read it alongside the full record.
| Source | Years from | High /decade | High /decade, last 10 yrs | Low /decade | Low /decade, last 10 yrs | Rain/yr 1991–2000 | Rain/yr 2016–25 |
|---|
Monthly average temperature, all stations
Average of the daily means for each month at each station. Playa Grande is the Palm Beach Inn station and Palo Seco the Quepos station until stations are installed there.
Monthly rain, all stations
Sum of daily rain for each month.
Monthly average humidity
Average relative humidity, %.
Playa Grande only. Water levels are for the Flamingo Marina tide gauge, about 20 km north, measured since September 2025 and rebuilt for 2015–2025 from the Quepos gauge (see About the data). Heights are relative to the average sea level of 2016–2025, not to a chart datum.
—
High and low water, every day
The band runs from each day's lowest to highest water. Dashed line: the level reached on only 5 % of days in 2016–2025, used here as "unusually high".
Unusually high tides per year
Days whose high water passed the dashed line above, compared with the days the astronomical tide alone would have passed it. The gap is sea level. 2015 starts in June, 2021 is mostly missing at the gauge, and 2026 runs to early October.
Sea level compared with normal, by month
Monthly average water level after removing the tide. It rises during El Niño, as warm water piles up along the Pacific coast of Central America.
Waves offshore, daily maximum
Significant wave height from ERA5, about 25 km offshore. Dashed line: the size reached on 10 % of days. Orange dots: days with both an unusually high tide and big waves, when erosion is most likely.
Average daily tide range, by year
The difference between each day's high and low water, averaged over the year. It rises and falls over the moon's 18.6-year cycle and is near a low point now.
The stations
- Nonosi, San Antonio de Escazú, about 1,410 m. Ambient Weather WS-2902, called Spaceishuge-1 in the app. 5-minute readings from 25 Dec 2024.
- Sun Mesa, Yucca Valley near Joshua Tree, about 1,000 m. WS-2902. Readings from 31 Dec 2025, the start of its history on Ambient Weather. Before that, daily values are estimated from NOAA (below).
- Palo Seco has no station of its own yet either. Its daily data comes from the public Quepos Ambient station (in town, 26 m, about 18 km southeast), from July 2024. For 1990 onward, ERA5 at Palo Seco is adjusted to that station using the months both cover: highs raised by about 4 °C (ERA5's grid square averages in cooler sea and hillside), lows almost unchanged, and rain multiplied by 0.54. Raw ERA5 is about 1.8 times too wet here because its square includes the mountains behind Parrita. NASA POWER's rainfall for this spot climbs from about 1,100 mm to 4,200 mm a year over the record, a fault in that dataset rather than real change, so it is not used.
- Playa Grande has no station of its own yet. Until it does, the page uses the public Palm Beach Inn station (Palm Beach Estates, about 12 m) for the same period.
Filling Nonosi's gaps
Nonosi's outdoor sensor dropped out on and off between May and August 2026. Missing hours are filled from a neighbouring station, adjusted for the usual difference between the two at that hour of the day and season:
- Bebedero, 0.65 km away at 1,467 m, used first (from July 2026). Typical error after adjustment: about 0.8 °F.
- Lg weather station, Escazú, 3 km away and about 300 m lower, used otherwise. Typical error after adjustment: about 1 °F.
The same method replaces readings that are present but wrong. From 12 to 28 June and on 21–22 August 2026, Nonosi read 4–8 °F too warm and 10–15 points too dry compared with its neighbour and with ERA5, the signature of a sensor heating in the sun. On those 19 days its temperature and humidity are replaced from the neighbour, and its own rain gauge readings are kept. The check runs on every update, and the list is saved in qc_excluded_days.csv.
Rain in filled hours is the neighbour's rain, unadjusted. On the daily chart, filled days are marked along the bottom. The original readings are kept untouched in the raw files.
Extending Sun Mesa back to Dec 2024
For 25 Dec 2024 to 31 Dec 2025, Sun Mesa's daily high, low and rain come from the NOAA Yucca Mesa volunteer station, about 11 km away at 1,050 m. NOAA's reading is taken in the morning, so its high is shifted back one day to the afternoon it happened. Each month's values are then adjusted by the typical difference between the two stations: Sun Mesa's nights run about 4 °F warmer, and its days within about 1 °F. Checked against 2026, when both stations ran, the estimates are typically within 1.2 °F for highs and 1.9 °F for lows. Rain is NOAA's as recorded; in 2026 the two totals were 1.62 in and 1.57 in. These days carry no humidity, wind or hourly data, and they are excluded from the hottest and coldest records.
Long-term records
- ERA5 / ERA5-Land (ECMWF reanalysis, via Open-Meteo). A reconstruction of past weather built from observations, on a grid of about 9 km for temperature and 28 km for rain. Same method every year, so it is the main source for trends. Absolute values can be off at a single spot; changes over time are more reliable.
- ERA5 rain at Nonosi runs high. Its rain grid square is about 28 km across and centred about 16 km northwest of Nonosi, so it averages wetter and drier ground together. In 2020–2025 it came out 17–30 % above the median of nearby Ambient home stations (2025: about 2,370 mm against about 1,820 mm). The year-to-year pattern matches well, so its percentage changes are usable, but its totals are not. Home tipping-bucket gauges also tend to undercount in heavy downpours, so the true figure probably lies a little above the gauges.
- NASA POWER (MERRA-2): a second, coarser reanalysis (about 50 km), shown for temperature. Its grid cell sits far lower than Nonosi, so its values run hot there. Its Costa Rica rainfall has a jump that doesn't match any station, so it is left out.
- NOAA GHCN-Daily stations: Juan Santamaría airport for Nonosi; Liberia airport for Playa Grande (45 km inland, so hotter and drier); Yucca Mesa, Joshua Tree and Twentynine Palms for Sun Mesa. The Costa Rica airports report rain on too few days for yearly totals, so only their temperatures are used. Station moves and instrument changes can create jumps.
Coast and tides (Playa Grande)
- Flamingo Marina tide gauge (Costa Rica's national network, via UNESCO's sea-level monitoring service), about 20 km north of Playa Grande. Radar sensor, readings since September 2025, averaged to hourly.
- Quepos tide gauge (University of Hawaii Sea Level Center), hourly since May 2015. Its tide is about 12 % larger than Flamingo's and arrives about half an hour earlier; hour by hour the two gauges correlate at 0.97.
- How the 10 years are rebuilt. A harmonic tide analysis (UTide) splits each gauge into the astronomical tide, driven by the moon and sun, and everything else: seasonal and El Niño sea-level changes, storm surge, and long-term rise. Flamingo's own year gives its astronomical tide, which is predicted back to 2015. Quepos supplies the "everything else", which is shared along this coast. Tested on Flamingo's measured year, the rebuilt levels are typically within 9 cm.
- Waves: ERA5 ocean wave reanalysis (via Open-Meteo), significant wave height for the grid point at 10.5° N, 86.0° W, about 25 km offshore.
- An 11-year record is too short for a reliable sea-level-rise rate. After allowing for El Niño, the trend is about 4 mm per year, in line with the global average, but uncertain by several millimetres either way.
El Niño and La Niña shading
Shaded periods come from NOAA's Oceanic Niño Index (ONI), the three-month average sea-surface temperature anomaly in the central Pacific. NOAA calls a period El Niño when the index stays at +0.5 °C or above for at least five overlapping three-month seasons, and La Niña at −0.5 °C or below. On the yearly rain chart, a year is shaded when at least six of its months fall in one phase. Because the index is a three-month average, its latest value is centred about six weeks back. When an episode is still running at the end of the record, the lighter shading after it, ending in a dashed line, carries it up to today and is labelled ongoing. In Costa Rica, El Niño usually means a drier, hotter Pacific side, including Guanacaste and the Central Valley, with a stronger mid-year dry spell (veranillo). In the California desert its effect is mostly on winter rain and is less consistent. The El Niño that began in May 2026 is still developing and is already very strong, which helps explain this year's warm months.
Keeping it up to date
New readings are downloaded from Ambient Weather Network and the reference datasets, processed with the same steps, and the data files behind this page are replaced. The page itself does not change.