New York ACTUALLY HAS 12 SEASONS

This web page is based on this meme.

And, while the meme is simple, this web page is... not. It may even be a little over-engineered.

How this Works

My goal with this page to capture, not the astrological or meterological season of New York, but the experiential season of the New Yorker. A creature who lives their lives outside of the traditional American comforts of climate-controlled cars.

For example, if it's a hot in late astrological Winter, the New Yorker's brain knows it's not spring, but is fooled into thinking they can wear a light jacket. Only to be sorely corrected by the city later that night.

This is the New Yorker experiencing Fool's Spring.

Therefore, this web-page updates dynamically based on the day of the year, the hour of the day, and the real-time temperature in NYC.

It uses both live and 53 years of historical data from OpenWeatherMap.org.

Methodology

To start, I mapped the meme's seasons to a two-dimensional grid of astrological sub-season and the relative temperature of that season.

This part is more feeling than science.

Winter

Spring

Summer

Fall

Next, I needed to define these axes.

Sub-Seasons

For the early, mid, and late I divided the seasons by into three roughly equal segments using their astrological definitions.

Ranges are inclusive start and exclusive end.

Historical Temperature Distributions

Next, I built a baseline for NYC weather by day of year and hour of day. I pulled hourly temperature data for NYC from Jan 1 1979 to Jan 1 2023 and computed a median and standard deviation for every hour of every day of the year.

If you're interested in exploring this data, it's available in the repo here.

import pandas as pd
weather_df = pd.read_csv("nyc_weather.csv")
weather_df["dt_datetime"] = pd.to_datetime(weather_df.dt_iso.str.replace(" UTC", ""))
weather_df["day_hour"] = weather_df.dt_datetime.dt.strftime("%j-%H")
weather_df = weather_df.groupby(by="day_hour").aggregate({"feels_like": ["mean", "std"]})
hour_temps_df = df.groupby(by="day_hour").aggregate({"feels_like": ["mean", "std"]})["feels_like"]
import json
hour_temps_json = json.dumps(hour_temps_df.to_dict("index"))

From here, we say that a "hot" day is anything that is 1.5 standard deviations above the mean and a "cold" day is anything that is 1.5 standard deviations below the mean.

Now that we have a baseline for what is "normal" we can dynamically compare the current temperature our baseline and decide if it's hot out by calling OpenWeatherMap.org every time the page loads. We use Time Square for our live temperature reading and compare it to our historical distribution.

June 2023 - Fire Season Update

After the events of June 7th, 2023, I've updated the page to reflect the new reality. When the air quality index is "Moderate" or worse in the summer/spring/fall, the season is now 'Hell's Front Porch'.

Deciding the Season

Using these features we update the arrow using the above logic. If it's late Winter, and we're 1.5 stddevs above the average for this day and hour, we're in 'Fool's Spring'. An explainer to the data we used to make this decision is appended below the header.

This page was made by Devon Peticolas. If you'd like to see the source, fork, or suggest a change please check out its repo on github.