Code
TARGET_SITE_ID = 3Monte Tamaro
1500 masl
Code
from datetime import date, datetime, timedelta, timezone
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
from startleiter.utils import get_engine, get_flights, get_predictions
engine = get_engine()
df_flight = get_flights(engine, TARGET_SITE_ID)
df_pred = get_predictions(engine, TARGET_SITE_ID)Code
today_idx = df_pred.reftime==pd.to_datetime(date.today())
df_pred[today_idx].drop(["id", "reftime", "source_id", "leadtime_days"], axis=1)| validtime | flying_probability | max_altitude_masl | max_distance_km | |
|---|---|---|---|---|
| 25659 | 2024-12-22 | 0.774242 | 4000.0 | 100.0 |
| 25666 | 2024-12-23 | 0.022727 | 1500.0 | 0.0 |
| 25673 | 2024-12-24 | 0.022727 | 1500.0 | 0.0 |
| 25680 | 2024-12-25 | 0.554593 | 1500.0 | 10.0 |
| 25687 | 2024-12-26 | 0.905200 | 3500.0 | 10.0 |
Code
sns.set()
ts = df_flight.set_index("datetime").sort_index()
keep_cols = ["date", "max_altitude_m", "length_km", "airtime_hours"]
ts = ts[keep_cols]
empty_row = pd.Series(name=pd.Timestamp.today(tz="UTC"), data=[None,] * len(keep_cols), index=keep_cols)
ts = pd.concat([ts, empty_row.to_frame().transpose()])
# compute reference statistics
n_years = 6
clim = ts.resample("1D").mean(numeric_only=True)
clim = clim.groupby(clim.index.isocalendar().week)
clim = clim.rolling(window=f"{n_years * 365}D", center=False).mean().reset_index(0, drop=True)
clim = clim.sort_index().rolling("20D", center=True).mean()
clim_counts = ts.date.resample("1D").count()
clim_counts = clim_counts.where(clim_counts > 0)
clim_counts = clim_counts.groupby(clim_counts.index.isocalendar().week)
clim_counts = clim_counts.rolling(window=f"{n_years * 365}D", center=False).mean().reset_index(0, drop=True)
clim_counts = clim_counts.sort_index().rolling("20D", center=True).mean()
end = datetime.now(timezone.utc)
start = end - timedelta(days=90)
clim = clim[clim.index > start]
clim_counts = clim_counts[clim_counts.index > start]
dfs = df_flight[df_flight.datetime > start]
dfg = dfs.groupby(df_flight.date).datetime.count()
fig, axs = plt.subplots(4, sharex=True, figsize=(7, 10))
# reference lines
axs[0].plot(clim_counts.index, clim_counts.values, color="tab:red")
axs[1].plot(clim.index, clim.max_altitude_m, color="tab:red")
axs[2].plot(clim.index, clim.length_km, color="tab:red")
axs[3].plot(clim.index, clim.airtime_hours, color="tab:red")
axs[3].tick_params(axis="x", labelrotation=45)
# predictions
pred = df_pred[df_pred.leadtime_days==0].set_index("validtime").sort_index()
axs[0].twinx().step(pred.index, pred.flying_probability, where="post", color="gray", alpha=0.5)
twinx = plt.gca()
twinx.grid(None)
twinx.set_ylim([0, 1])
twinx.set_ylabel("Flyability []")
axs[1].step(pred.index, pred.max_altitude_masl, where="post", color="gray", alpha=0.5)
axs[2].step(pred.index, pred.max_distance_km, where="post", color="gray", alpha=0.5)
# plot individual flights
ax0 = sns.scatterplot(x=dfg.index, y=dfg.values, marker="x", ax=axs[0])
ax1 = sns.scatterplot(x="datetime", y="max_altitude_m", marker="x", data=dfs, ax=axs[1])
ax2 = sns.scatterplot(x="datetime", y="length_km", marker="x", data=dfs, ax=axs[2])
ax3 = sns.scatterplot(x="datetime", y="airtime_hours", marker="x", data=dfs, ax=axs[3])
ax0.set(ylabel="No. flights per day")
ax1.set(ylabel="Max altitude [m]")
ax2.set(ylabel="Distance [km]")
ax3.set(ylabel="Airtime [h]")
ax0.set_xlim([start, end])
ax1.set_xlim([start, end])
ax2.set_xlim([start, end])
ax3.set_xlim([start, end])
plt.tight_layout()