Load flight data

Create Flight data structure for working with flight trajectories.

import pandas as pd
import numpy as np

from pycontrails import Flight
from pycontrails.datalib.ecmwf import ERA5

Create Flight instance

From Numpy Arrays

# waypoints
longitude = np.linspace(0, 50, 100)
latitude = np.linspace(0, 10, 100)
altitude = np.linspace(11000, 11500, 100)
time = pd.date_range("2022-03-01 00:00:00", "2022-03-01 02:00:00", periods=100)

fl = Flight(longitude=longitude, latitude=latitude, altitude=altitude, time=time, flight_id="id")
fl

Flight [4 keys x 100 length, 2 attributes]

Attributes
time	[2022-03-01 00:00:00, 2022-03-01 02:00:00]
longitude	[0.0, 50.0]
latitude	[0.0, 10.0]
altitude	[11000.0, 11500.0]
flight_id	id
crs	EPSG:4326

	longitude	latitude	time	altitude
0	0.000000	0.00000	2022-03-01 00:00:00.000000000	11000.000000
1	0.505051	0.10101	2022-03-01 00:01:12.727272727	11005.050505
2	1.010101	0.20202	2022-03-01 00:02:25.454545454	11010.101010
3	1.515152	0.30303	2022-03-01 00:03:38.181818181	11015.151515
4	2.020202	0.40404	2022-03-01 00:04:50.909090909	11020.202020
...	...	...	...	...
95	47.979798	9.59596	2022-03-01 01:55:09.090909090	11479.797980
96	48.484848	9.69697	2022-03-01 01:56:21.818181818	11484.848485
97	48.989899	9.79798	2022-03-01 01:57:34.545454545	11489.898990
98	49.494949	9.89899	2022-03-01 01:58:47.272727272	11494.949495
99	50.000000	10.00000	2022-03-01 02:00:00.000000000	11500.000000

100 rows × 4 columns

From Pandas DataFrame

# Example flight
df = pd.DataFrame()
df["longitude"] = np.linspace(0, 50, 100)
df["latitude"] = np.linspace(0, 10, 100)
df["altitude"] = 11000
df["time"] = pd.date_range("2022-03-01 00:00:00", "2022-03-01 02:00:00", periods=100)
fl = Flight(data=df, flight_id="ABC")
fl

Flight [4 keys x 100 length, 2 attributes]

Attributes
time	[2022-03-01 00:00:00, 2022-03-01 02:00:00]
longitude	[0.0, 50.0]
latitude	[0.0, 10.0]
altitude	[11000.0, 11000.0]
flight_id	ABC
crs	EPSG:4326

	longitude	latitude	altitude	time
0	0.000000	0.00000	11000.0	2022-03-01 00:00:00.000000000
1	0.505051	0.10101	11000.0	2022-03-01 00:01:12.727272727
2	1.010101	0.20202	11000.0	2022-03-01 00:02:25.454545454
3	1.515152	0.30303	11000.0	2022-03-01 00:03:38.181818181
4	2.020202	0.40404	11000.0	2022-03-01 00:04:50.909090909
...	...	...	...	...
95	47.979798	9.59596	11000.0	2022-03-01 01:55:09.090909090
96	48.484848	9.69697	11000.0	2022-03-01 01:56:21.818181818
97	48.989899	9.79798	11000.0	2022-03-01 01:57:34.545454545
98	49.494949	9.89899	11000.0	2022-03-01 01:58:47.272727272
99	50.000000	10.00000	11000.0	2022-03-01 02:00:00.000000000

100 rows × 4 columns

Create Flight without Waypoints

# Example flight
attrs = dict(flight_id="1234", equip="A532")
fl = Flight.create_empty(attrs=attrs)
fl

Flight [4 keys x 0 length, 3 attributes]

Attributes
flight_id	1234
equip	A532
crs	EPSG:4326

	longitude	latitude	time	altitude

Create from CSV file

# load flight
df = pd.read_csv("data/flight.csv")
fl = Flight(data=df, flight_id="csv")
fl

Flight [4 keys x 175 length, 2 attributes]

Attributes
time	[2022-03-01 00:50:00, 2022-03-01 03:47:00]
longitude	[-97.026, -77.036]
latitude	[32.931, 38.854]
altitude	[190.5, 11582.4]
flight_id	csv
crs	EPSG:4326

	longitude	latitude	altitude	time
0	-77.035950	38.829315	236.22	2022-03-01 00:50:00
1	-77.038223	38.772675	708.66	2022-03-01 00:51:00
2	-77.114231	38.744568	9471.66	2022-03-01 00:52:00
3	-77.201965	38.739888	2019.30	2022-03-01 00:53:00
4	-77.286191	38.745117	3032.76	2022-03-01 00:54:00
...	...	...	...	...
170	-97.025925	32.931379	190.50	2022-03-01 03:43:00
171	-97.025922	32.930649	190.50	2022-03-01 03:44:00
172	-97.025922	32.930649	190.50	2022-03-01 03:45:00
173	-97.025922	32.930649	190.50	2022-03-01 03:46:00
174	-97.025922	32.930649	190.50	2022-03-01 03:47:00

175 rows × 4 columns

Using the Flight instance

The flight.data attribute is a dictionary with np.ndarray values

# waypoints
longitude = np.linspace(0, 50, 10)
latitude = np.linspace(0, 10, 10)
altitude = np.linspace(11000, 11500, 10)
time = pd.date_range("2022-03-01 00:00:00", "2022-03-01 02:00:00", periods=10)
attrs = {"flight_id": "ABC123"}
fl = Flight(longitude=longitude, latitude=latitude, altitude=altitude, time=time, attrs=attrs)

fl.data.keys()

dict_keys(['longitude', 'latitude', 'time', 'altitude'])

Flight attributes are stored in a dictionary on the attrs attribute. The crs attribute is always added by default, if not specified.

fl.attrs

{'flight_id': 'ABC123', 'crs': 'EPSG:4326'}

Data can be set / get from the Flight like a dictionary

# get
lat = fl["latitude"]

# set
lat[5] = 20
fl["latitude"] = lat

# get updated
fl["latitude"][5]

20.0

The Flight class contains the following convenience properties

# Pressure altitude, in hPa
fl.level

array([226.3170091 , 224.34300442, 222.3862176 , 220.44649846,
       218.52369813, 216.61766904, 214.7282649 , 212.85534072,
       210.99875274, 209.15835847])

# Altitude, in ft
fl.altitude_ft

array([36089.23884514, 36271.5077282 , 36453.77661126, 36636.04549431,
       36818.31437737, 37000.58326043, 37182.85214348, 37365.12102654,
       37547.38990959, 37729.65879265])

# Values that are constant along the flight path

# set constant value along flight waypoints
fl["constant"] = np.full(shape=fl.shape, fill_value=100)

fl.constants

{'constant': 100, 'flight_id': 'ABC123', 'crs': 'EPSG:4326'}

# Flight distance, in meters
fl.length

7818835.115366629

# Time start/end
print(fl.time_start)
print(fl.time_end)

2022-03-01 00:00:00
2022-03-01 02:00:00

# Flight duration, as a pandas Timedelta
fl.duration

Timedelta('0 days 02:00:00')

# Max time gap between waypoints, as a pandas Timedelta
fl.max_time_gap

Timedelta('0 days 00:13:20')

# Max distance gap between waypoints, in meters
fl.max_distance_gap

1831403.3492360476

Intersect with Met data

# waypoints
longitude = np.linspace(0, 50, 50)
latitude = np.linspace(0, 10, 50)
altitude = np.linspace(11000, 11500, 50)
time = pd.date_range("2022-03-01 00:00:00", "2022-03-01 02:00:00", periods=50)
fl = Flight(longitude=longitude, latitude=latitude, altitude=altitude, time=time, flight_id="ABC")

# domain
time = ("2022-03-01 00:00:00", "2022-03-01 03:00:00")
variables = ["t", "q", "u", "v", "w", "ciwc", "z", "cc"]
pressure_levels = [300, 250, 200]

# get met data
era5 = ERA5(time=time, variables=variables, pressure_levels=pressure_levels)
met = era5.open_metdataset()

# interpolate to nearest grid member
fl.intersect_met(met["air_temperature"], method="nearest")

array([231.6297 , 231.5037 , 231.73001, 231.08833, 218.8252 , 218.8393 ,
       219.00676, 218.07823, 218.66353, 219.61859, 219.48346, 219.66501,
       219.8988 , 219.86067, 219.95186, 220.16989, 220.15166, 220.06793,
       219.71309, 219.5191 , 219.5647 , 219.54646, 219.39972, 219.1908 ,
       219.0192 , 219.02084, 219.36572, 219.42708, 219.39308, 219.49672,
       219.63931, 219.63351, 219.80347, 219.72304, 218.88406, 218.68758,
       219.5904 , 219.66667, 219.75786, 219.57713, 219.53734, 219.87642,
       219.74045, 219.47931, 219.345  , 219.28697, 219.14354, 219.04074,
       218.90147, 218.93463], dtype=float32)

# linear interpolation
fl.intersect_met(met["air_temperature"], method="linear")

array([225.77794, 225.39075, 225.21887, 224.88387, 225.13017, 224.75333,
       224.69849, 224.51218, 224.65642, 225.06851, 225.2779 , 225.23737,
       225.27412, 225.21637, 225.02275, 225.16205, 225.02489, 224.83875,
       224.50832, 224.36188, 224.2705 , 224.0278 , 223.71686, 223.553  ,
       223.37677, 223.35237, 223.489  , 223.45067, 223.46031, 223.51036,
       223.47456, 223.33977, 223.33707, 223.19572, 222.68132, 222.43277,
       222.99377, 222.90991, 222.94348, 222.70607, 222.66989, 222.84317,
       222.59155, 222.31973, 222.06894, 221.91066, 221.70157, 221.5247 ,
       221.33913, 221.26541], dtype=float32)

Get Lengths

# total flight length in meters
fl.length

5642421.5973290345

# intersect flight with air temperature
fl["temp"] = fl.intersect_met(met["air_temperature"], method="nearest")

# get the length of the flight where ambient temperature is > 226 K
fl.length_met("temp", threshold=226)

462850.7958842697

Plot and Resample

fl.dataframe.plot.scatter(x="longitude", y="latitude", figsize=(12, 8))
fl.max_distance_gap

115715.16936791067

# resample with 10 minute waypoints
fl = fl.resample_and_fill("10min")

fl.dataframe.plot.scatter(x="longitude", y="latitude", figsize=(12, 8))
fl.max_distance_gap

472488.41576537804

# resample with 10 second waypoints
fl = fl.resample_and_fill("10s")

fl.dataframe.plot.scatter(x="longitude", y="latitude", figsize=(12, 8))
fl.max_distance_gap

7874.969421029368

Clean and smooth a noisy flight with gaps

df = pd.read_csv("data/flight-noisy.csv")
df["time"] = df["time"].map(lambda t: np.datetime64(t))
f = Flight(df, drop_duplicated_times=True)
_ = f.dataframe.plot.scatter(x="time", y="altitude_ft")

f = f.clean_and_resample(nominal_rocd=20)
_ = f.dataframe.plot.scatter(x="time", y="altitude")