Segmentation of flights during ORCESTRA

The research flights during ORCESTRA can be divided into segments which are specific time intervals with coherent characteristics during a flight. The coherent characteristics, referred to as "kinds", either refer to the aircraft state during a flight maneuver (e.g. fairly contant heading, roll angle, and altitude on a straight leg as read out from the aircraft position and attitude data), or to colocated measurements (in space and time) with the EarthCARE (EC) satellite or other campaign platforms such as the METEOR research vessel.

The goal of the flight segmentation is to provide a commonly agreed upon, consistent, and easy-to-use subsampling functionality that can be used by reserachers who work with the ORCESTRA measurement data. For instance, all studies that analyse data taken on circular flight paths can use the flight segmentation to select segments of kind circle, and are thereby guaranteed to refer to the same commonly agreed upon subset of the campaign data.

All segments are specified in a yaml-file. While each segment, defined by a given start and end time, is uniquely labelled with a segment_id, any given time during a flight may belong to one or several segments. For instance, segments of kind ec_underpass typically lie within another segment of kind straight_leg. This allows flights to be segmented in multiple ways and at multiple levels of granularity. However, as a convention, it is advised that a time or time window may not belong to more than one segment of the same kind. Apart from the example of overpasses and underpasses, the first version of the flight segmentation will not contain overlapping segments and coarser segments can be added by collaborators lateron as they see fit.

Structure of yaml-files

Every yaml-file begins with a header consisting of the flight's nickname taken from flight reports if available, mission (ORCESTRA), platform (HALO, ATR, METEOR,..), flight_id (e.g. HALO-20240813a), and segments. The latter lists all identified flight segments.

The general structure of a segment within the yaml files looks as follows: Mandatory components:

• segment_id: unique identifier of the segment, constructed as the combination of the flight_id and the last four digits of a hash computed from the start and end time of the segment
• start: start time of the segment in format YYYY-MM-DD HH:MM:SS, e.g. 2024-08-13 14:56:37.
• end: end time of the segment in same format as start time. Optional components:
• kinds: Can be thought of as tags which label coherent characteristics of the flight segment. See a list of all tags currently in use below.
• name: Short (1 - 3 words) qualitative description of the segment, does not have to be unique
• irregularities: lists irregularities such as deviations from envisioned flight track due to deep convection, or circles in which no sondes were dropped. Generally meant to be a free text field for proper explanations, this category may contain some standardized irregularity tags (to be decided on) for automatic checking, and these should be prepended to the explanatory string of the irregularity.
• comments: lists custom comments such as the distance to the exact EC overpass position in the case of an ec_overpass segment

First entries of an example file:

nickname: EarthCARE echo
mission: ORCESTRA
platform: HALO
flight_id: HALO-20240816a
takeoff: 2024-08-16 11:06:34
landing: 2024-08-16 21:59:33
segments:
- kinds:
  - straight_leg
  name: ferry ascent
  segment_id: HALO-20240816a_kd78
  start: 2024-08-16 11:39:05
  end: 2024-08-16 12:06:58
  irregularities: [strange spike in roll angle at 12:01:37]
  comments: []
- kinds:
  - straight_leg
  - ec_track
  name: ec track southwards
  segment_id: HALO-20240816a_0k2e
  start: 2024-08-16 12:06:58
  end: 2024-08-16 12:51:55
  irregularities: []
  comments: []

Note: Time ranges are defined as semi-open intervals, i.e. while the start time is inside the segment, the end time is not. This allows for an unambiguous definition of exactly consecutive segments. In the case of underpass and overpass segments, the start and end times are identical.

Definition of kinds

Kinds may or may not apply/be adopted for all platforms. For version 1 of the segmentation of HALO flights, one or more of the following kinds are assigned to each segment:

circle:

• can be identified by a constant change of aircraft heading, a duration of about 60 min, as well as a roughly constant roll angle of 1.5-3 degrees (positive or negative, depending on turning clockwise or counter-clockwise) for the standard circle with a radius of about 133km. Duration and roll angle change with the chosen circle radius. The smaller ATR circle has a radius of approximately 70km and a larger roll angle of 3-5 degree depending on and varying with the wind speed and direction at the respective altitude.
• in most cases associated with the launch of dropsondes, typically 12 per circle at every 30° heading. In cases where the first/last dropsonde of the circle was launched before/after the roll angle and change in heading match the circle characteristics described above, the start/end time of the circle segment is taken to be the launch_time of the respective dropsonde minus/plus 1 sec.
• Circle dropsondes were sometimes droped before/after the roll angle reached its circle value in which case the launch time of the sonde marks the beginning/end of the circle period (minus/plus a few seconds, so that the sonde is included in the circle segment).
• Some circles were highly irregular due to detours around convective towers. In these cases, an irregularity note need to be added.

ATR_coordination:

• circle that was also flown by the ATR aircraft on the same day

straight_leg:

• can be identified by an approximately constant aircraft heading and a roll angle close to 0°
• short-term deviations of the roll angle can occur due to turbulence and extreme cases should be mentioned as irregularities; strong deviations my also be radar calibration maneuvers in which case they need to be listed as a separate radar_calibration segment.
• while segments in which the aircraft was ascending or descending are often of kind straight_leg, straight_leg segments with roughly constant altitude need to be separated from those with continuous ascent/descent.

ec_track:

• straight leg along the EarthCARE track

ascent:

• straight leg or circle during which the altitude was constantly increased (slight altitude changes of a few meters do not count)

descent:

• straight leg or circle during which the altitude was constantly decreased (slight altitude changes of a few meters do not count)

radar_calibration:

• maneuver typically conducted during straight legs, where the aircraft tilts to a roll angle of first about -20° and then +20°.
• if conducted during a straight leg, the straight leg is split into three flight segments: 1.) straight_leg, 2.) radar_calibration, 3.) straight_leg.
• segments start and end at about 0° roll angle.

ec_underpass:

• defined as the time when the EC satellite is almost exactly above HALO

pace_underpass:

• defined as the time when the PACE satellite is almost exactly above HALO

meteor_overpass:

• defined as the time when HALO flies above or passes close by METEOR

bco_overpass:

• defined as the time when HALO flies above or passes close by the Barbados Cloud Observatory (BCO)

mindelo_overpass:

• defined as the time when HALO flies above or passes close by the Mindelo research station

Workflow for HALO segmentation - general overview

The flight segmentation workflow broadly consists of three phases:

1. Segmentation of a flight
2. Review of the segmentation in pairs (the person who performed the segmentation + one other colleague)
3. Handover of the final flight segmentation (along with suggestions for how to improve the flight report where necessary) to the corresponding flight-PI for cross-checking.

For developers

The following workflow for generating the flight segmentation YAML files is suggested:

1. Install the requirements noted here as well as the IPFS Desktop App, e.g. on Mac via brew install --cask ipfs.
2. Build a respective python environment, e.g. mamba env create -f environment.yaml
3. Assign yourself to an open segmentation issue of this repository.
4. Use a copy of the file scripts/segmentation_template.md, open it as ipython notebook, and load the BAHAMAS and dropsonde data to determine the individual segments, using a mix of bokeh and other plots of roll angle, altitude, heading or other measures.
5. Save your segments from the notebook to a YAML file.
6. test and check the YAML file using the scripts/report.py: python3 scripts/report.py flight_segment_files/HALO-20240813a.yaml reports/HALO-20240813a.html. This will create an HTML file that you can be opened in any browser and check the details of the flight segments.
7. If necessary, adjust the times and further info in the notebook and update the YAML file. Redo step 4 until you are satisfied with all segments.
8. Add your final YAML file to the repository by creating a pull request and assigning a reviewer. Don't add the reports/*.html files. THey will be generated automatically when you do the pull request and serve as a first check to validate the new YAML file.
9. Review your yaml file together with one other colleague
10. Check for inconsistencies with the flight report and send corresponding suggestions for improvement, together with your reviewed flight segmentation to the respective flight-PI for final feedback.

Reading the files

The flight segmentation data is provided in YAML files. YAML is a text based, human readable data format that uses python-like indentation for structuring its contents. It is often used for configuration files and due to its great human readability very suited for version control, which is one reason we wanted to use it here. For python users, the module PyYAML (included in Anaconda) offers an easy to use way to read the data from the yaml files into plane python objects like lists and dictionaries. Here is an example to read a file and print the circle start and end times from that file:

import yaml
flightinfo = yaml.load(open("HALO_20240813a.yaml"))
print([(c["start"], c["end"]) for c in flightinfo["segments"] if "circle" in c["kinds"]])

Further information

The segmentation copies and adapts many ideas from the flight segmentation conducted for the EUREC4A field campaign.

Further information on the respective flight can be found in the flight reports.