sea-ice-phenology

Sea Ice Phenology Detection Developed by the ICE Remote Sensing Lab at University of Victoria (UVic).

Author: Sangwon Lim

Sea ice phenological indicators define starting and ending periods of seasonal ice formation and melt cycle:

• Melt Onset (MO): Starting date of the melt season
• Pond Onset (PO): Starting date of the pond formation
• Pond Drainage (PD): Starting date of the pond drainage
• Open Water (OW): Ending date of the melt season
• Freeze Onset (FO): Starting date of the ice formation

The package offers multiple processes to retrieve such information from MODIS and Landsat satellite optical data.

Getting Started

Downloading the Repository

Download this repository using the green Code button at the top right OR

$ git clone https://github.com/sum1lim/sea-ice-phenology.git

Install the Package in a Python Virtual Environment

Prerequisite

• gcloud CLI

Windows

Double-click on install batch script to install.

Double-click on uninstall batch script to uninstall.

Apple Silicon (M1)

Navigate to the parent directory in terminal, sea-ice-phenology, in the command line interface and run the following commands.

$ sh M1_install.sh
$ source ~/miniforge3/bin/activate

Remove sip-venv to uninstall

$ conda remove -n sip-venv --all

Linux & OSX (Intel)

Navigate to the parent directory in terminal, sea-ice-phenology, in the command line interface and run the following commands.

$ python -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt
$ pip install .

Remove venv to uninstall

$ rm -r venv

Graphical User Interface (GUI)

Windows

Double-click on gui batch script

Linux & OSX

$ gui

Command-line User Interface (CLI*)

*CLI is not recommended for Windows

Authenticating Google Earth Engine Credentials

Authentication is required prior to retrieving remote sensing data from Google Earth Engine Data Catalogue. Start the process by running the following command in a CLI:

$ authenticate

Get Time Series

Retrieve optical reflectance time series data at a user-defined location from MOD09GQ.006 Terra Surface Reflectance Daily Global 250m and USGS Landsat 8 Collection 2 Tier 1 TOA Reflectance products. Wavelengths of 0.64-0.67µm and 0.64-0.67µm are selected, respectively. Reflectance mean of 9 points, center point surrounded by points in 8 directions (N, NE, E, SE, S, SW, W and NW) distanced at 125/$\tan$($\pi$/8) meters, are calculated for each date. Invalid pixels with cloud coverage, low solar angle and land contamination are not selected.

$ get_timeseries --help
usage: get_timeseries [-h] [--coords COORDS COORDS]
                      [--coords-csv COORDS_CSV COORDS_CSV COORDS_CSV COORDS_CSV]
                      --start-date START_DATE --end-date END_DATE
                      [--output OUTPUT]

optional arguments:
  -h, --help            show this help message and exit
  --coords COORDS COORDS
                        Area of Interest (AOI) defined by lon, lat
                        coordiates
  --coords-csv COORDS_CSV COORDS_CSV COORDS_CSV COORDS_CSV
                        Input file (CSV) followed by column names of
                        lon, lat and label
  --start-date START_DATE
                        YYYY-MM-DD
  --end-date END_DATE   YYYY-MM-DD
  --output OUTPUT       Output file/dir path

Command line to retrieve time series from a user-defined location:

$ get_timeseries --coords [Lon] [Lat] --start-date [YYYY-MM-DD] --end-date [YYYY-MM-DD] --output [Path to output CSV file]

Command line to retrieve time series from a CSV file with multiple (Lat, Lon) coordinates:

$ get_timeseries --coords-csv [Path to CSV file] [Longitude column] [Latitude column] [Label column] --start-date [YYYY-MM-DD] --end-date [YYYY-MM-DD] --output [Path to output directory]

Example(s):

# To retrieve time series from a user-defined location:
$ get_timeseries --coords -105.304 69.0435 --start-date 2017-01-01 --end-date 2022-01-01 --output ./example/Cambridge_Bay.csv

# To retrieve time series from a CSV file with multiple (Lat, Lon) coordinates:
$ get_timeseries --coords-csv ./CommunityList.csv Lon_ice Lat_ice Community --start-date 2017-01-01 --end-date 2022-01-01 --output ./example

Output: CSV file(s)

system:time_start	B1
2017-02-07	0.7299
2017-02-09	0.6476444444444445
2017-02-10	0.8302444444444445
2017-02-11	0.8168555555555556
2017-02-12	0.8367444444444444
2017-02-18	0.7521222222222222
2017-02-20	0.5673111111111111
2017-02-22	0.6707

Time-Series Interpolation

Data retrieved in the previous step may include voids in the time series, which can be problematic for further processes. Multiple interpolation methods are provided to fill in the gaps.

$ interpolate --help
usage: interpolate [-h] --input INPUT [INPUT ...]
                   [--type {linear,slinear,median,gaussian,random}]
                   --output OUTPUT

optional arguments:
  -h, --help            show this help message and exit
  --input INPUT [INPUT ...]
                        Path to time-series data file (csv)
  --type {linear,slinear,median,gaussian,random}
                        Interpolation type
  --output OUTPUT       Output dir path

Command Line:

$ interpolate --input [Path to input file] --type [Interploation method] --output [Path to output directory]

Example(s):

# Single Input
$ interpolate --input ./example/Cambridge_Bay.csv --type median --output ./example

# Multiple inputs
$ interpolate --input ./example/Cambridge_Bay.csv ./example/Alert.csv --type median --output ./example

# Multiple inputs (Every file in a directory)
$ interpolate --input ./example/* --type median --output ./example

Output: CSV file(s) and graph visualization

system:time_start	B1	B1_interpolate
2017-02-07	0.7299	0.7299
2017-02-08		0.6887722222222222
2017-02-09	0.6476444444444445	0.6476444444444445
2017-02-10	0.8302444444444445	0.8302444444444445
2017-02-11	0.8168555555555556	0.8168555555555556
2017-02-12	0.8367444444444444	0.8367444444444444
2017-02-13		0.7944333333333333
2017-02-14		0.8060799999999999

Phenology Detection

Detects sea ice phenological indicators of MO, PO, PD, OW and FO. 1D-smoothing, Hampel and Lowess filtering on the time series data, is performed before the detection to enhance the performance.

$ phenology --help
usage: phenology [-h] --input INPUT [INPUT ...] [--type {slope-diff}]
                 --output OUTPUT

optional arguments:
  -h, --help            show this help message and exit
  --input INPUT [INPUT ...]
                        Path to time-series data file (csv)
  --type {slope-diff}   Phenology detection method
  --output OUTPUT       Output dir path

Command Line:

$ phenology --input [Path to input file] --type [Phenology detection method] --output [Path to output directory]

Example(s):

# Single Input
$ phenology --input ./example/Cambridge_Bay_interpolate.csv --type slope-diff --output ./example

# Multiple inputs
$ phenology --input ./example/Cambridge_Bay_interpolate.csv ./example/Alert_interpolate.csv --type slope-diff --output ./example

# Multiple inputs (Every file in a directory)
$ phenology --input ./example/* --type slope-diff --output ./example

Output: CSV file(s) and graph visualization

system:time_start	B1	B1_interpolate	B1_phenology
2017-02-07	0.7299	0.7299
2017-02-08		0.6887722222222222
2017-02-09	0.6476444444444445	0.6476444444444445
2017-02-10	0.8302444444444445	0.8302444444444445
2017-02-11	0.8168555555555556	0.8168555555555556
2017-02-12	0.8367444444444444	0.8367444444444444
2017-02-13		0.7944333333333333
2017-02-14		0.8060799999999999
...	...	...	...
2017-05-08		0.9531222222222222	MO
...	...	...	...
2017-06-05	0.3274563252925873	0.3274563252925873	PO
...	...	...	...
2017-06-22	0.4115777777777777	0.4115777777777777	PD
...	...	...	...
2017-07-19	0.0041111111111111	0.0041111111111111	OW
...	...	...	...
2017-10-16	0.2260666666666666	0.2260666666666666	FO

Phenological Trend

Calculates linear trends of annual phenological indicators detected.

$ trend --help
usage: trend [-h] --input INPUT [INPUT ...] --output OUTPUT

optional arguments:
  -h, --help            show this help message and exit
  --input INPUT [INPUT ...]
                        Path to time-series data file (csv)
  --output OUTPUT       Output dir path

Command Line:

$ trend --input [Path to input file] --output [Path to output directory]

Example(s):

# Single Input
$ trend --input ./example/Cambridge_Bay_phenology.csv --output ./example

# Multiple inputs
$ trend --input ./example/Cambridge_Bay_phenology.csv ./example/Alert_interpolate.csv --output ./example

# Multiple inputs (Every file in a directory)
$ trend --input ./example/* --output ./example

Output: CSV file(s) and graph visualization

Year	MO	OW	FO
2017			288
2018	145	207	287
2019	146		289
2020	148
2021	149	209
2022	144	206	289