Table of Content

1. Table of Content
2. Notebooks
3. Examples
   1. NEXRAD L2 Reflectivity
   2. NEXRAD L3 Hydrometeor Classification
   3. GOES16/17/18 & Himawari 8/9 ABI L2
   4. GOES16/17/18 & Himawari 8/9 RRQPEF L2
   5. GOES16/17/28 GLM L2 Events
   6. MSG Seviris
   7. ERA5 Wind Speed
   8. Sentinel3 A/B OLCI
   9. Sentinel1 Deep Learning Models
4. Usage details
   1. Arguments for zone/time selection
   2. Arguments for the sensor selection
   3. Logins

Notebooks

Link to notebooks presenting colocations with Sentinel-1:

• NEXRAD L2 Reflectivity
• NEXRAD L3 DPR/N0M/N0H
• GOES16/17/18 & Himawari 8/9 ABI L2
• MSG 1&2 SEVIRIS 10.8 µm
• ERA5
• GOES16/17/18 & Himawari 8/9 RRQPEF L2
• GOES16/17/28 GLM L2 Events
• Sentinel3 A/B OLCI
• Sentinel1 Rain & Biological Slicks

These notebooks have been generated on Google Colab.

Examples

Here the list of the commands used in these notebooks:

NEXRAD L2 Reflectivity

python colocalize.py 20170108t015819 --data NEXRAD_L2 --create_gif True

NEXRAD L3 Hydrometeor Classification

python colocalize_nexrad_l3.py 20170108t015819 --channel HHC --create_gif True

GOES16/17/18 & Himawari 8/9 ABI L2

python colocalize.py 20210913t092920 --data ABI --channel C14 --create_gif True

GOES16/17/18 & Himawari 8/9 RRQPEF L2

python colocalize.py 20210913t092920 --data RRQPEF --create_gif True

GOES16/17/28 GLM L2 Events

python colocalize.py 20200827t000933 --data GLM --create_gif True

MSG Seviris

with open('seviris.txt', 'w') as file:
    file.write('test\t20210105T010101\t40.6\t7.6\t40.6\t10.3\t42.5\t10.3\t42.5\t7.6')

python colocalize_seviris.py --requests_filename seviris.txt --create_gif True

ERA5 Wind Speed

python colocalize.py 20200827t000933 --data ERA5 --channel northward_wind_at_10_metres --create_gif True --time_step 60 --max_timedelta 360

Sentinel3 A/B OLCI

python colocalize_s3.py 20180117T180947

Sentinel1 Deep Learning Models

! python process_s1.py 20170108t015819 NEXRAD

! python process_s1.py BiologicalSlicks --key 20180117t180947 

Usage details

Arguments for zone/time selection

The collocation scripts can be called on SAR products. In that case, the ISO datetime should be given as the first argument. By default, the programs will consider it to be a IW-VV and search in the res/listing_[IW][VV].zip file. It currently contains only from 2015-01-01 to 2021-12-31. EW-VV are also supported. In this case, --sensor_operational_mode should be set to EW.

It is possible to run the program for multiple collocations with the sentinel1_keys_filename and --requests_filename arguments. The first one will read a file containing the ISO datetime on each line. The second a file where each line is key ISO_datetime lat1 lon1 lat2 lon2 lat3 lon3 lat4 lon4 (separated by tabulations). For example, the following block will run collocation betwen NEXRAD and ERA5:

with open('foo.txt', 'w') as file:
    file.write("GLM_NEXRAD_ERA5_20210101T000000\t20210101T000000\t22.52\t-83.97\t22.52\t-79.43\t26.67\t-79.43\t26.67\t-83.97\n")
    file.write("GLM_NEXRAD_ERA5_20210101T010000\t20210101T010000\t22.52\t-83.97\t22.52\t-79.43\t26.67\t-79.43\t26.67\t-83.97\n")
    file.write("GLM_NEXRAD_ERA5_20210101T020000\t20210101T020000\t22.52\t-83.97\t22.52\t-79.43\t26.67\t-79.43\t26.67\t-83.97\n")
    
os.system("python colocalize.py --requests_filename foo.txt --data NEXRAD_L2 --create_gif False --delta_factor 0 --verbose 1")
os.system("python colocalize.py --requests_filename foo.txt --data ERA5 --channel eastward_wind_at_10_metres --create_gif False --delta_factor 0 --verbose 1")

will create collocations between NEXRAD and ERA5 on a square centered on KBYX.

Collocation with Sentinel3 only supports collocation with either a single ISO datetime or a filename containing them. In both cases, they are the first argument. Same for deep learning models.

Arguments for the sensor selection

This repo support collocations with the geostationnary satellites GOES (16 to 18), Himawari (8 & 9) and MSG. For the two formers, ABI 1 to 16 could be selected, by we are personnaly using only 13 & 14. To support them, add them in check_args.py in CHANNELS['ABI'] and add colormap parameters (name, vmin and vmax) in the platform_cmap_args function of utils/misc.py. GLM is available for GOES 16 to 18. Collocation for MSG have to be launched with the colocalize_seviris.py rather than colocalize.py. It only supports dataset EO:EUM:DAT:MSG:HRSEVIRI and EO:EUM:DAT:MSG:HRSEVIRI-IODC. Similarly with ABI, several channels are available and could be used with minimal modifications, but we are only interested in the 10.8 µm channel.

ERA5 collocations only supports northward_wind_at_10_metres and eastward_wind_at_10_metres. Other parameters should be added to check_args.CHANNELS['ERA5'] if needed (and to the misc.platform_cmap_args).

NEXRAD collocations can be obtained by either the colocalize.py (for the L2 products available on AWS) and with colocalize_nexrad_l3.py for L3 products. The latter supports the following channels:

• DPR: Digital Instantaneous Precipitation Rate
• NYQ: Base Reflectivity Data Array
• NXQ: Base Reflectivity Data Array
• NZM: Melting Layer
• NXH: Digital Hydrometeor Classification
• HHC: Hybrid Hydrometeor Classification

Sentinel3 OLCI "chl_oc4me" collocations can be obtained by colocalize_s3.py. This one only require a SAR ISO datetime and will search and plot the Sentinel3 on the same zone in a timeframe of +- 24h.

The notebooks should cover most use cases, check them to see which lib to install.

Logins

Some programs need log to download data. Those are Sentinel3, Sentinel1 and Seviris. They are read from a logins.py file not included in the repo for obvious reasons. It contains the following lines:

# Download Sentinel3 from finder.creodias.eu
SENTINEL_USERNAME = 
SENTINEL_PASSWORD = 

# Download Sentinel1 from ASF
ASF_USERNAME = 
ASF_PASSWORD = 

# Download ERA5 from eumdac api
EUMETSAT_CONSUMER_KEY = 
EUMETSAT_CONSUMER_SECRET = 

Deep Learning models

This repo can also apply deep learning models on Sentinel1 IW. Current models are available for rainfall estimation and biological slick segmentation. However, they need to download a 800 Mo L1 GRD.

If your GPU is not currently used, you can comment the os.environ['CUDA_VISIBLE_DEVICES'] = "-1" in utils/deep_learning.py.