Open SAR Toolkit - Tutorial 4c, version 1.2, August 2020. Andreas Vollrath, ESA/ESRIN phi-lab

title

OST Tutorial IV - C¶

Create country-wide mosaic time-series and timescan. Introduction to GRD Batch processing part III.¶

Short description

This notebook is very similar to the Tutorial IVa, with the difference that you will process GRD data over a larger area and create time-series and timescan mosaics. It therefore represents the workflow for large-scale processing.

Requirements

a PC/Mac with at least 16GB of RAM
about 150GB of free disk space
a Copernicus Open Data Hub user account, valid for at least 7 days (https://scihub.copernicus.eu)

NOTE: all cells that have an * after its number can be executed without changing any code.

0* - Install OST and dependencies¶

NOTE: Applies only if you haven’t fully installed OST and its dependencies yet, e.g. on Google Colab, so uncomment the lines in this case.

[ ]:

# !apt-get -y install wget
# !wget https://raw.githubusercontent.com/ESA-PhiLab/OST_Notebooks/master/install_ost.sh
# !bash install_ost.sh

1* - Import of Libraries¶

In this step we import some standard python libraries for OS independent path handling as well as the Sentinel1_GRDBatch class thta handles the full workflow from search, download and processing of multiple GRD scenes. In addition, the OST helper module vector is loaded to create an AOI based on Point coordinates, and the raster module for creating a time-series animation.

[ ]:

from pathlib import Path
from pprint import pprint

from ost import Sentinel1Batch
from ost.helpers import vector, raster

2* - Set up the project¶

Here you going to initialize the Sentinel1_GRDBatch class by determining the project folder, the AOI and the start and end date. Since you should be already familiar with the search and refine functions, we execute them within the same cell.

[ ]:

# define a project directory
home = Path.home()
# create a processing directory
project_dir = home / "OST_Tutorials" / "Tutorial_4c"

# define aoi with helper function, i.e. get a buffered area around point coordinates
aoi = "IRL"

# define the start and end date
start = "2019-02-01"
end = "2019-04-30"

# initialize the class to s1_grd instance
s1_grd = Sentinel1Batch(project_dir=project_dir, aoi=aoi, start=start, end=end, product_type="GRD")

# trigger the search
s1_grd.search()

# optional: once you did the search the first time, you can load
# the full inventory uncommenting the follwoing 2 lines
# and commenting the search command
# s1_grd.inventory_file = s1_grd.inventory_dir/"full.inventory.gpkg"
# s1_grd.read_inventory()

# do the refinement
s1_grd.refine_inventory()

3* - Plot refined data inventory¶

Here you will visualize the resultant dataframes from the refined search inventory based on the product key.

[ ]:

# ---------------------------------------------------
# for plotting purposes we use this iPython magic
%matplotlib inline
%pylab inline
pylab.rcParams["figure.figsize"] = (19, 19)
# ---------------------------------------------------

# search command
key = "ASCENDING_VVVH"
nb_frames = len(s1_grd.refined_inventory_dict[key])
print(f"Our refined inventory holds {nb_frames} frames to process.")

# we plot the full Inventory on a map
s1_grd.plot_inventory(s1_grd.refined_inventory_dict[key], transparency=0.1)

4* - Download of GRD scenes¶

As already shown in Tutorial II, you will download the scenes based on the refined inventory dataframe for the respective produckt key.

[ ]:

s1_grd.download(s1_grd.refined_inventory_dict[key])

5* - Set ARD parameters¶

Similar to the Sentinel1-Scene class (Tutorial I and III), the Sentinel1-GRDBatch class handles the defintion of ARD types in a hierarchical dictionary structure. You can use the same types and steps to customize as for the Sentinel1-Scene class.

[ ]:

# single scene ARD parameters
s1_grd.ard_parameters["single_ARD"]["resolution"] = 50
s1_grd.ard_parameters["single_ARD"]["product_type"] = "GTC-gamma0"
s1_grd.ard_parameters["single_ARD"]["create_ls_mask"] = False

# time-series ARD
s1_grd.ard_parameters["time-series_ARD"]["remove_mt_speckle"] = False

# our borders for Ireland are quite rough. We therefore won't clip the final mosaics
s1_grd.ard_parameters["mosaic"]["cut_to_aoi"] = False

# s1_grd.config_dict['temp_dir'] = '/ram'
pprint(s1_grd.ard_parameters)

6* - Run the batch routine¶

To process all the data, including time-series and timescans is as easy as one command. All the complexity is handled behind, and you just have to wait, since processing can take quite a while.

[ ]:

s1_grd.grds_to_ards(
    inventory_df=s1_grd.refined_inventory_dict[key],
    timeseries=True,
    timescan=True,
    overwrite=False,
)

7* - Create a time-series animation¶

For interactive presentations it is nice to have animated “movies”. The following command allows you to create animated time-series of oyur processed data.

[ ]:

# define Time-series folder
ts_folder = s1_grd.processing_dir / "23" / "Timeseries"

# create the animation
raster.create_timeseries_animation(
    timeseries_folder=ts_folder,
    product_list=["bs.VV", "bs.VH"],
    out_folder=s1_grd.processing_dir,
    shrink_factor=10,
    add_dates=True,
)