Inspect S1 Time-Series for Area of Interest¶

Authors:

• Victor Tang

Reviewed/Edited by:

• Marcos Kavlin
• Dr. Andrew Dean

Purpose¶

This notebook is the first in the SWIFT-C Tutorial. The goal of this notebook is to show you, the user, how to inspect Sentinel 1 time series for an area of interest. This will be done as part of a series of Notebooks, designed to walk you through the steps involved in assessing Wetland Function through time.

Workflow¶

1. Import necessary packages.
2. Define a normalization function for your S1 imagery.
3. Open the S1 imagery, apply the normalization, and visualize the time series.

Notes¶

The composites that were used to run this notebook are placeholders used to demonstrate this workflow. The data required to run the steps demonstrated in this notebook are 10 day Sentinel 1 median composites. In this example the images were produced for the year 2022. The images were grouped into 10 day composites as that was most conducive to a complete year-long time series, while still removing speckle and noise. To best follow along with this tutorial we recommend all Sentinel 1 images for a single year are all put into a folder for that year.

1. Import required packages¶

In [ ]:

import numpy as np
import matplotlib.pyplot as plt

import glob
from tqdm.notebook import tqdm

import rioxarray as rxr

import xrspatial.multispectral as ms

base_dir = f"{path_to_data}"

import numpy as np import matplotlib.pyplot as plt import glob from tqdm.notebook import tqdm import rioxarray as rxr import xrspatial.multispectral as ms base_dir = f"{path_to_data}"

2. Create a normalization function¶

In this step we create a function to normalize our S1 images to ensure that they can be accurately compared.

In [ ]:

def normalize(val: np.ndarray, lower: float, upper: float) -> np.ndarray:
    """Normalize a raster band

    Args:
        val: np.ndarray
            The desired raster band to
            normalize.

        lower: float
            The percentile to use as
            a minimum value for the
            normalization.

        upper: float
            The percentile to use as
            a maximum value for the
            normalization

    Returns:
        normalized_band: np.ndarray
            A normalized np.ndarray of the input
            band.

    """
    vmin = np.nanpercentile(val, lower)
    vmax = np.nanpercentile(val, upper)
    b_copy = val.copy()
    b_copy[b_copy < vmin] = vmin
    b_copy[b_copy > vmax] = vmax
    normalized_band = (b_copy - vmin) / (b_copy - vmin)
    return normalized_band

def normalize(val: np.ndarray, lower: float, upper: float) -> np.ndarray: """Normalize a raster band Args: val: np.ndarray The desired raster band to normalize. lower: float The percentile to use as a minimum value for the normalization. upper: float The percentile to use as a maximum value for the normalization Returns: normalized_band: np.ndarray A normalized np.ndarray of the input band. """ vmin = np.nanpercentile(val, lower) vmax = np.nanpercentile(val, upper) b_copy = val.copy() b_copy[b_copy < vmin] = vmin b_copy[b_copy > vmax] = vmax normalized_band = (b_copy - vmin) / (b_copy - vmin) return normalized_band

3. Open S1 images, apply the normalization function, and visualize the imagery¶

The following two cells focus on the following steps:

• List the images in the directory of interest that we wish to normalize
• Open each image using rioxarray
• Using the VV and VH bands of each image calculate the VV/VH index
• Apply the normalization function we defined in the previous section to each band
• Plot our normalized False-Color Composites, in a sequential fashion.

In [ ]:

target_year = folder_with_data_for_target_year
fpath_list = glob.glob(base_dir + "/*.tif".format(target_year))

target_year = folder_with_data_for_target_year fpath_list = glob.glob(base_dir + "/*.tif".format(target_year))

In [4]:

fpath_list = fpath_list[:36]

fig, axes = plt.subplots(6, 6, figsize=[20, 20], facecolor="white")
axes = axes.ravel()

for fpath in tqdm(fpath_list):
    i = int(fpath.split("_")[-1].split(".")[0])

    da = rxr.open_rasterio(fpath).assign_coords(band=["vh", "vv"])

    vv = da.sel(band="vv")
    vh = da.sel(band="vh")
    rvh = vv / vh

    vv.values = normalize(vv.values, 1, 99)
    vh.values = normalize(vh.values, 1, 99)
    rvh.values = normalize(rvh.values, 1, 99)

    img = ms.true_color(
        r=vv,
        g=vh,
        b=rvh,
        nodata=np.nan,
    )

    ax = axes[i - 1]
    img.plot.imshow(ax=ax)
    ax.set_title("Period {}".format(i), fontsize=18, fontweight="bold")
    ax.set_axis_off()

fig.suptitle(
    "{} Sentinel-1 False Color (VV:VH:VV/VH)".format(target_year),
    fontsize=36,
    fontweight="bold",
    y=1.0,
)
fig.tight_layout()

fpath_list = fpath_list[:36] fig, axes = plt.subplots(6, 6, figsize=[20, 20], facecolor="white") axes = axes.ravel() for fpath in tqdm(fpath_list): i = int(fpath.split("_")[-1].split(".")[0]) da = rxr.open_rasterio(fpath).assign_coords(band=["vh", "vv"]) vv = da.sel(band="vv") vh = da.sel(band="vh") rvh = vv / vh vv.values = normalize(vv.values, 1, 99) vh.values = normalize(vh.values, 1, 99) rvh.values = normalize(rvh.values, 1, 99) img = ms.true_color( r=vv, g=vh, b=rvh, nodata=np.nan, ) ax = axes[i - 1] img.plot.imshow(ax=ax) ax.set_title("Period {}".format(i), fontsize=18, fontweight="bold") ax.set_axis_off() fig.suptitle( "{} Sentinel-1 False Color (VV:VH:VV/VH)".format(target_year), fontsize=36, fontweight="bold", y=1.0, ) fig.tight_layout()

  0%|          | 0/36 [00:00<?, ?it/s]