The raw Sentinel-2 satellite image appears dark and details are obscured to the human eye. To better visualize and interpret key features, contrast enhancement and color transformation techniques common in remote sensing image processing are applied. Specifically, linear contrast stretching improves the dynamic range devoted to intensities within the region of interest. Additionally, converting the native spectral band combinations into displays more intuitive for human perception based on long-standing color modeling research facilitates more rapid digitization and annotation. Such pre-processing steps are essential to convert the raw sensor observations into derivations by visual analytics. They remove barriers to geospatial time series analysis while retaining the underlying radiometric fidelity. Appropriate image transformations enable the integration of human semantic understanding about landscape change with automated algorithms. This helps advance spatiotemporal detection of phenomena related to sustainability challenges.
vmin, vmax = np.nanpercentile(red, (5,95)) # 5-95% contrast stretch
plt.figure(figsize=[10,10])
show(red, cmap='gray', vmin=vmin, vmax=vmax)
plt.show()