Introduction to Sentinel 1: SAR(Synthetic-Aperture Radar) Data
Synthetic Aperture Radar(SAR) data are images of the earth’s landscape created by the reflective signals, called back-scatter, that return to the radar. SAR emits its own energy to retrieve later, making it an active sensor. In contrast, passive sensors such as optical images retrieve the information reflected from another source(such as the sun).
Advantages/Disadvantages
SAR has many useful applications over the optical image mainly because the energy transmitted is in the microwave range and is not hindered by day/light, clouds, or rain(for larger wavelengths). Also, SAR(at larger wavelengths) can penetrate through dense forests, while optical images will only capture the trees’ top layer.
However, the image generated won’t describe the “greenness” as SAR can only describe an object’s presence. Therefore, SAR images are sometimes difficult to interpret. Another disadvantage of SAR is that the images have to be taken at an angle, resulting in a loss of information gain if a shadow covers the land. The image may also have alternating pixels of strong-receive/no receive, creating a “salt-and-pepper-like” image due to the alignment of the object relative to the angle facing the satellite.
SAR Wavelengths
Radar sensors emit energy in wavelength greater than the infrared wavelength and less than the radio wavelength. SAR wavelengths can range from 1cm — 100cm and a range of wavelengths are designated to a specific band type:
In short, bands with greater wavelengths penetrate deeper into the ground surface. For example, X-bands might volume scatter with the leaves and secondary branches, C-bands will volume scatter with secondary branches and primary branches, and L-bands scatter with primary branches and the trunk.
For this blog, I will explicitly talk about the open-sourced Sentinel 1 SAR that has a C-Band Radar. The dataset can be found on the google earth engines catalog found here: https://developers.google.com/earth-engine/datasets/catalog/COPERNICUS_S1_GRD
Backscatter
If the surface is flat, the signal will bounce off the flat surface, and the receive will return no signal. When a signal is returned, the receive’s magnitude differs based on the direction of the transmitting waves and receiving waves. Sometimes, the sensor can transmit-receive in all directions, called circular polarized, but typically the waves transmit either horizontally or vertically and receive horizontally and vertically. The math behind these interactions can be complex, but the HH(Horizontal-transmit horizontal-receive), HV, VV, and VH values can be intuitively evaluated as; the HH value is the horizontal received given horizontal transmit.
Sentinel 1, for example, transmits horizontally when monitoring polar environments and sea ice zones and transmits vertically for all else.
The value of the backscattering may be explained by three different types of backscattering:
1. Rough Surface — Rocky bare soil or water bodies that have allowed at least some of the signals to be returned. The rough surface is most sensitive to VV.
2. Volume — Typically trees, where the signal is bounced off the branches of the tree or its neighbors and are returned back. Volume scattering is most receptive to VH or HV
3. Double Bounce — Typically two flat surfaces such as buildings, cliffs, or inundated vegetation where the signal bounces twice and returns a strong signal. Double Bounce shows the highest signals on HH.
It’s a bit difficult to say which transmit-receive combo associates better with certain types of objects. The interpretation of the values can be interpreted case by case. For example, larger VV values on a tropical forest after rain can be explained by inundation. (The double bounce from the flat flooded water and tree.)
Incidence Angle/Shadow
It’s also important to note that SAR images are taken at an angle. The image might need preprocessing as pixels that are further away from the satellite have less penetration. The return value might be less for landscapes further away from the satellite than similar landscapes close. Also, because the image is taken at an angle, there might be regions (typically over a mountain) where the signal never reached.
Future SAR data
Currently, Sentinel 1 open sources C-Band SAR data and is publicly available. Anyone can pull the dataset from the Google Earth Engine hub, which holds publicly available data and can process the data on their server. In the future, more open-sourced SAR data will be publicly available, specifically from NISAR(2022) and BIOMASS(2022). NISAR is a joint NASA-ISRO(Indian Space Research Organisation) project which will carry L-band and S-band radars. BIOMASS is a project by the ESA(European Space Agency) with P-band radars.
Sources:
SAR Handbook(J. Kellndorfer)— https://www.servirglobal.net/Global/Articles/Article/2674/sar-handbook-comprehensive-methodologies-for-forest-monitoring-and-biomass-estimation
Comparative flood area analysis of C-band VH, VV, and L-band HH polarizations SAR data (R. Manavalan, Y. Rao & K. Mohan) — https://www.tandfonline.com/doi/abs/10.1080/01431161.2017.1325534
What is Synthetic Aperture Radar?
https://earthdata.nasa.gov/learn/backgrounders/what-is-sar
