Forest cover refers to the extent of land area covered by forests. It can be expressed either as a percentage relative to the total land area or in absolute terms measured in square kilometers or square miles (ScienceDirect). As of 2020, globally, forests account for 31 percent of the land area with roughly half of this area considered relatively intact. The total forest coverage is 4.06 billion hectares. The majority of world's forests, over half, are concentrated in five countries (the Russian Federation, Brazil, Canada, the United States of America, and China), and the top ten countries collectively harbor two-thirds (66 percent) of the world's forested areas as shown in figure 1 and 2 (FAO, 2020).

Importance and challenges of monitoring water under forest cover

Forests play a vital role in nature's water management system by collecting and storing rain and snow before delivering it to streams, wet meadows, and aquifers. In the face of rapid climate change (Pacific Forest Trust, 2022), maintaining the health of forest watersheds is important for ensuring water security, resilience to fires, and flood control Detecting flooded vegetation is necessary for monitoring wetlands and floods (Tsyganskaya et al., 2018). Wetlands offer essential services like flood control, sediment storage, wildlife habitat, and more (Reid et al., 2005).

Tropical forests and wetlands are important and their exploration is challenging for Earth scientists because they are difficult to access (NASA Earth Observatory, 2013). The dense canopy of trees obstructs direct observation and monitoring of the ground from the air.
Hence, the monitoring of small-scale, gradual, or seasonal changes in vegetation and water accumulation beneath the tree cover is challenging.

Remote sensing enables extensive data collection across large, forested regions (Spatial Post, 2023). It helps in monitoring water beneath forest cover by using various sensors and technologies (Le et al., 2023). Synthetic Aperture Radar (SAR) can penetrate dense canopies, provide insights into underlying surfaces, and detect inundation (Gašparović and Klobučar, 2021).

Synthetic-Aperture Radar (SAR)

Synthetic-aperture radar (SAR) is a radar technique employed for generating two-dimensional images or three-dimensional reconstructions of various objects such as landscapes (Kirscht and Rinke, 1998). It is referred to as an active sensor as it generates its own energy and measures the reflected amount of energy after its interaction with the Earth (NASA Earth Data). SAR transmits and receives energy at microwave frequencies (McNairn and Shang, 2016). The SAR antenna simultaneously transmits multiple pulses, typically chirps, and then awaits their return at a predetermined time interval because it cannot receive signals during transmission (Galassi, 2021). An advantage of SAR is its ability to penetrate cloud cover. It therefore can be used in all-weather conditions and offers high-resolution images independent of day or night. SAR is used in various domains such as geoscience, climate change research, environmental monitoring, Earth system observation, 2-D and 3-D mapping, change detection, and even planetary exploration (Moreira et al., 2013).

In a SAR system, whether space-borne or airborne, electromagnetic microwave pulses are directed toward the Earth's surface in a side-looking geometry as shown in Figure 4. The SAR, moving along a presumed straight path, illuminates a ground swath at an angle known as the incidence angle (θ) (Lauknes, 2011). Echoes from the surface interact with objects like trees or rocks and are captured by the SAR receiver. The resulting SAR-focused image, termed a single-look complex (SLC) is a 2D array with a slant range (R) representing distance and azimuth (x) indicating the scatterer's position along the SAR flight path (Lauknes, 2011).

SAR backscatter

Backscatter refers to the portion of the radar signal that the target reflects directly back to the radar antenna. Backscattering is the mechanism through which this reflective signal, known as backscatter, is generated (Sentinel Online). Detecting flooding under tree canopies, such as in dense and closed forests is challenging using the optical spectrum unless the canopy is open or incomplete (Bourgeau-Chavez et al., 2009). Therefore, SAR has proven to be advantageous for flood detection, especially under forest canopies, and where optical sensors may face limitations due to weather conditions (Gašparović and Klobučar, 2021). The backscatter signals from SAR interact differently with flooded and non-flooded areas. The presence of water under the vegetation canopy enhances the signal’s feedback. The integration of SAR data capturing vegetation structure, soil moisture, and flooded areas with data is considered beneficial for accurate wetland mapping (Gašparović and Klobučar, 2021). The factors affecting SAR backscattering include band wavelength, polarization, and moisture.Figure 3 shows interaction of optical and SAR sensors with forest canopy (Boyd and Danson, 2005).

SAR band wavelength

The backscatter received by SAR sensor is influenced by various factors. The wavelength employed in SAR affects the signal's penetration and, consequently, the imaged content. Surface roughness acts as a modulation factor (a variable that alters the intensity of the signal) for backscatter returns, ranging from minimal to intense, thereby decreasing or increasing the brightness of the resulting pixel (Alaska Satellite Facility, 2023).
Figure 5 illustrates how the sensor wavelength, λ affects signal penetration across surfaces signal penetration across surfaces With increasing wavelength, radar signals penetrate deeper (Meyer, 2019).

Table 1 summarizes SAR applications based on frequency bands. X-band sensors are primarily employed for urban and infrastructure monitoring due to their high resolution. However, their limited penetration into vegetation makes X-band rarely suitable for characterizing forest canopies or monitoring activities beneath vegetation (Meyer, 2019). C-band sensors have been predominant in SAR monitoring for the past 30 years, and with moderate-to-high resolution and improved vegetation penetration compared to X-band, it serves as a practical compromise. It allows for wider swath imaging, suitable for regional and global-scale applications. While C-band enhances canopy penetration, its signals may not fully penetrate dense vegetation, limiting its effectiveness for analyzing activities beneath canopy layers (Meyer, 2019; Kellndorfer et al., 2019). Historical SAR systems mainly operated in C-band. The upcoming generation of SAR sensors is however, predominantly focused on the L-band frequency range, and although L-band SARs lack the high-resolution of shorter wavelengths, their capacity to penetrate vegetation offers significant advantages for Earth observation. With a better chance of observing the ground, L-band SARs are valuable for mapping activities beneath canopies such as flooding. Due to their superior penetration into vegetation, L-band SARs are well-suited for characterizing canopy structures, particularly in denser forests (Meyer, 2019; Kellndorfer et al., 2019).

Table 1: Microwave bands categorized based on their frequencies. Spaceborne SARs commonly function within the frequency bands highlighted in green (Meyer, 2019).

Polarization

As an electromagnetic wave travels through time and space, its properties may undergo alterations due to interactions along its path and with the target object. This may result in a change in polarization (Galassi, 2021).

SAR sensors commonly utilize four polarization combinations: VV, VH, HV, and HH, as outlined in Table 2. The initial letter denotes the transmitted signal's polarization, while the second letter indicates the polarization of the received return, as depicted in Figure 6 (Alaska Satellite Facility, 2023).

Co-polarization (VV, HH) is strong for surface scattering, while cross-polarization (VH, HV) is associated with measuring volume scattering. Cross-polarized observations are essential for biomass applications and tracking changes in forests (Alaska Satellite Facility, 2023) (NISAR JPL, 2020).

Figure 7 shows an example of a polarimetric imagery captured by the Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) instrument over Rosamond, California. The transmission of horizontal and vertical polarized signals, along with the reception of resulting backscatter signals, generated three channels of imagery. By colorizing and overlaying these distinct images, details and variations in surface features are easily distinguishable (NISAR JPL, 2020)

Soil Moisture and its signature on SAR images

The interaction of SAR signals with the Earth's surface is influenced by soil moisture. Accurate retrieval requires accounting for various surface properties such as land cover, soil texture, and surface roughness (Gharechelou et al., 2021). Elevated moisture content in soils and vegetation correlates with increased backscatter signals. Standing open water appears very dark in SAR images as most of its energy scatters away from the sensor. Shorter wavelengths, such as those in C- and X-bands have strong backscatter from rough water surfaces (Kellndorfer et al., 2019).

Figure 8 illustrates the impact of moisture on Sentinel-1 C-band data over Ecuador. The darkening effects result from intense tropical convection systems during active rain causing signal attenuation. Conversely, the brightening effects are attributed to wet vegetation and soils from rain events associated with the tropical frontal system. Riverbeds remain visible amid the brightened backscatter areas in the affected image from February 27, 2017, confirming that SAR signals indeed reflect an increase in vegetation and soil moisture (Kellndorfer et al., 2019).

Table 3 provides a summary of the expected backscatter characteristics across various vegetation transition scenarios and has values ranging from very low, low, medium, high, to very high (Kellndorfer et al., 2019).

Table 3: SAR backscatter and forest type (Kellndorfer et al., 2019).

Detecting flooding under tree cover

“Flooding is difficult to reliably detect with optical imagery because the forest canopy obscures the view. Another problem is clouds, especially during the rainy season”, - Bruce Chapman, NASA Earth Observatory, 2013.

NASA's Jet Propulsion Laboratory, led by experts like Chapman, utilized the Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) to capture hidden landscapes beneath cloud cover and dense vegetation. UAVSAR employs microwaves with a wavelength within the L-Band (20 centimeters) which easily penetrates tree canopies and clouds. This enables year-round data acquisition regardless of weather conditions. The radar imagery as shown in Figure 9 produced over the Napo River in Ecuador and Peru revealed flooded areas beneath forest canopies in yellow, orange, and blue, while healthy vegetation appeared green. The technology's ability to distinguish inundation and changes in wetlands not only aids in flood detection but also contributes to climate change research (NASA Earth Observatory, 2013) (NASA JPL UAVSAR).

Another study named “Using C-Band Synthetic Aperture Radar Data to Monitor Forested Wetland Hydrology in Maryland’s Coastal Plain, USA” explored the potential of C-band synthetic aperture radar (SAR) data, specifically C-HH and C-VV polarizations in detecting and monitoring levels of inundation and soil moisture under forest canopies (Lang and Kasischke, 2008). Focused on a Mid-Atlantic floodplain, the study analyzed the backscatter coefficient's relationship with hydrological factors, such as inundation, soil moisture, height, and forest canopy closure. The findings of the study reveal that C-HH SAR data performed better in correlation with inundation and soil moisture than C-VV data. The authors emphasized the utility of C-band SAR for monitoring forested wetland hydrology throughout the year and provided valuable insights for regions vulnerable to wetland loss and climate change impacts.

Conclusion

Monitoring areas beneath forest cover, especially for detecting flooding poses significant challenges due to the obstructive nature of dense canopies and the limitations of traditional remote sensing technologies. However, Synthetic Aperture Radar (SAR) emerges as a powerful tool in overcoming these challenges. SAR's ability to penetrate through vegetation canopies, aided by factors such as wavelength, polarization, and sensitivity to moisture content, makes it an asset for monitoring activities beneath tree cover.