What is RADARSAT Constellation Mission (RCM)?
The RADARSAT Constellation Mission (RCM), a government-owned and controlled constellation and a continuation of the wildly successful RADARSAT-1 and RADARSAT-2 missions, was launched by Canada in 2019.
RCM's primary goal is to assist Canada's operational initiatives for disaster management, ecological monitoring, and marine surveillance.
When all RCM satellites are employed, the precise revisit time decreases to four days from each satellite's 12-day revisit time. As a result, RCM can provide daily coverage of Canada's territorial and bordering seas with numerous daily re-looks achievable at higher latitudes. Depending on the beam mode, spatial resolutions may vary but typically range from 3-100m.
FP Mode of RCM
Each satellite has a fully polarimetric (FP) mode that RCM maintains. The FP mode broadcasts and receives horizontal (H) and vertical (V) orthogonal polarizations while maintaining the phase.
The pulse repetition frequency (PRF) must be twice as high as that needed for single- or dual-polarized SARs to interleave these orthogonal polarizations during transmission. The FP swath width must be cut in half when the PRF is doubled. The short FP swath is not useful for operational mapping, despite the entire scattering matrix providing a substantial dataset.
Compact Polarimetry (CP)
Compact Polarimetry (CP), a partial answer for customers that need large swath coverage but wish to maintain polarization richness, was put out during the creation of RCM.
Radar astronomy had already proved the potential of CP. Government scientists were encouraged to examine simulated CP data in response to this. Their study proved that the CP might offer products useful for government operations, including precise agricultural inventories.
The Stokes parameters describe a partly polarized electromagnetic (EM) field's polarization state. The initial Stokes parameter (S0) is the sum of the powers of the two orthogonally polarized received waves and represents the overall intensity of the radar backscatter (unpolarized and polarized). The other three parameters describe the polarized region properties of the electromagnetic field (S1, S2, and S3).
Evaluating CP Data for Crop Inventory Operations' Improvement
The primary purpose of this study is to evaluate CP data obtained by RCM for the improvement of crop inventory operations. RCM CP data affect operations and other regional and international agricultural monitoring projects. This research compares the performance of CP to the current processes that need inputs of optical imagery by performing classification utilizing the four Stokes parameters and m-chi decomposition parameters produced from a temporal stack of RCM CP pictures.
Site Location
Kenaston situated in Saskatchewan, Canada, was the region of interest for this study. Kenaston has played a significant role in agricultural remote sensing research in Canada. The climate in this prairie area is generally chilly, dry to subhumid, with sporadic summer rainfall.
How the Research was Conducted
Seventeen RCM pictures were collected in CP ScanSAR mode. These pictures have a baseline spatial resolution of 30 m and a sweep width of 125 km. The four Stokes parameters, three elements of the m-chi decomposition (volume scattering, double bounce, and surface), and backscatter intensities (RH; RV) were all determined from this CP Single Look Complex (SLC) data.
The accuracy of classifying seven crop types (lentils, peas, flaxseed, canola, wheat, barley and pasture/forage) was examined using a Random Forest (RF) classifier. Eight classification runs were produced from testing various CP input combinations. The accuracy of these RF runs was compared to a classification that simply employed satellite optical pictures.
Significant Findings of the Study
Accuracies
For classification runs that used all Stokes or all m-chi characteristics, a 95% overall mapping accuracy was attained. These accuracy levels were around 2% less than those of a classification made using optical vision and post-processing filtering.
Classification runs based on inputs of backscatter intensity underperformed classification trials based on Stokes and m-Chi parameter inputs to the RF classifier.
Parameters comparison
The RF classifier's rankings of variable relevance showed the contributions made by each input to the achievement of both general and crop-specific classifications.
S1 and S2 of the Stokes parameters were significant for all crops, particularly for barley, wheat, and canola.
S3 is highly important for detecting wheat and lentils, two crops with a lower organized canopy. Peas were an exception, requiring entire scattering (S0) to obtain excellent accuracy.
Conclusion
This research proved that crop classifications from RCM CP imaging could be made with high accuracy, supporting prior findings based on simulated CP data.
The potential of CP to monitor crop growth is supported by assessing the significance of the Stokes parameters to these classifications. Even though this research only included one test location, the results need to spur further attempts to use CP for operational mapping.
Reference
Laura Dingle Robertson, Heather McNairn, Xianfeng Jiao, Connor McNairn & Samuel O. Ihuoma (2022) Monitoring Crops Using Compact Polarimetry and the RADARSAT Constellation Mission. Canadian Journal of Remote Sensing. https://www.tandfonline.com/doi/full/10.1080/07038992.2022.2121271
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.