UTILISATION OF HYBRID POLARIMETRIC SAR DATA FOR NATURAL RESOURCE MONITORING – A NEW DIMENSION

Natural resource mapping and its monitoring are the key elements for sustainable rural development. Satellite based remote sensing with Synthetic Aperture Radar (SAR) is widely accepted for natural monitoring because of its synoptic view and better temporal coverage due to its unique capabilities to provide day and night measurements, almost independent of atmospheric/climatic conditions. ISRO’s Radar Imaging Satellite (RISAT-1) with hybrid polarimetry added a new dimension to explore its utilisation for various applications including natural resource monitoring and further to its management with the inclusion of Geographical Information System. A pilot study has been carried out by considering hybrid polarimetric single look complex SAR datasets acquired over Annaram village in Mahabubnagar district. In the current work, hybrid polarimetric data from RISAT-1 have been analysed and processed for different standard radar targets like corner reflectors and various land cover features using Stokes parameter based decomposition techniques like M-delta, M-chi and M-alpha. The importance of Stokes vector and its derived value added parameter based decompositions are well demonstrated statistically as they aid in characterising target properties based on dominant basic scattering mechanism. Derived Stokes Polarimetric parameters that are decomposed from hybrid polarimetric data are elucidated as the essential basis for better target discrimination and classification. The preliminary results and experience gained in implementation of different emerging hybrid polarimetric decompositions for deriving characteristics of scattering mechanism involved in natural and man-made targets towards sustainable rural development are discussed in this paper.


Introduction
For the next few decades, the vision and plan of a developing nation is going to be socioeconomic security which will be one of the pillars and the entire space technology would be application-driven instead of technology-driven.
The gravity of the present situation is to monitor the natural resources to have sustainable rural development in rapidly developing nations like India and to explore various means and ways in the sphere of remote sensing with geospatial technology towards further development. Usually, during monsoon period (June to October), specifically in India, utilisation of optical remote sensing data is highly hampered by clouds. In this light, microwave remote sensing plays a significant role in supervising the available natural resources, especially during monsoon seasons due to its unique capability to penetrate through the clouds. In Indian scenario, Synthetic Aperture Radar (SAR) data with varying incidence angles, frequency bands, polarisations, resolutions and swath customised for various land, coastal and oceanic applications is expected to enhance the scope of microwave remote sensing primarily for natural resource monitoring and flood mapping during monsoon season.
Over past many years, SAR has received considerable attention for exploiting many applications due to its distinctive capabilities to provide day and night and all-weather imaging system. With recent advances in polarimetry (Boerner W.M, 1998), SAR with hybrid-polarity architecture (R. K Raney, 2006), transmitting circular polarisation (H±iV) and receiving two orthogonal mutually coherent polarisations, which is an excellent demonstration of compact polarimetry(J.C. Souyris, 2005) (R. Touzi, 2009) leading to reduced downlink data rate, bare minimum sensitivity to cross-talks between the channels(Beckmann, 1968) and simple hardware realisation without any trade-off to the swath and resolution(T.L. Ainsworth, 2007). In view of the above capabilities, there is a high demand to understand and further explore the characteristics of polarimetric SAR data to support many terrestrial and oceanic applications.

Generation of Stokes Parameters
In response to left/right circularly polarised transmitted signal, it provides Stokes parameters by using data received in two mutually orthogonal channels. Based on the derived Stokes parameters, several useful quantitative derivatives equation [2] to [5] are extracted like, Degree of Polarisation (DoP) (m, representative of polarised and diffused scattering), Circular Polarisation Ratio (CPR, representative of scattering associated with dihedral reflection) and relative phase (δ) between the two linear E-vectors of the backscattered field (an indicator of double bounce scattering).The polarisation state of an electromagnetic wave can be mainly characterised by the DoP, elipticity and relative phase expressed in terms of Stokes parameters.

Degree of Polarisation (m):
Degree of polarisation varies from 0 to 1. It is the ratio between polarized power to the total received power of partially polarised wave which is an indicator of polarised and diffused scattering, fundamentally related to entropy. [2] Relative Phase Difference (δ δ δ δ δ): Its values variy from -180 0 to +180 0 and is well known for single bounce, the backscattered wave is a mirror image of transmitted wave and for double bounce, the backscattered wave is the same as that of transmitted wave. [3]

Results and Discussion
RISAT-1 data acquired over Annaram village and its environs including Shadnagar are considered for implementing polarimetric decompositions, to interpret scattering mechanism over various standard and distributed natural targets where the ground truth is considerably available. Figure 2 shows surface scattering, double bounce and volumetric scattering for various targets using m-delta decomposition.  It is quite interesting to note that barren land with shrubs signifying volume scattering (green) in 4d has been fully cleared (4e and 4f ) and modified into residential plots during 2016.
However, the water body (4g) near Shadnagar remained unaltered over these three years, except for the increase in built-up area around its surroundings (4i). Table 3  these corner reflectors as shown in Figure 5 demonstrate the correctness of implementation of decomposition methodology for land use land cover monitoring.

Table 3: Statistical Representation of Scattering Mechanisms on Various Targets
It is observed that Degree of Polarisation is close to 1 as expected for corner reflectors because they have the tendency to preserve the polarisation information based on its structure and its scattering properties. The relative phase is observed to be close to -900 because of double bounce effect which is predominant in dihedral structures and +900 due to odd bounce scattering mechanism persisting in trihedral corner reflector. Slight difference in characteristics of Stokes parameters can be accounted for CR orientation.

Conclusion
In this current work, hybrid polarimetric data of RISAT-1 have been analysed and processed for various land cover features using Stokes parameter based decomposition techniques. The scattering mechanisms derived using decomposition techniques are also validated using standard radar calibration target like corner reflectors. The experience gained in the implementation and analysis of hybrid polarimetric data will help in developing a methodology for classifying natural targets based on various scattering mechanisms. Thus, hybrid polarimetric decompositon methodology will be highly constructive and provides one of the best solutions in monitoring natural resources, periodically leading to sustainable rural development in a structured way, especially during monsoon seasons.The hybrid polarimetric data can be ordered systematically through NRSC data centre. As soon as the data are acquired, it can be processed, decomposed and classified regularly and then related with geo-spatial information to access and monitor sustainable rural development. The potential scope of polarimetric decomposition for target classification can be extended for future national and international space-borne SAR sensors like RISAT-1A, NASA ISRO Synthetic Aperture Radar (NISAR) and Canada's RADARSAT Constellation Mission (RCM), etc., which are also operating in hybrid polarimetry.