1. Introduction
Soil is a mixture of organic as well as inorganic parts and their amount change from place to place or within the same place (Jandl, Rodeghiero et al. 2014). Owing to this, estimation of soil components (both quantitative as well as qualitative) is a burdensome process (Gehl and Rice 2007). Some logical as well as coherent datum intended to obtain information regarding soil organic carbon content (SOC) estimation must be required for the purpose to optimize monitoring as well as mapping capacity (Angelopoulou, Tziolas et al. 2019). SOC holds an integral part on carbon cycle and approximately 1500 Gt of carbon is stored in soils at 1 m depth (Jobbágy and Jackson 2000), (Scharlemann, Tanner et al. 2014). Additionally, SOC is a part of Organic Matter (OM) and has influence on physical, chemical, biological characteristics of soil ecosystem (Ontl and Schulte 2012). Eswaran et al., (Eswaran, Van Den Berg et al. 1993) stated some hurdles to estimate correct global carbon content. These are because of high spatial changeability of soil organic carbon, variableness of soil kinds which contains unpredictable estimates, inaccessibility of valid data as well as changes in plants and land usage.
Various traditional processes were used for SOC monitoring but they are laborious as well as expensive (Omran 2017). Studies were conducted to explore and apply other revolutionary procedures for all types of environments and all kinds of soil (Jandl, Rodeghiero et al. 2014) i.e., use of Remote Sensing (RS) practice is efficient, low cost in addition to environmentally sound aspect for analysis of various soil characteristics (Xu, Smith et al. 2017) such as SOC estimation etc. (Vaudour, Gilliot et al. 2016). Visible near infrared-shortwave infrared (VNIR-SWIR) sensors (find RS usages) work on principle of energy-matter interaction (Schwartz, Ben-Dor et al. 2012). Part of electromagnetic radiations (falling on soil exterior) that is reflected from soil surface is recorded as a spectrum which is enough to produce information (both qualitative as well as quantitative) regarding soil characteristics (Nocita, Stevens et al. 2015). In VNIR-SWIR, characteristic vibrations take place (Mohamed, Saleh et al. 2018) in visible region (400-700nm) i.e., electronic transitions occur which produce absorption bands linked to chromophore while on the other hand, in NIR-SWIR (700-2500 nm) weak overtones or such vibrations take place owing to extending as well as bending of some bonds such as N-H, O-H, as well as C-H bonds etc. (Rossel, Walvoort et al. 2006), (Stuart 2004). Association amongst SOC as well as electromagnetic radiations in VNIR, SWIR region has already been reported in laboratory conditions (Bartholomeus, Schaepman et al. 2008), (Stevens, Nocita et al. 2013), (Nocita, Stevens et al. 2013). In addition to laboratory trials, numerous studies have also been conducted in real field conditions founded on manned as well as unmanned airborne system, and satellite platforms (Gomez, Rossel et al. 2008). However, these practices have restrictions for direct SOC estimation including vegetation cover, soil wetness etc. (Nocita, Stevens et al. 2013), (Bartholomeus 2009). Furthermore, there is a demand of some multivariate statistical procedures called Chemometrics to associate spectral signatures with soil characteristics (Geladi 2003).
One of the most widely employed practice is application of partial least squares regression (PLSR) for reporting direct association among variables (Peng, Shi et al. 2014). That’s the basic reason of increased usage of machine learning algorithms for correlation procedures (Stenberg, Rossel et al. 2010), (Liakos, Busato et al. 2018).
This review article emphasizes on current research of remote sensing procedures, state-of- art procedures as well as instruments for current SOC estimation.