| Summary, etc |
A key indicator of soil quality is Soil organic carbon (SOC) which is affected by the various factors including topography and land and crop management, precise estimation of SOC will be useful in understanding the influence of agro-ecosystem on it. The conventional estimation in laboratory is laborious, expensive and time consuming. Alternate method of SOC estimation using remote sensing and geographic information systems (GIS) is non-intrusive, low cost and provides spatially continuous information over large areas on a repetitive basis. In the present study, SOC content was attempted to estimate using Sentinel-2 datPa (290 km swath, 5 days revisit time and 10 m spatial resolution). About 10 random soil samples at 0 – 15 cm depth from 10 m ×10 m area of identified each paddy field were collected and prepared one composite soil sample, a total of 100 composite soil samples were collected from the identified paddy field at different slopes [Plain (<2% slope), Gently undulating (2-5% slope), Rolling (10-15% slope), Hilly (16-30% slope)] of Bhoirymbong block, Meghalaya. The sampling sites were recorded using a handheld Global Positioning System unit (GERMIN GPSMAP-64). The soil samples were divided into two sets (i) 75% of the samples for SOC prediction and 25% of the samples for model validation. The soil samples were processed and analyzed the properties (eg.SOC, pH, EC, available N, available p, available K) using standard protocols. The cloud free October and November SENTINAL -2 data of was accessed from the official website of Copernicus Open Access Hub. The 15 different indices such as (i) R-NIR base indices [Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Soil Adjusted Vegetation Index (SAVI), Green Red Vegetation Index (GRVI), Enhanced Vegetation Index (EVI), Modified Soil Adjusted Vegetation Index (MSAVI), Vegetation Index (V), Renormalized Difference Vegetation Index (RDVI), Second Modified Soil Adjusted Vegetation Index (MSAVI2), Weighted Difference Vegetation Index (WDVI)] (ii) G-R base indices [Brightness Index (BI), Second Brightness Index (BI2), Redness Index (RI), Soil Brightness Index (SBI), Colour Index (CI)] were derived for SOC prediction. <br/>The result is described that the indices (i.e. NDVI, GNDVI, SAVI, GRVI, EVI, MSAVI, WDVI, BI, BI2, RI, V, SBI, RDVI, MSAVI2 and CI) values are varied from 0.56-0.87, 0.46-0.76, 0.84-1.31, 0.001-0.40, 1.14-3.00, 0.84-1.31, 0.007-472.00, 233.56-582.43, 568.18-1320.7, 0.0002-0.001, 3-14, 1632.7-3865.6, 26.09-54.11, 1.43-1.87, -0.40-0.001 for the month of October respectively, and 0.28-0.89, 0.40- 0.77, 0.42-1.33, 0.14- 0.37, 0.52-2.78, 0.42- 1.33, 15-1156, 200.24- 922.10, 135.81-615.51, 0.0009-0.002, 1-16, 1995.80-44490.6, 16.99-49.31, 0.88-1.88 and 0.37-0.14 for the month of November respectively. Among Red-NIR base indices, GRVI and WDVI were found the lowest (0.26±0.06 for October, 0.02±0.07 for November) and highest value (173.87±99.02, 415.10±277.19) respectively, whereas CI and SBI were found the lowest (-0.06±0.06 for October, -0.02±0.07 for November) and highest (2569.60±471.65 for October, 2839.80±406.38 for November) respectively among the Green-Red based indices. The pH, EC, SOC, available N, available P and available K values varied from 4.99 to 5.55, 5 to 20 μS/m, 1.86 (Hilly, 16-30% slope) to 2.19% (Plain, <2% slope), 198.56 to 307.76 kg/ha, 8.20 to 25.56 kg/ha, and 102.3 to 394.45 kg/ha respectively. The highest correlation coefficients (r) of the following indices between SOC and NDVI (r = 0.72), SAVI (r = 0.72), MSAVI (r=0.72) and V (r = 0.72) for the month of October, whereas, SBI (r = 0.33), BI (r = 0.275), BI2 (r = 0.275) and RI (r = -0.266) for the month of November. The best SLR models for SOC prediction was SOC (%) = -0.375+3.315*NDVI followed by SOC (%) = -0.026+3.445*GNDVI or MSAVI (SOC (%) = -0.375+2.21*MSAVI in the October month, however, green-NIR base independent variable could only able to explained only 12.3% of SOC. The MLR model for SOC prediction was SOC (%) = 8.41+20.55*NDVI+ 1.14*GNDVI-12.34*MSAVI2+0.04*EVI(PCR) followed by SOC (%) = -0.99+1.04*NDVI+1.10*GNDVI+0.76*MSAV I2+0.14*EVI (PLSR), SOC (%) = 8.82+21.51*NDVI-0.98*GNDVI-1*MSAVI2 +0.002*EVI (OLS). <br/>It is concluded that the Red-NIR base indices such as NDVI, GNDVI, MASAVI2 and EVI are sensitive to highly fragile topography with vegetation system especially in the month of October. The Green-NIR indices are the best suitable for SOC prediction of the low crop area and no crop areas for fragile topographical system. Further it is also concluded that the best suitable SOC prediction model of the fragile topographical system are principal components regression model i.e. <br/>For October, SOC (%) =8.41+20.55*NDVI+1.14*GNDVI-12.34*MSAVI2+0.04.98*EVI <br/>For November, SOC (%) = 1.76+0.0003*BI-85.77*RI+0.0001*SBI <br/> |
