CORRELATION BETWEEN SOIL CARBON STOCK AND RGB VEGETATION INDEX OBTAINED FROM IMAGES ACQUIRED BY AN UNMANNED AERIAL VEHICLE

Abstract

Remote sensing using Unmanned Aerial Vehicles (UAVs) has become an effective tool for monitoring vegetated areas at high spatial resolution. This study aimed to correlate Soil Carbon Stock (SCS) with Vegetation Indices (VIs) derived from RGB images of different vegetation covers, obtained using an optical sensor mounted on a UAV, based on four indices from the visible portion of the electromagnetic spectrum (ExG, GLI, NGRDI, and VARI). The methodology comprised three stages: (1) soil sampling and acquisition of orthomosaics in the field; (2) laboratory determination of organic carbon using Walkley–Black (WB) and CHNS methods, along with digital image processing; and (3) Pearson correlations (r) between SCS and the mean response of the indices within the delimited areas. Mean SCS (CHNS) values were 40.94, 54.90, and 99.93 Mg ha-1 for wetland, pasture, and forest, respectively; while SCS (WB) values were 52.61, 48.20, and 68.74 Mg ha-1. Correlations between SCS (CHNS) and the vegetation indices were strong for VARI (r = 0.732) and NGRDI (r = 0.713), moderate for GLI (r = 0.519), and weak for ExG (r = –0.040). For SCS (WB), correlations were strong for GLI (r = 0.922) and NGRDI (r = 0.711), and moderate for VARI (r = 0.694) and ExG (r = 0.596). These results highlight the potential of UAVs for non-destructive, low-cost environmental monitoring with high spatial and temporal resolution. This approach enables frequent surveys to monitor seasonal dynamics, detect degraded areas, and support soil carbon management strategies.