Impact of various horticultural farming practices on the spatial variability of Soil Organic Carbon in a small Mediterranean watershed using spectrometric measurements and Sentinel data.

Perennial crops such as orchards and vineyards are typical mediterranean crops which present a high variability in water and inter-row management, both having strong impacts on soil organic carbon (SOC) fate. The different irrigation options are drippers, micro-sprinklers or flood irrigation. The inter-rows can be grassed to facilitate the pruning, harvest or various operations in the crops, or soil can be ploughed to avoid water competition with trees. Assessing SOC variability at territorial scales is a real challenge to promote practices that will increase SOC storage and therefore alleviate climate change and improve the soil health. This study aims to assess (i) the intra-field variability of SOC in orchards and vineyards with various inter-row and irrigation practices from lab spectral measurements, and (ii) the spatial variability of SOC at the watershed scale according to different agricultural practices using Sentinel data. Soil samples were collected over a two-year period in 2024 and 2025 on 17 plots covering a large variability of irrigation and inter-row management in a small watershed in southeastern France (Ouvèze-Ventoux WS). Physical chemical analyses were made on air-dried, 2mmn sieved soil samples. Spectral measurements were made using a lab spectrometer (SM-3500) in the range of visible near-infrared to mid-infrared (350-2500 nm). Sentinel 1 and 2 images acquired over the Ouveze-Ventoux WS were selected in winter period for the two studied years, when NDVI presented the lowest values. Three methods were compared to estimate SOC variability: PSLR, DNN and cubist. The results from lab spectrometer measurements showed a high accuracy to obtain SOC (R²=0.89, correlation between observed and simulated SOC) allowing to quantify the impact of row management on the SOC variability. This accuracy decreased if only Sentinel data are used (as expected) but the correlations between observations and simulations remained significant. The cubist model appeared the most effective to map SOC at the watershed scale Such approaches based on spectral measurements performed at different scales represent inexpensive and useful tools to assess the spatial variability of the soil properties which can be applied in various environmental contexts.