Geodaysit 2023

Investigating the Correlation between Sentinel-2 Multispectral Images and Ground-Based Field Measurements of Soil Moisture (Case Study: Mendatica, Liguria, Italy)
06-13, 16:45–17:00 (Europe/London), Sala Biblioteca @ PoliBa

Surface Soil Moisture (SSM) is an essential climate variable that links the atmospheric and surface processes, controlling the exchange of water, partitioning the available energy at the ground surface and biochemical process. SSM plays also a crucial role in controlling hydro-geological hazards, like rainfall-triggered landslides.
SSM can be monitored using various methods: ground-based measurements, proximal methods, or air-borne/ space-borne remote sensing. Traditional methods are mainly ground-based measurements through contact sensors; they provide accurate but single-point measurements and require manual placement and intensive maintenance, especially in large-scale studies. Because SSM is a heterogeneous variable in terms of space and time, data acquisition with traditional single-point measurement methods is very limited, especially at large scale.
Advances in satellite Remote Sensing (RS) bring the possibility of continuous land surface observation and characterization over time. In addition to the geometric condition, and optical and mineral properties of the earth surface, SSM is one of the influential factors that control the radiation emitted from the earth’s surface. All parts of the electromagnetic (EM) spectrum that are normally used for earth observation can be used for quantitative SSM extraction. Considering the potential of penetration depth of the EM wavelength, RS methods can be classified into three categories: thermal, microwave, and optical RS. Thermal RS can be used individually or in combination of vegetation indices, like the Crop Water Stress Index (CWSI). The acquisition of thermal data has high costs, and, in addition, the differentiation between soil temperature and tree canopy temperature is not easily achieved. Most globally available SSM products are derived from microwave RS, due to the ability of microwave radiation to penetrate cloud cover, but they are highly sensitive to surface roughness and have coarse spatial resolution, making them inefficient for studies at small scale. Optical RS in the visible, near-infrared, and shortwave infrared ranges measures the reflected radiation from the earth surface, which can be related as a function of soil moisture to provide very high spatial resolution data.
In the present study, the potential of multispectral satellite images acquired by Sentinel-2 (S-2) for SSM extraction is investigated. For this purpose, a yearly dataset of hourly SSM measurements, acquired at four different depths (-10cm, -35cm, -55cm, -85cm) from a monitoring network in Mendatica (Liguria, Italy) from 1st of July 2020 to 30th of June 2021, was used to look for correlation with S-2 images. Data acquired by the sensors were previously calibrated, taking into account the soil-specific characteristics of the areas (Bovolenta et al., 2020), and the reliability of the dataset was verified. After performing the required preprocessing on satellite images, the correlation coefficients between each band of S-2 images and ground-based measurements were calculated. The results represent the potential of each band or a combination of them to estimate SSM from RS through linear estimators.