Downscaling and validating SMAP soil moisture using a machine learning algorithm over the Awash River basin, Ethiopia

Abstract

Microwave remote sensing instrument like Soil Moisture Active Passive ranging from 1 cm to 1 m has provided spatial soil moisture information over the entire globe. However, Soil Moisture Active Passive satellite soil moisture products have a coarse spatial resolution (36km x 36km), limiting its application at the basin scale. This research, subsequently plans to; (1) Evaluate the capability of SAR for the retrieval of surface roughness variables in the Awash River basin; (2) Measure the performance of Random Forest (RF) regression model to downscale SMAP satellite soil moisture over the Awash River basin; (3) validate downscaled soil moisture data with In-situ measurements in the river basin. Random Forest (RF) based downscaling approach was applied to downscale satellite-based soil moisture product (36km x 36km) to fine resolution (1km x 1km). Fine spatial resolution (1km) soil moisture data for the Awash River basin was generated. The downscaled soil moisture product also has a strong spatial correlation with the original one, allowing it to deliver more soil moisture information than the original one. In-situ soil moisture and downscaled soil moisture had a 0.69 Pearson correlation value, compared to a 0.53 correlation between the original and In-situ soil moisture. In-situ soil moisture measurements were obtained from the Middle and Upper Awash sub-basins for validation purposes. In the case of Upper Awash, downscaled soil moisture shows a variation of 0.07 cm3 /cm3, -0.036 cm3 /cm3, and 0.112 cm3 /cm3 with Root Mean Square Error, Bias error, and Unbiased Root Mean Square Error respectively. Following that, the accuracy of downscaled soil moisture against the Middle Awash Sub-basin reveals a variance of 0.1320 cm3 /cm3, -0.033 cm3 /cm3, and 0.148 cm3 /cm3 with Root Mean Square Error, Bias error, and Unbiased Root Mean Square Error respectively. Future studies should take into account the temporal domain of Soil Moisture Active Passive satellite soil moisture product downscaling over the study region.

Spatial Downscaling of Satellite-Based Soil Moisture Products Using Machine Learning Techniques: A Review

“…In ML-based downscaling models, typically, a relationship is established between SM with driving forces such as surface temperature, vegetation indices, surface albedo, soil texture, meteorological factors, and topographic variables [33,66,[98][99][100][101][102] in order to capture the soil hydrologic properties and processes. Therefore, data related to these driving forces are often used as ancillary variables in downscaling SM data.…”

“…The downscaled data showed r = 0.97 and RMSE = 0.048 m 3 /m −3 when compared against airborne SM retrievals. In addition, Bai et al [67], Wakigari and Leconte [114], and Sishah et al [66] used RF for downscaling SMAP data. Li et al [115] downscaled a 25 km resolution microwave-based surface SM dataset developed by fusing SMAP SM data with European Space Agency Climate Change Initiative (ESA-CCI) data using an RF algorithm.…”

“…One of the challenges in utilising optical/thermal-imagebased ancillary variables is the data gaps due to the cloud contamination. Some studies have introduced ancillary variables derived using synthetic-aperture radar (SAR) as a solution to overcome this issue [66,67]. Even though ML techniques have become a popular approach for downscaling SM from satellite observations (mainly C/X/L-band brightness temperatures for passive sensors and C-band backscattering coefficients for active sensors), in recent years, e.g., [68,69], comprehensive reviews on ML-based SM downscaling methods are limited in the existing literature.…”

Watershed scale soil moisture estimation model using machine learning and remote sensing in a data-scarce context

“…When 𝜃 𝐸𝑅𝑅𝑂𝑅 is closer to zero, the corresponding 𝜃 𝐷𝑊𝑆 approaches 𝜃 𝐹𝐼𝐸𝐿𝐷 and the downscaling results area satisfactory. Previous validation studies (Bai et al, 2019;Colliander et al, 2017;Liu et al, 2020;Singh et al, 2019;Sishah et al, 2023;Xu, 2019) used the correlation coefficient σ for analytical comparison between in-situ measurements and remote soil moisture estimates.…”

“…This makes them less suitable for agricultural, hydrological, and environmental applications requiring daily and high spatial detail information (Vergopolan et al, 2021). Several methods have been proposed to enhance the spatial resolution of remote soil moisture estimates through a process called "downscaling" (Abbaszadeh et al, 2019;Bai et al, 2019;Cui et al, 2019;Fang et al, 2019;Guevara & Vargas, 2019;Hernandez-Sanchez et al, 2020;Liu et al, 2020;Mao et al, 2019;Montzka et al, 2020;Peng et al, 2017;Shangguan et al, 2024;Sishah et al, 2023;Xu et al, 2024;Zhu et al, 2023). Recently, machine learning techniques such as random forest (Hengl et al, 2018) have achieved advancements in the downscaling of remote soil moisture estimates, either spatially (Bai et al, 2019;Chen et al, 2019;Zappa et al, 2019;Zhao et al, 2018) or temporally (Lu et al, 2015;Mao et al, 2019;Xing et al, 2017).…”