The paper introduces a novel approach for estimating soil moisture in vegetated surfaces, specifically focusing on sugarcane crops throughout various growth stages in agriculture applications. While existing models typically address bare soil scenarios, this model utilizes data from P-, L-, and C-band Synthetic Aperture Radar (SAR) to estimate soil moisture. The semi-empirical Dubois model forms the basis of the proposed model, which has been adapted to accommodate multiband operation and crop height variations. Synthetic datasets are generated using the adjusted model to train two neural networks incorporated into the overall model. Additionally, a linear expression for estimating crop height is integrated into the model. The model is validated in an Experimental Site at the School of Agricultural Engineering, UNICAMP, and an independent area at the Sugarcane Technology Center in Piracicaba, Brazil. The model utilizes a multiband drone-borne SAR system with a 3-meter image resolution and radiometric accuracy of 0.5 dB. The results indicate that the model can estimate soil moisture with root-mean-square errors of 0.05 cm3.cm-3 (5 vol. %) across crop heights ranging from zero to 2.5 meters.
Soil moisture estimation is an important task to enable precision agriculture in creating optimal plans for irrigation, fertilization, and harvest. It is common to utilize statistical and machine learning models to estimate soil moisture from traditional data sources such as weather forecasts, soil properties, and crop properties. However, there is a growing interest in utilizing aerial and geospatial imagery to estimate soil moisture. Although these images capture high-resolution crop details, they are expensive to curate and challenging to interpret. Imagine, an AI-enhanced software tool that predicts soil moisture using visual cues captured by smartphones and statistical data given by weather forecasts. This work is a first step towards that goal of developing a multi-modal approach for soil moisture estimation. In particular, we curate a dataset consisting of real-world images taken from ground stations and their corresponding weather data. We also propose MIS-ME - Meteorological & Image based Soil Moisture Estimator, a multi-modal framework for soil moisture estimation. Our extensive analysis shows that MIS-ME achieves a MAPE of 10.14%, outperforming traditional unimodal approaches with a reduction of 3.25% in MAPE for meteorological data and 2.15% in MAPE for image data, highlighting the effectiveness of tailored multi-modal approaches. Our code and dataset will be available at https://github.com/OSU-Complex-Systems/MIS-ME.git.
Characterizing soil moisture (SM) around drip irrigation pipes is crucial for precise and optimized farming. Machine learning (ML) approaches are particularly suitable for this task as they can reduce uncertainties caused by soil conditions and the drip pipe positions, using features extracted from relevant datasets. This letter addresses local moisture detection in the vicinity of dripping pipes using a portable microwave imaging system. The employed ML approach is fed with two dimensional images generated by two different microwave imaging techniques based on spatio-temporal measurements at various frequency bands. The study investigates the performance of K-Nearest Neighbor (KNN) and Convolutional Neural Networks (CNN) algorithms for moisture classification based on these images, both before and after performing soil clutter reduction. We also explore the potentials of CNN and KNN for moisture estimation around the plant roots and in the presence of pebbles. The results demonstrate the more accurate moisture estimation using CNN when it is applied after clutter reduction considering back projection algorithm (BPA) as the imaging technique.
Accurate soil moisture information is crucial for developing precise irrigation control strategies to enhance water use efficiency. Soil moisture estimation based on limited soil moisture sensors is crucial for obtaining comprehensive soil moisture information when dealing with large-scale agricultural fields. The major challenge in soil moisture estimation lies in the high dimensionality of the spatially discretized agro-hydrological models. In this work, we propose a performance-triggered adaptive model reduction approach to address this challenge. The pro- posed approach employs a trajectory-based unsupervised machine learning technique, and a prediction performance-based triggering scheme is designed to govern model updates adaptively in a way such that the prediction error between the reduced model and the original model over a prediction horizon is maintained below a predetermined threshold. An adaptive extended Kalman filter (EKF) is designed based on the reduced model for soil moisture estimation. The applicability and performance of the proposed approach are evaluated extensively through the application to a simulated large-scale agricultural field.
Soil moisture is a key hydrological parameter that has significant importance to human society and the environment. Accurate modeling and monitoring of soil moisture in crop fields, especially in the root zone (top 100 cm of soil), is essential for improving agricultural production and crop yield with the help of precision irrigation and farming tools. Realizing the full sensor data potential depends greatly on advanced analytical and predictive domain-aware models. In this work, we propose a physics-constrained deep learning (P-DL) framework to integrate physics-based principles on water transport and water sensing signals for effective reconstruction of the soil moisture dynamics. We adopt three different optimizers, namely Adam, RMSprop, and GD, to minimize the loss function of P-DL during the training process. In the illustrative case study, we demonstrate the empirical convergence of Adam optimizers outperforms the other optimization methods in both mini-batch and full-batch training.
Soil moisture is a crucial hydrological state variable that has significant importance to the global environment and agriculture. Precise monitoring of soil moisture in crop fields is critical to reducing agricultural drought and improving crop yield. In-situ soil moisture sensors, which are buried at pre-determined depths and distributed across the field, are promising solutions for monitoring soil moisture. However, high-density sensor deployment is neither economically feasible nor practical. Thus, to achieve a higher spatial resolution of soil moisture dynamics using a limited number of sensors, we integrate a physics-based agro-hydrological model based on Richards' equation in a physics-constrained deep learning framework to accurately predict soil moisture dynamics in the soil's root zone. This approach ensures that soil moisture estimates align well with sensor observations while obeying physical laws at the same time. Furthermore, to strategically identify the locations for sensor placement, we introduce a novel active learning framework that combines space-filling design and physics residual-based sampling to maximize data acquisition potential with limited sensors. Our numerical results demonstrate that integrating Physics-constrained Deep Learning (P-DL) with an active learning strategy within a unified framework--named the Physics-constrained Active Learning (P-DAL) framework--significantly improves the predictive accuracy and effectiveness of field-scale soil moisture monitoring using in-situ sensors.
The microwave imaging system(MIS) stands out among prominent imaging tools for capturing images of concealed obstacles. Leveraging its capability to penetrate through heterogeneous environments MIS has been widely used for subsurface imaging. Monitoring subsurface drip irrigation(SDI) as an efficient procedure in agricultural irrigation is essential to maintain the required moisture percentage for plant growth which is a novel MIS application. In this research, we implement a laboratory-scale MIS for SDI reflecting real-world conditions to evaluate leakage localization and quantification in a heterogeneous area. We extract a model to quantify the moisture content by exploiting an imaging approach that could be used in a scheduled SDI. We employ the subspace information of images formed by back projection and Born approximation algorithms for model parametrization and estimate the model parameters using a statistical curve fitting technique. We then compare the performance of these imaging techniques in the presence of environmental clutter such as plant roots and pebbles. The proposed approach can well contribute to efficient mechanistic subsurface irrigation for which the local moisture around the root is obtained noninvasively and remotely with less than 20% estimation error.
Improving the accuracy of soil moisture estimation is required for advancing irrigation scheduling and water conservation efforts. Central to this task are soil hydraulic parameters, which govern moisture dynamics but are rarely known precisely and must therefore be inferred from observational data. In large-scale agricultural fields, estimating the complete set of these parameters is often impractical due to the sparse and noisy nature of available measurements. To address this challenge, this work develops a framework that uses sensitivity analysis and orthogonal projection to identify parameters that are both reliably estimable from available data. These parameters, together with the spatial distribution of soil moisture, are jointly estimated by assimilating observational data into a cylindrical-coordinate version of the Richards equation using an extended Kalman filter. The soil moisture measurements are obtained from microwave remote sensors mounted on center pivot irrigation systems - an emerging and practical technology for capturing field-scale variability. Numerical simulations and field experiments conducted on a large-scale site in Lethbridge, Alberta, Canada, demonstrate that the proposed method improves soil moisture estimation accuracy by 24-43% and enhances predictive model performance by 50%. Furthermore, the estimated parameters - particularly saturated hydraulic conductivity - exhibit good agreement with experimental measurements.
Synergetic use of sensors for soil moisture retrieval is attracting considerable interest due to the different advantages of different sensors. Active, passive, and optic data integration could be a comprehensive solution for exploiting the advantages of different sensors aimed at preparing soil moisture maps. Typically, pixel-based methods are used for multi-sensor fusion. Since, different applications need different scales of soil moisture maps, pixel-based approaches are limited for this purpose. Object-based image analysis employing an image object instead of a pixel could help us to meet this need. This paper proposes a segment-based image fusion framework to evaluate the possibility of preparing a multi-scale soil moisture map through integrated Sentinel-1, Sentinel-2, and Soil Moisture Active Passive (SMAP) data. The results confirmed that the proposed methodology was able to improve soil moisture estimation in different scales up to 20% better compared to pixel-based fusion approach.
We present a methodology based on interferometric synthetic aperture radar (InSAR) time series analysis that can provide surface (top 5 cm) soil moisture (SSM) estimations. The InSAR time series analysis consists of five processing steps. A co-registered Single Look Complex (SLC) SAR stack as well as meteorological information are required as input of the proposed workflow. In the first step, ice/snow-free and zero-precipitation SAR images are identified using meteorological data. In the second step, construction and phase extraction of distributed scatterers (DSs) (over bare land) is performed. In the third step, for each DS the ordering of surface soil moisture (SSM) levels of SAR acquisitions based on interferometric coherence is calculated. In the fourth step, for each DS the coherence due to SSM variations is calculated. In the fifth step, SSM is estimated by a constrained inversion of an analytical interferometric model using coherence and phase closure information. The implementation of the proposed approach is provided as an open-source software toolbox (INSAR4SM) available at www.github.com/kleok/INSAR4SM. A case study over an arid region in California/Arizona is presented. The proposed workflow was applied in Sentinel- 1 (C-band) VV-polarized InSAR observations. The estimated SSM results were assessed with independent SSM observations from a station of the International Soil Moisture Network (ISMN) (RMSE: 0.027 $m^3/m^3$ R: 0.88) and ERA5-Land reanalysis model data (RMSE: 0.035 $m^3/m^3$ R: 0.71). The proposed methodology was able to provide accurate SSM estimations at high spatial resolution (~250 m). A discussion of the benefits and the limitations of the proposed methodology highlighted the potential of interferometric observables for SSM estimation over arid regions.
We develop a deep learning based convolutional-regression model that estimates the volumetric soil moisture content in the top ~5 cm of soil. Input predictors include Sentinel-1 (active radar), Sentinel-2 (optical imagery), and SMAP (passive radar) as well as geophysical variables from SoilGrids and modelled soil moisture fields from GLDAS. The model was trained and evaluated on data from ~1300 in-situ sensors globally over the period 2015 - 2021 and obtained an average per-sensor correlation of 0.727 and ubRMSE of 0.054, and can be used to produce a soil moisture map at a nominal 320m resolution. These results are benchmarked against 13 other soil moisture works at different locations, and an ablation study was used to identify important predictors.
In this contribution, we investigate the potential of hyperspectral data combined with either simulated ground penetrating radar (GPR) or simulated (sensor-like) soil-moisture data to estimate soil moisture. We propose two simulation approaches to extend a given multi-sensor dataset which contains sparse GPR data. In the first approach, simulated GPR data is generated either by an interpolation along the time axis or by a machine learning model. The second approach includes the simulation of soil-moisture along the GPR profile. The soil-moisture estimation is improved significantly by the fusion of hyperspectral and GPR data. In contrast, the combination of simulated, sensor-like soil-moisture values and hyperspectral data achieves the worst regression performance. In conclusion, the estimation of soil moisture with hyperspectral and GPR data engages further investigations.