Large-scale Soil Moisture Monitoring Research Articles

Downscaling soil moisture in regions with high soil heterogeneity: the solution based on ensemble learning with sequential and parallel learner

Soil moisture plays an important role in the water and heat exchanges between the land surface and atmosphere, and it has great importance for agricultural production, ecological planning, and water resources management. Although microwave remote sensing has been widely used in large-scale soil moisture monitoring, the accuracy of the downscaled retrieval results cannot be guaranteed for regions with high vegetation coverage and high soil heterogeneity. To address these challenges, this study built soil moisture indice set based on MODIS and elevation data by calculating the Pearson correlation coefficient (R) and Maximum Information Coefficient (MIC), then constructed decision tree models (Gradient Boosting Decision Tree and Random Forest) about the indice set and low-resolution Soil Moisture Active Passive (SMAP) by using two ensemble learning methods (Bagging and Boosting). The models were applied to the high-resolution soil moisture indices in Jilin Province for the years 2017 to 2020 to generate 1 km-resolution products. In the validation process, Triple Collocation Analysis (TCA), comparison of soil moisture maps with coarse and fine resolution, and in-situ measurements in Lishu County, Tongyu County, and Jilin City were used to evaluate the differences between downscaling soil moisture results and ground observations at network, seasonal and point scales. The results were as follows: (1) The correlation coefficient (R2) calculated by the TCA method was 0.733 (GBDT_36km) > 0.649 (RF_36km), and the error variance was 0.0004 (GBDT_36km) < 0.00058 (RF_36km). (2) R at network scale was 0.798 (GBDT_SM) > 0.662 (RF_SM), RMSE was 0.040 (GBDT_SM) < 0.044 (RF_SM), the point scale R was 0.864 (GBDT_SM) > 0.833 (RF_SM), RMSE was 0.029 (GBDT_SM) < 0.039 (RF_SM). The R in four stages of the growth period was GBDT_SM > RF_SM, RMSE was GBDT_SM < RF_SM. In conclusion, the GBDT and RF models can reliably downscale soil moisture in Jilin Province, and the Boosting ensemble learning method represented by GBDT had a better estimation performance.

Adaptive proper orthogonal decomposition for large scale reliable soil moisture estimation

A major challenge in automatic irrigation of extensive agricultural fields is large scale soil moisture monitoring. Proper orthogonal decomposition (POD) is a widespread data-driven dimension reduction technique which can be combined with QR pivoting method for estimation of high-dimensional signals and optimal sensor placement. However, it requires computation of tailored basis functions which should be extracted from known training data. This is feasible for problems with constant features such as face recognition. However, using fixed bases (and probably fixed sensor selection) may not be an appropriate approach for estimation of signals with time-variant features. This paper demonstrates that the POD algorithm can be implemented adaptively in a reliable fashion to address this issue. The probability of the mean estimation error to remain within a predetermined threshold is studied and theoretical results on permitted update intervals are provided. The adaptive tailored basis and sensor selection increases suitability to large scale applications with time-variant features. The performance of the algorithm on a NASA soil moisture dataset is analyzed.

Soil Moisture Estimation With SVR and Data Augmentation Based on Alpha Approximation Method

Soil moisture content is an important parameter in hydrological, meteorological, and agricultural applications. Balenzano et al. proposed the alpha approximation method in 2011 for solving some complex issues during the retrieval of soil moisture over agricultural crops with synthetic aperture radar data. However, determining the constraints and solving the underdetermined system of equations in this method add new challenges. Considering the questions of constraints and underdetermined system of equations, the alpha approximation method is used to augment the measured data, and can avoid solving the underdetermined system of equations with constraints directly. Then, these data are applied in a support vector regression machine for soil moisture estimation. It is found that when an optimal model is determined, the method proposed in this article is superior to the direct use of the alpha approximation method, and the root-mean-squared error (RMSE) decreased from 0.0775 to 0.0339 and R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> increased from 0.0467 to 0.6491. In addition, the method obtained a good result from a data set collected that included a different growing period of crops by changing the standardized method from StandardScaler to Scale, where the RMSE is 0.0501 and R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> is 0.3204. This indicates the good generalization capability of this method. In conclusion, the proposed method solves the two questions effectively and provides a potential way for long-time or large-scale soil moisture monitoring with much less in situ measurements.

Modeling transient soil moisture dichotomies in landscapes with intermixed land covers

Large-scale in situ soil moisture monitoring networks are becoming increasingly valuable research tools, but existing networks feature almost exclusive deployment of stations in grassland vegetation. These grassland soil moisture observations are unlikely to adequately represent the real soil moisture patterns in landscapes with intermixed land cover types. Here we demonstrate the severity of the problem for one particular landscape and introduce a flexible new method for solving the problem. The specific objectives of this study were (i) to compare root-zone soil moisture dynamics of two dominant vegetation types across Oklahoma, grassland (observed) and winter wheat cropland (simulated); (ii) to relate the soil moisture dynamics of grassland and cropland vegetation using an artificial neural network (ANN) as an observation operator; and (iii) to use the resulting ANN to estimate the soil moisture spatial patterns for a landscape of intermixed grassland and wheat cropland. Root-zone soil moisture was represented by plant available water (PAW) in the top 0.8 m of the soil profile. PAW under grassland was calculated from 18 years of soil moisture observations at 83 stations of the Oklahoma Mesonet, whereas PAW under winter wheat was simulated for the same 83 locations using a calibrated and validated soil water balance model. Then, we trained an ANN to reproduce the simulated PAW under winter wheat using only six inputs: day of the year, latitude and longitude, measured PAW under grassland, and percent sand and clay. The resulting ANN was used, along with grassland soil moisture observations, to estimate the detailed soil moisture pattern for a 9 × 9 km2 Soil Moisture Active Passive (SMAP) grid cell. The seasonal dynamics of root-zone PAW for grassland and winter wheat were strongly asynchronous, so grassland soil moisture observations rarely reflect cropland soil moisture conditions in the region. The simple ANN approach facilitated efficient and accurate prediction of the simulated PAW under winter wheat, RMSD = 21 mm and normalized RMSD = 0.17, using observed PAW under grassland as an input. This new method for estimating soil moisture under adjacent, contrasting land covers at a relatively low computational cost could be employed for any region and land cover pairing with training data available, and it may significantly enhance the applications of existing large-scale soil moisture monitoring networks.

The Observation Record Length Necessary to Generate Robust Soil Moisture Percentiles

AbstractThe ability to use in situ soil moisture for large-scale soil moisture monitoring, model and satellite validation, and climate investigations is contingent on properly standardizing soil moisture observations. Percentiles are a useful method for homogenizing in situ soil moisture. However, very few stations have been continuously monitoring in situ soil moisture for 20 years or more. Therefore, one challenge in evaluating soil moisture is determining whether the period of record is sufficient to produce a stable distribution from which to generate percentiles. In this study daily in situ soil moisture observations, measured at three separate depths in the soil column at 15 stations in the United States and Canada, are used to determine the record length that is necessary to generate a stable soil moisture distribution. The Anderson–Darling test is implemented, both with and without a Bonferroni adjustment, to quantify the necessary record length. The authors evaluate how the necessary record length varies by location, measurement depth, and month. They find that between 3 and 15 years of data are required to produce stable distributions, with the majority of stations requiring only 3–6 years of data. Not surprisingly, more years of data are required to obtain stable estimates of the 5th and 95th percentiles than of the first, second, and third quartiles of the soil moisture distribution. Overall these results suggest that 6 years of continuous, daily in situ soil moisture data will be sufficient in most conditions to create stable and robust percentiles.

Feasibility analysis of using inverse modeling for estimating natural groundwater recharge from a large-scale soil moisture monitoring network

Summary Despite the importance of groundwater recharge (GR), its accurate estimation still remains one of the most challenging tasks in the field of hydrology. In this study, with the help of inverse modeling, long-term (6 years) soil moisture data at 34 sites from the Automated Weather Data Network (AWDN) were used to estimate the spatial distribution of GR across Nebraska, USA, where significant spatial variability exists in soil properties and precipitation (P). To ensure the generality of this study and its potential broad applications, data from public domains and literature were used to parameterize the standard Hydrus-1D model. Although observed soil moisture differed significantly across the AWDN sites mainly due to the variations in P and soil properties, the simulations were able to capture the dynamics of observed soil moisture under different climatic and soil conditions. The inferred mean annual GR from the calibrated models varied over three orders of magnitude across the study area. To assess the uncertainties of the approach, estimates of GR and actual evapotranspiration (ETa) from the calibrated models were compared to the GR and ETa obtained from other techniques in the study area (e.g., remote sensing, tracers, and regional water balance). Comparison clearly demonstrated the feasibility of inverse modeling and large-scale (>104 km2) soil moisture monitoring networks for estimating GR. In addition, the model results were used to further examine the impacts of climate and soil on GR. The data showed that both P and soil properties had significant impacts on GR in the study area with coarser soils generating higher GR; however, different relationships between GR and P emerged at the AWDN sites, defined by local climatic and soil conditions. In general, positive correlations existed between annual GR and P for the sites with coarser-textured soils or under wetter climatic conditions. With the rapidly expanding soil moisture monitoring networks around the globe, this study may have important applications in aiding water resources management in different regions.

The Capacitive Soil Moisture Sensor Research

To meet the requirements of a large-scale soil moisture monitoring system, this dissertation developed a new type of capacitive soil moisture sensor based on the detection and amplification of the singal peak-to-peak value named PA-1 sensor. PA-1 sensor was calibrated and evaluated under static laboratory conditions. A new type of soil moisture sensor named PA-1 is developed, which is based on the relationship between soil moisture and soil relative permittivity. Circuit design, operating features and exterior design of the sensor are introduced. PA-1 is calibrated in solution samples using the standard calibration method. Calibration experiment results show that PA-1 has a very linear measurement curve. The performance of PA-1 is studied and compared to the widely used Decagon Company’s EC-5 capacitive soil moisture sensor. Experiment results show that PA-1 has advantages of high sensitivity, low temperature drift, strong tolerance to electromagnetic interference and low power consumption. Keywords: sensor, Precision Agriculture, low temperature drift, low power consumption.

New Soil Property Database Improves Oklahoma Mesonet Soil Moisture Estimates*

Abstract Soil moisture data from the Oklahoma Mesonet are widely used in research efforts spanning many disciplines within Earth sciences. These soil moisture estimates are derived by translating measurements of matric potential into volumetric water content through site- and depth-specific water retention curves. The objective of this research was to increase the accuracy of the Oklahoma Mesonet soil moisture data through improved estimates of the water retention curve parameters. A comprehensive field sampling and laboratory measurement effort was conducted that resulted in new measurements of the percent of sand, silt, and clay; bulk density; and volumetric water content at −33 and −1500 kPa. These inputs were provided to the Rosetta pedotransfer function, and parameters for the water retention curve and hydraulic conductivity functions were obtained. The resulting soil property database, MesoSoil, includes 13 soil physical properties for 545 individual soil layers across 117 Oklahoma Mesonet sites. The root-mean-square difference (RMSD) between the resulting soil moisture estimates and those obtained by direct sampling was reduced from 0.078 to 0.053 cm3 cm−3 by use of the new water retention curve parameters, a 32% improvement. A &amp;gt;0.15 cm3 cm−3 high bias on the dry end was also largely eliminated by using the new parameters. Reanalysis of prior studies that used Oklahoma Mesonet soil moisture data may be warranted given these improvements. No other large-scale soil moisture monitoring network has a comparable published soil property database or has undergone such comprehensive in situ validation.

State of the Art in Large-Scale Soil Moisture Monitoring

Soil moisture is an essential climate variable influencing land-atmosphere interactions, an essential hydrologic variable impacting rainfall-runoff processes, an essential ecological variable regulating net ecosystem exchange, and an essential agricultural variable constraining food security. Large-scale soil moisture monitoring has advanced in recent years, creating opportunities to transform scientific understanding of soil moisture and related processes. These advances are being driven by researchers from a broad range of disciplines, but this complicates collaboration and communication; and, for some applications, the science required to utilize large-scale soil moisture data is poorly developed. In this review, we describe the state of the art in large-scale soil moisture monitoring and identify some critical needs for research to optimize the use of increasingly available soil moisture data. We review representative examples of (i) emerging in situ and proximal sensing techniques, (ii) dedicated soil moisture remote sensing missions, (iii) soil moisture monitoring networks, and (iv) applications of large-scale soil moisture measurements. Significant near-term progress seems possible in the use of large-scale soil moisture data for drought monitoring. Assimilation of soil moisture data for meteorological or hydrologic forecasting also shows promise, but significant challenges related to spatial variability and model structures remain. Little progress has been made in the use of large-scale soil moisture observations within the context of ecological or agricultural modeling. Opportunities abound to advance the science and practice of large-scale soil moisture monitoring for the sake of improved Earth system monitoring, modeling, and forecasting.

Potential of Synthetic Aperture Radar for Large-Scale Soil Moisture Monitoring: A Review

RESUMELe radar imageur a synthese d'ouverture (SAR) est un outil de teledetection qui peut servir a mesurer l'humidite du sol. La cartographie des teneurs en eau de surface a partir des niveaux de gris d'une image est une application possible bien que plusieurs facteurs peuvent interferer lors de la conversion. Ils doivent donc etre identifies et pris en compte. Les plus importants facteurs sont la rugosite de surface, la couverture vegetale et la configuration du radar. Une fois ces facteurs quantifies soit par des modeles, des algorithmes ou un systeme expert, l'information peut servir a plusieurs applications comme le suivi de l'evolution des teneurs en eau, l'initialisation et la validation de modeles d'humidite des sols, la distribution spatiale des precipitations, l'estimation des reserves en eau au printemps et a l'automne, la detection de la variabilite hydrique des champs et le rehaussement des problemes de drainage. Le potentiel du SAR et de RADARSAT pour le suivi de l'humidite des sols est eval...

The influence of polarization on canopy transmission properties at 6.6 GHz and implications for large scale soil moisture monitoring in semi-arid environments

Nimbus/SMMR 6.6 GHz data were used together with an extensive data base of surface moisture to study the influence of polarization on radiative transfer characteristics of savannah vegetation in south-eastern Botswana. Ratios of virtual and horizontal polarization optical depths and ratios of albedos were found to be almost constant throughout a three-year period. This conclusion led to a new approach to eliminate the vegetation components in the remotely sensed brightness temperatures, which is important for soil moisture monitoring from space. Using the same data set, this dual-polarization approach was compared with an earlier described synergistic approach, in which NDVI was used to eliminate the vegetation effects. The root mean square error between ground based soil moisture and satellite estimated surface soil moisture was reduced to 1.2% by volume from 5% by volume using the dual polarization approach. >

Microwave vegetation optical depth and inverse modelling of soil emissivity using Nimbus/SMMR satellite observations

A radiative transfer model has been used to determine the large scale effective 6.6 GHz and 37 GHz optical depths of the vegetation cover. Knowledge of the vegetation optical depth is important for satellite-based large scale soil moisture monitoring using microwave radiometry. The study is based on actual observed large scale surface soil moisture data and observed dual polarization 6.6 and 37 GHz Nimbus/SMMR brightness temperatures over a 3-year period. The derived optical depths have been compared with microwave polarization differences and polarization ratios in both frequencies and with Normalized Difference Vegetation Index (NDVI) values from NOAA/AVHRR. A synergistic approach to derive surface soil emissivity from satellite observed brightness temperatures by inverse modelling is described. This approach improves the relationship between satellite derived surface emissivity and large scale top soil moisture fromR 2=0.45 (no correction for vegetation) toR 2=0.72 (after correction for vegetation). This study also confirms the relationship between the microwave-based MPDI and NDVI earlier described and explained in the literature.