Exploring the use of machine-learning techniques to integrate ground- and remote sensing-based observations for efficient near-surface soil moisture mapping

2020 ◽  
Author(s):  
Sarah Schönbrodt-Stitt ◽  
Paolo Nasta ◽  
Nima Ahmadian ◽  
Markus Kurtenbach ◽  
Christopher Conrad ◽  
...  
Keyword(s):  
Machine Learning ◽  
Soil Moisture ◽  
Spatial Resolution ◽  
Surface Soil ◽  
Bare Soil ◽  
Small Scale ◽  
Surface Soil Moisture ◽  
Near Surface ◽  
Campania Region

<p>Mapping near-surface soil moisture (<em>θ</em>) is of tremendous relevance for a broad range of environment-related disciplines and meteorological, ecological, hydrological and agricultural applications. Globally available products offer the opportunity to address <em>θ</em> in large-scale modelling with coarse spatial resolution such as at the landscape level. However, <em>θ</em> estimation at higher spatial resolution is of vital importance for many small-scale applications. Therefore, we focus our study on a small-scale catchment (MFC2) belonging to the “Alento” hydrological observatory, located in southern Italy (Campania Region). The goal of this study is to develop new machine-learning approaches to estimate high grid-resolution (about 17 m cell size) <em>θ</em> maps from mainly backscatter measurements retrieved from C-band Synthetic Aperture Radar (SAR) based on Sentinel-1 (S1) images and from gridded terrain attributes. Thus, a workflow comprising a total of 48 SAR-based <em>θ</em> patterns estimated for 24 satellite overpass dates (revisit time of 6 days) each with ascendant and descendent orbits will be presented. To enable for the mapping, SAR-based <em>θ</em> data was calibrated with in-situ measurements carried out with a portable device during eight measurement campaigns at time of satellite overpasses (four overpass days in total with each ascendant and descendent satellite overpasses per day in November 2018). After the calibration procedure, data validation was executed from November 10, 2018 till March 28, 2019 by using two stationary sensors monitoring <em>θ</em> at high-temporal (1-min recording time). The specific sensor locations reflected two contrasting field conditions, one bare soil plot (frequently kept clear, without disturbance of vegetation cover) and one non-bare soil plot (real-world condition). Point-scale ground observations of <em>θ</em> were compared to pixel-scale (17 m × 17 m), SAR-based <em>θ</em> estimated for those pixels corresponding to the specific positions of the stationary sensors. Mapping performance was estimated through the root mean squared error (RMSE). For a short-term time series of <em>θ</em> (Nov 2018) integrating 136 in situ, sensor-based <em>θ</em> (<em>θ</em><sub>insitu</sub>) and 74 gravimetric-based <em>θ</em> (<em>θ</em><sub>gravimetric</sub>) measurements during a total of eight S1 overpasses, mapping performance already proved to be satisfactory with RMSE=0.039 m³m<sup>-</sup>³ and R²=0.92, respectively with RMSE=0.041 m³m<sup>-</sup>³ and R²=0.91. First results further reveal that estimated satellite-based <em>θ</em> patterns respond to the evolution of rainfall. With our workflow developed and results, we intend to contribute to improved environmental risk assessment by assimilating the results into hydrological models (e.g., HydroGeoSphere), and to support future studies on combined ground-based and SAR-based <em>θ</em> retrieval for forested land (future missions operating at larger wavelengths e.g. NISARL-band, Biomass P-band sensors).</p>

scholarly journals The impact of near-surface soil moisture assimilation at subseasonal, seasonal, and inter-annual timescales

2015 ◽  
Vol 19 (12) ◽  
pp. 4831-4844 ◽  
Author(s):  
C. Draper ◽  
R. Reichle
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Surface Soil ◽  
Root Zone ◽  
Surface Model ◽  
Surface Soil Moisture ◽  
Near Surface ◽  
In Situ Observations ◽  
The Impact

Abstract. A 9 year record of Advanced Microwave Scanning Radiometer – Earth Observing System (AMSR-E) soil moisture retrievals are assimilated into the Catchment land surface model at four locations in the US. The assimilation is evaluated using the unbiased mean square error (ubMSE) relative to watershed-scale in situ observations, with the ubMSE separated into contributions from the subseasonal (SMshort), mean seasonal (SMseas), and inter-annual (SMlong) soil moisture dynamics. For near-surface soil moisture, the average ubMSE for Catchment without assimilation was (1.8 × 10−3 m3 m−3)2, of which 19 % was in SMlong, 26 % in SMseas, and 55 % in SMshort. The AMSR-E assimilation significantly reduced the total ubMSE at every site, with an average reduction of 33 %. Of this ubMSE reduction, 37 % occurred in SMlong, 24 % in SMseas, and 38 % in SMshort. For root-zone soil moisture, in situ observations were available at one site only, and the near-surface and root-zone results were very similar at this site. These results suggest that, in addition to the well-reported improvements in SMshort, assimilating a sufficiently long soil moisture data record can also improve the model representation of important long-term events, such as droughts. The improved agreement between the modeled and in situ SMseas is harder to interpret, given that mean seasonal cycle errors are systematic, and systematic errors are not typically targeted by (bias-blind) data assimilation. Finally, the use of 1-year subsets of the AMSR-E and Catchment soil moisture for estimating the observation-bias correction (rescaling) parameters is investigated. It is concluded that when only 1 year of data are available, the associated uncertainty in the rescaling parameters should not greatly reduce the average benefit gained from data assimilation, although locally and in extreme years there is a risk of increased errors.

scholarly journals Statistical Exploration of SENTINEL-1 Data, Terrain Parameters, and in-situ Data for Estimating the Near-Surface Soil Moisture in a Mediterranean Agroecosystem

2021 ◽  
Vol 3 ◽  
Author(s):  
Sarah Schönbrodt-Stitt ◽  
Nima Ahmadian ◽  
Markus Kurtenbach ◽  
Christopher Conrad ◽  
Nunzio Romano ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Empirical Model ◽  
Surface Soil ◽  
Surface Soil Moisture ◽  
Channel Network ◽  
Near Surface ◽  
Short Period ◽  
Terrain Parameters ◽  
Simple Empirical Model

Reliable near-surface soil moisture (θ) information is crucial for supporting risk assessment of future water usage, particularly considering the vulnerability of agroforestry systems of Mediterranean environments to climate change. We propose a simple empirical model by integrating dual-polarimetric Sentinel-1 (S1) Synthetic Aperture Radar (SAR) C-band single-look complex data and topographic information together with in-situ measurements of θ into a random forest (RF) regression approach (10-fold cross-validation). Firstly, we compare two RF models' estimation performances using either 43 SAR parameters (θNovSAR) or the combination of 43 SAR and 10 terrain parameters (θNovSAR+Terrain). Secondly, we analyze the essential parameters in estimating and mapping θ for S1 overpasses twice a day (at 5 a.m. and 5 p.m.) in a high spatiotemporal (17 × 17 m; 6 days) resolution. The developed site-specific calibration-dependent model was tested for a short period in November 2018 in a field-scale agroforestry environment belonging to the “Alento” hydrological observatory in southern Italy. Our results show that the combined SAR + terrain model slightly outperforms the SAR-based model (θNovSAR+Terrain with 0.025 and 0.020 m3 m−3, and 89% compared to θNovSAR with 0.028 and 0.022 m3 m−3, and 86% in terms of RMSE, MAE, and R2). The higher explanatory power for θNovSAR+Terrain is assessed with time-variant SAR phase information-dependent elements of the C2 covariance and Kennaugh matrix (i.e., K1, K6, and K1S) and with local (e.g., altitude above channel network) and compound topographic attributes (e.g., wetness index). Our proposed methodological approach constitutes a simple empirical model aiming at estimating θ for rapid surveys with high accuracy. It emphasizes potentials for further improvement (e.g., higher spatiotemporal coverage of ground-truthing) by identifying differences of SAR measurements between S1 overpasses in the morning and afternoon.

scholarly journals New Downscaling Approach Using ESA CCI SM Products for Obtaining High Resolution Surface Soil Moisture

2020 ◽  
Vol 12 (7) ◽  
pp. 1119 ◽  
Author(s):  
Jovan Kovačević ◽  
Željko Cvijetinović ◽  
Nikola Stančić ◽  
Nenad Brodić ◽  
Dragan Mihajlović
Keyword(s):  
Soil Moisture ◽  
High Resolution ◽  
Random Forest ◽  
Spatial Resolution ◽  
Surface Soil ◽  
Initial Step ◽  
Surface Soil Moisture ◽  
In Situ Data ◽  
Random Forest Models

ESA CCI SM products have provided remotely-sensed surface soil moisture (SSM) content with the best spatial and temporal coverage thus far, although its output spatial resolution of 25 km is too coarse for many regional and local applications. The downscaling methodology presented in this paper improves ESA CCI SM spatial resolution to 1 km using two-step approach. The first step is used as a data engineering tool and its output is used as an input for the Random forest model in the second step. In addition to improvements in terms of spatial resolution, the approach also considers the problem of data gaps. The filling of these gaps is the initial step of the procedure, which in the end produces a continuous product in both temporal and spatial domains. The methodology uses combined active and passive ESA CCI SM products in addition to in situ soil moisture observations and the set of auxiliary downscaling predictors. The research tested several variants of Random forest models to determine the best combination of ESA CCI SM products. The conclusion is that synergic use of all ESA CCI SM products together with the auxiliary datasets in the downscaling procedure provides better results than using just one type of ESA CCI SM product alone. The methodology was applied for obtaining SSM maps for the area of California, USA during 2016. The accuracy of tested models was validated using five-fold cross-validation against in situ data and the best variation of model achieved RMSE, R2 and MAE of 0.0518 m3/m3, 0.7312 and 0.0374 m3/m3, respectively. The methodology proved to be useful for generating high-resolution SSM products, although additional improvements are necessary.

An offline reanalysis of land surface variables with LDAS-Monde forced by a kilometric scale NWP system

2020 ◽  
Author(s):  
Bertrand Bonan ◽  
Clément Albergel ◽  
Adrien Napoly ◽  
Yongjun Zheng ◽  
Jean-Christophe Calvet
Keyword(s):  
Soil Moisture ◽  
Spatial Resolution ◽  
Land Surface ◽  
Surface Soil ◽  
Weather Prediction ◽  
Land Surface Model ◽  
Research Centre ◽  
Small Scale ◽  
Surface Model ◽  
Surface Soil Moisture

<p>LDAS-Monde is the offline land data assimilation system (LDAS) developed by Météo-France’s research centre (CNRM) aiming to monitor the evolution of land surface variables (LSVs) at various scales, from regional to global. It combines numerical simulations from the multilayer and interactive vegetation ISBA land surface model and satellite-derived observations of surface soil moisture and leaf area index (LAI). LDAS-Monde has been successfully validated over the globe.</p><p>In this work, we study the possibility to set up LDAS-Monde to the context of the kilometric spatial resolution. In this context, we assimilate satellite observations of LAI from the Copernicus Global Land Service (CGLS) into the ISBA land surface model forced with Météo-France’s small scale numerical weather prediction system AROME. We produce a reanalysis of LSVs at 2.5-km spatial resolution over the AROME domain centred on France starting from 2017. The quality of this reanalysis is assessed by comparing the obtained reanalysis with satellite products of LAI and surface soil moisture from e.g. CGLS and in-situ measurements of soil moisture from various networks (SMOSMANIA, …). We also show the ability of our system to monitor the evolution of LSVs in the context of the severe drought that France suffered during the summer 2018. LDAS-Monde at 2.5-km spatial resolution displays a great potential for agricultural monitoring at high resolution. We also plan to adapt our framework to 1.0-km spatial resolution.</p>

scholarly journals A Machine Learning-Based Approach for Surface Soil Moisture Estimations with Google Earth Engine

2021 ◽  
Vol 13 (11) ◽  
pp. 2099
Author(s):  
Felix Greifeneder ◽  
Claudia Notarnicola ◽  
Wolfgang Wagner
Keyword(s):  
Machine Learning ◽  
Soil Moisture ◽  
Moisture Content ◽  
Spatial Resolution ◽  
Soil Moisture Content ◽  
Surface Soil ◽  
High Spatial Resolution ◽  
Google Earth ◽  
Surface Soil Moisture ◽  
Google Earth Engine

Due to its relation to the Earth’s climate and weather and phenomena like drought, flooding, or landslides, knowledge of the soil moisture content is valuable to many scientific and professional users. Remote-sensing offers the unique possibility for continuous measurements of this variable. Especially for agriculture, there is a strong demand for high spatial resolution mapping. However, operationally available soil moisture products exist with medium to coarse spatial resolution only (≥1 km). This study introduces a machine learning (ML)—based approach for the high spatial resolution (50 m) mapping of soil moisture based on the integration of Landsat-8 optical and thermal images, Copernicus Sentinel-1 C-Band SAR images, and modelled data, executable in the Google Earth Engine. The novelty of this approach lies in applying an entirely data-driven ML concept for global estimation of the surface soil moisture content. Globally distributed in situ data from the International Soil Moisture Network acted as an input for model training. Based on the independent validation dataset, the resulting overall estimation accuracy, in terms of Root-Mean-Squared-Error and R², was 0.04 m3·m−3 and 0.81, respectively. Beyond the retrieval model itself, this article introduces a framework for collecting training data and a stand-alone Python package for soil moisture mapping. The Google Earth Engine Python API facilitates the execution of data collection and retrieval which is entirely cloud-based. For soil moisture retrieval, it eliminates the requirement to download or preprocess any input datasets.

Optimal Spatial Resolution of Sentinel-1 Surface Soil Moisture Evaluated Using Intensive in Situ Observations

2021 ◽  
Author(s):  
Theresa C. Van Hateren ◽  
Marco Chini ◽  
Patrick Matgen ◽  
Luca Pulvirenti ◽  
Nazzareno Pierdicca ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Spatial Resolution ◽  
Surface Soil ◽  
Surface Soil Moisture ◽  
In Situ Observations

scholarly journals Estimating root zone soil moisture using near-surface observations from SMOS

2014 ◽  
Vol 18 (1) ◽  
pp. 139-154 ◽  
Author(s):  
T. W. Ford ◽  
E. Harris ◽  
S. M. Quiring
Keyword(s):  
United States ◽  
Soil Moisture ◽  
Great Plains ◽  
Surface Soil ◽  
Root Zone ◽  
The United States ◽  
Surface Soil Moisture ◽  
Near Surface ◽  
Cross Correlation Analysis

Abstract. Satellite-derived soil moisture provides more spatially and temporally extensive data than in situ observations. However, satellites can only measure water in the top few centimeters of the soil. Root zone soil moisture is more important, particularly in vegetated regions. Therefore estimates of root zone soil moisture must be inferred from near-surface soil moisture retrievals. The accuracy of this inference is contingent on the relationship between soil moisture in the near-surface and the soil moisture at greater depths. This study uses cross correlation analysis to quantify the association between near-surface and root zone soil moisture using in situ data from the United States Great Plains. Our analysis demonstrates that there is generally a strong relationship between near-surface (5–10 cm) and root zone (25–60 cm) soil moisture. An exponential decay filter is used to estimate root zone soil moisture using near-surface soil moisture derived from the Soil Moisture and Ocean Salinity (SMOS) satellite. Root zone soil moisture derived from SMOS surface retrievals is compared to in situ soil moisture observations in the United States Great Plains. The SMOS-based root zone soil moisture had a mean R2 of 0.57 and a mean Nash–Sutcliffe score of 0.61 based on 33 stations in Oklahoma. In Nebraska, the SMOS-based root zone soil moisture had a mean R2 of 0.24 and a mean Nash–Sutcliffe score of 0.22 based on 22 stations. Although the performance of the exponential filter method varies over space and time, we conclude that it is a useful approach for estimating root zone soil moisture from SMOS surface retrievals.

scholarly journals Different Rates of Soil Drying after Rainfall Are Observed by the SMOS Satellite and the South Fork in situ Soil Moisture Network

2015 ◽  
Vol 16 (2) ◽  
pp. 889-903 ◽  
Author(s):  
Wesley J. Rondinelli ◽  
Brian K. Hornbuckle ◽  
Jason C. Patton ◽  
Michael H. Cosh ◽  
Victoria A. Walker ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Surface Soil ◽  
Soil Drying ◽  
Surface Soil Moisture ◽  
South Fork ◽  
The South ◽  
Near Surface ◽  
Soil Moisture Sensors ◽  
Smos Satellite

Abstract Soil moisture affects the spatial variation of land–atmosphere interactions through its influence on the balance of latent and sensible heat fluxes. Wetter soils are more prone to flooding because a smaller fraction of rainfall can infiltrate into the soil. The Soil Moisture Ocean Salinity (SMOS) satellite carries a remote sensing instrument able to make estimates of near-surface soil moisture on a global scale. One way to validate satellite observations is by comparing them with observations made with sparse networks of in situ soil moisture sensors that match the extent of satellite footprints. The rate of soil drying after significant rainfall observed by SMOS is found to be higher than the rate observed by a U.S. Department of Agriculture (USDA) soil moisture network in the watershed of the South Fork Iowa River. This leads to the conclusion that SMOS and the network observe different layers of the soil: SMOS observes a layer of soil at the soil surface that is a few centimeters thick, while the network observes a deeper soil layer centered at the depth at which the in situ soil moisture sensors are buried. It is also found that SMOS near-surface soil moisture is drier than the South Fork network soil moisture, on average. The conclusion that SMOS and the network observe different layers of the soil, and therefore different soil moisture dynamics, cannot explain the dry bias. However, it can account for some of the root-mean-square error in the relationship. In addition, SMOS observations are noisier than the network observations.

Sign in / Sign up

Export Citation Format

Share Document