Several targets of Sustainable Development Goals depend on the soil condition, as they impact ecosystems functioning and food, fibre and timber production. Soil condition regulates the climate, hydrological and nutrient cycle and provide resilience against floods and droughts. Addressing these global issues also requires reliable tools for global soil monitoring, such as Earth Observation (EO) products for mapping and monitoring soils, including their uncertainty. This contemporary research would greatly benefit from synergies between pedometrics and spectroscopy/remote sensing scientists. There is the need to create models linking soil properties with ancillary environmental variables, such as proximal and remote sensing data and that help quantifying uncertainty. These models will allow to understand the processes happening in the soil and in the landscape with space-time patterns, at different scales. In this session, we aim to bring together scientists working on research related to using the full range of pedometrics and soil sensing techniques available for mapping and monitoring soils. A preliminary view indicates some pillars as follows: a) the link between spectroscopy and wet soil laboratory analysis, seeking for the best strategy to evolve soil quality analysis; b) the link between proximal and remote sensing, with soil analysis; c) the link between proximal/remote sensing and pedometrics for extrapolating relationships established at point support to the spatial and temporal extent covered by proximal/remote sensing. We will aim to identify priorities for the future in what is an active area of collaborative research.
vPICO presentations: Thu, 29 Apr
The concentration of micronutrients in staple crops varies spatially. Quantitative information about this can help in designing efficient interventions to address micronutrient deficiency. The concentration of a micronutrient in a staple crop can be mapped from limited samples, but the resulting statistical predictions are uncertain. Decision-makers must understand this uncertainty to make robust use of spatial information, but this is a challenge due to the difficulties of communicating quantitative concepts to a general audience. We proposed strategies to communicate uncertain information and present a systematic evaluation and comparison in the form of maps. We proposed to test five methods to communicate the uncertainty about the conditional mean grain concentration of an essential micronutrient, selenium (Se). Evaluation of the communication methods was done through questionnaire by eliciting stakeholder opinions about the usefulness of the methods of communicating uncertainty. We found significant differences in how participants responded to the different methods. In particular, there was a preference for methods based on the probability that concentrations are below or above a nutritionally-significant threshold compared with general measures of uncertainty such as the confidence interval of a prediction. There was no evidence that methods which used pictographs or calibrated verbal phrases to support the interpretation of probabilities made a different impression than probability alone, as judged from the responses to interpretative questions, although these approaches were ranked most highly when participants were asked to put the methods in order of preference.
How to cite: Chagumaira, C., Chimungu, J. G., Gashu, D., Nalivata, P. C., Broadley, M. R., Milne, A. E., and Lark, R. M.: Communicating uncertainties in spatial predictions of grain micronutrient concentration, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-709, https://doi.org/10.5194/egusphere-egu21-709, 2021.
Soils are crossroads of carbon and nitrogen geochemical cycles and were consequently identified as a potential sink for carbon (C) and a compartment storage for nitrogen (N). Monitoring the joint evolution over time of organic C and total N stocks in soils appears interesting because of the C/N ratio is an indicator of changes in the organic matter quality. Nevertheless, the temporal evolutions detected in most of the existing studies are in the order of a few gC.m-2.yr-1 (C) and mgN.m-2.yr-1 (N). This study aims to assess uncertainties of soil organic carbon (SSOC) and soil total nitrogen (SSTN) stocks in the topsoil layer (0-25 cm) using three different methods (stochastic, deterministic and experimental), in order to identify the main sources of uncertainty and to evaluate the significance of SSOC and SSTN evolutions over the time. This study was based on a 1200 ha agricultural catchment area in Brittany (France) where systematic soil sampling was repeated at 108 sites in 2013 and 2018. Moreover, soil sampling was repeated three times in 2020 at the same sites by 3 different teams of experienced samplers. Comparing the three methods of uncertainty assessment, we found they provided equivalent results with a SSOC standard deviation of 0.85, 0.74 and 0.68 kgC.m-2 respectively for stochastic, deterministic and experimental approaches and 0.08, 0.07 and 0.06 kgN.m-2 for SSTN. Variance decomposition identified variations of fine earth mass as the main source of uncertainty (77 % of total variance) and attributed at least 16% of the uncertainties due to the operator procedure and were therefore reducible. Using the stochastic approach, the width of the 90 % confidence interval was estimated at each sampling site for C, N and C/N temporal changes. Changes were considered significant at respectively 59, 77 et 99 sites for SSOC, SSTN and C/N: a majority of sites lost organic carbon (-0.03 ± 0.07 kgC.m-2.yr-1), gained total nitrogen (0.006 ± 0.005 kgN.m-2.yr-1) and the C.N-1 (-0.17 ± 0.09 yr-1) ratio decreased. Finally, stock measurements uncertainty was mainly explained by soil natural variability but may still be reduced by a better control of the measurement procedure. In the agricultural context of the study area, the accuracy of the direct measurement appeared sufficient to detect SSOC and SSTN evolution over a time span of 5 years.
How to cite: Tabaud, L., Walter, C., Blancfene, C., Gascuel, C., Lemercier, B., Michot, D., and Pichelin, P.: Assessment of nitrogen and organic carbon stocks in agricultural soils: uncertainties and significance of temporal evolution, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10588, https://doi.org/10.5194/egusphere-egu21-10588, 2021.
Obtaining detailed soil information at a scale suitable for variable rate application (VRA) of fertilizer requires intense sampling. Within this context, traditional wet chemistry analysis of the soil is too costly to provide sufficient detail on the spatial variation of soils within fields. Infrared and X-ray fluorescence spectroscopy of soil is much cheaper and might provide sufficient data to map the concentrations affordably. Estimates of the concentrations from spectroscopy are subject to error, however, and these errors in turn carry with them costs if they lead to under- or over-applications of fertilizer and resultant loss of potential crop yields or environmental pollution. Given a loss function and an error distribution for the estimates of the concentrations, it is possible to minimize the expected loss from VRA fertilizer management.
Within this study, we have estimated the variation of available P and K within horticultural fields and the effect of errors on the expected loss. The topsoil (0-25 cm) of four fields was sampled at numerous points and analysed by infrared and X-ray spectroscopy to provide estimates of available P and K. . The spatial variation in the estimates were modelled as mixtures of fixed and random effects by residual maximum likelihood (REML), with the spectroscopic model error accounted for in the parameter estimates. These models where then used to map the P and K content by universal kriging together with their kriging variances. Loss functions were computed based on the error distribution and a dose response curve from the literature. We then computed the total loss (compared to perfect information) for each field under VRA of fertilizer and a blanket fertilizer scheme based on the wet chemistry data.
Underestimation of the error variance in four out of eight linear mixed models underlines the importance of error propagation to estimate the short-scale spatial variance of the soil property correctly. As expected, the optimum fertilizer rate when accounting for uncertainty tends towards over-application. The asymmetry of the loss function described that underestimation generally leads to a higher loss compared to overestimation of soil P and K. The effect of error variance on the expected loss was further found to be dependent on the range of the kriging predictions relative to the parameterization of the dose response curve. There was a financial incentive for VRA of P fertilizer but not for K. Additionally there is an environmental incentive for VRA of P because much less fertilizer would be applied compared to a blanket fertilizer rate based on wet chemistry data.
How to cite: Breure, T., Haefele, S. M., Webster, R., Hannam, J. A., Corstanje, R., and Milne, A. E.: Agricultural decision-making under uncertainty: a loss function on the kriging variance from soil properties predicted by infrared and X-ray fluorescence spectroscopy, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12615, https://doi.org/10.5194/egusphere-egu21-12615, 2021.
There is a growing demand for high quality soil data to model soil processes and map soil properties. However, wet chemistry measurements on soil properties are subjected to many error sources, such as the observer, the instrument and lack of standardised methodologies. Consequently, soil data are imperfect and uncertain because of these error sources. Uncertainties in measurements of fundamental soil properties can propagate through, e.g., pedotransfer functions, spectroscopic models and digital soil mapping algorithms. Therefore, it is important to provide detailed uncertainty information about soil measurements to potential data users. In practice, uncertainty estimates are rarely specified by providers of analytical soil data.
In this research, we aimed to quantify uncertainties in synthetic and real-world pH (1:1 soil-water suspension) and Total Organic Carbon (TOC) measurements. We assumed that uncertainty can be represented by a normal distribution. A linear mixed-effects model was applied to estimate the parameters of the normal distribution, i.e., mean and standard deviation, of both synthetic and real-world datasets. The model included ‘sample ID’ as a fixed effect, and ‘batch’ and ‘laboratory’ as random effects. The use of synthetic datasets allowed us to investigate how well the model parameters could be estimated given a specific experimental measurement design, whereas the real-world case served to explore if the parameter estimates were still accurate for such unbalanced datasets.
For a balanced dataset (n=20, n=100, n=200 and n=500), using synthetic pH data for three hypothetical laboratories (two batches per laboratory), the mean estimated standard deviations (σ) of the random effects were σbatch=0.10, σlaboratory=0.24 and σresidual=0.2. These estimates were in agreement with the σ for the respective random effects used to generate the synthetic dataset, meaning that the model could accurately estimate the model parameters. Subsequently, changes were made to the experimental measurement design by randomly removing 20%, 50% and 80% of the data, resulting in unbalanced datasets. In general, the interquartile range (IQR) of σ for each random effect increased with a larger percentage of removed data. However, the increase in IQR was larger for n=20 compared to, e.g., n=200. When comparing 0% and 80% randomly removed data, the IQR for the batch effect increased with 60.3%. Conversely, for n=200 an increase of only 23.5% was observed.
Subsequently, the same model was fitted on real-world pH and TOC data, provided by the Wageningen Evaluating Programs for Analytical Laboratories (WEPAL). The unbalanced dataset structure was first reconstructed and filled with synthetically generated data, based on sample means and standard deviations derived from the measured data. The model was fitted on both datasets. For measured pH, the model yielded σbatch=0.27, σlaboratory=0.17 and σresidual=0.10. The IQRs of the estimated σ from synthetic WEPAL data were 0.04 (batch), 0.06 (laboratory) and 0.02 (residual). The model fitted on the measured TOC data estimated σbatch=5.3%, σlaboratory=2.8% and σresidual=2.1%. For the synthetic WEPAL data, IQRs of 1.3% (batch), 1.4% (laboratory) and 0.4% (residual) were determined for the estimated σ. These findings suggest that despite having a highly unbalanced dataset, realistic model parameter estimates can still be obtained.
How to cite: van Leeuwen, C., Mulder, T., Batjes, N., and Heuvelink, G.: Statistical modelling of measurement error in wet chemistry soil data, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-1272, https://doi.org/10.5194/egusphere-egu21-1272, 2021.
The use of remote sensing data can lead to great efficiencies when mapping soil variables across broad regions. However, remote sensors rarely make direct measurements of the soil property of interest. Instead, an empirical model is required to relate the remote sensing data to ground measurements of the property of interest. We discuss how a survey of ground measurements required to calibrate such a model can be optimized. We make reference to the mapping of peat depth within the Dartmoor National Park (UK) using radiometric potassium data from an airborne survey of the region (http://www.tellusgb.ac.uk/). We expand the standard linear mixed model to accommodate nonlinear relationships between radiometric potassium and peat depths. The attenuation of the radiometric signal is seen to increase with peat depth, but the depth is particularly uncertain when the radiometric signal is small. When a spatial simulated annealing algorithm is used to optimize the locations for a survey of peat depth measurements to minimize the errors in the maps of peat depth upon use of the radiometric data, the complete range of the radiometric data are sampled but ground measurements are particularly focussed where the radiometric signal is small. We see that an optimized survey of 30 ground measurements combined with the radiometric data lead to more accurate maps than can be achieved from interpolation of more than 200 peat depth measurements.
How to cite: Marchant, B.: Prediction of peat depths using airborne radiometric data: optimization of ground surveys., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5827, https://doi.org/10.5194/egusphere-egu21-5827, 2021.
When planning a geochemical survey, it is necessary to make decisions about the sampling density. Sampling density determines both the quality of predictions and the cost of field work. In geostatistical surveys, the relationship between sampling density and map quality, as measured by the kriging variance (mean square error of the prediction) can be computed. When the variogram is known, then the kriging variance at an unsampled site depends only on the spatial distribution of sampling points around that site. It is therefore possible to find the sample density such that the kriging variance is limited to acceptable values. However, the implications of kriging variances are not always straightforward for decision makers or sponsors of survey to understand. Here we present an alternative method to help end-users assess the implications of uncertainty in spatial prediction in so far as this is controlled by sampling. It is called the offset correlation and is a measure of how far the mapped spatial variation depends on the positioning of a reqular square sampling grid. The offset correlation increases as the uncertainty in the map, attributable to sample density, decreases. It is bounded on the interval [0,1], which makes it intuitively easy to interpret as an uncertainty measure. In this presentation we shall explain the offset correlation concept, illustrate it with some test cases, and provide session participants with an opportunity to join an elicitation of sampling density for a hypothetical survey of soil micronutrient status.
The offset correlation is an intuitive measure of the precision of a geostatistical mapping process because people can more easily grasp bounded measures like a correlation than unbounded ones like a variance.
How to cite: Milne, A., Chagumaira, C., and Lark, M.: Planning a geostatistical survey for mapping soil micronutrients: eliciting sampling densities, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5611, https://doi.org/10.5194/egusphere-egu21-5611, 2021.
The relationship between visible-near-infrared (Vis-NIR-SWIR) spectra and soil organic carbon (SOC) and the effects of preprocessing techniques on SOC predictive models have been shown in several studies. However, little attention has been given to the effect of analytical methods used to produce the SOC data used to calibrate those models. The predictive performance of Vis-NIR spectral models depends not only on the preprocessing technique and machine learning method but also on the analytical method employed to produce the SOC data. Our hypothesis is that some combinations of preprocessing and models may be more sensitive to laboratory (measurement) error than others. To test this hypothesis, we evaluated the leave-one-out cross-validation performance of three predictive models (Random Forest (RF), Cubist, and Partial Least Square Regression (PLSR)) calibrated using SOC data produced via three analytical methods (dry combustion (DC) and wet combustion with quantification by titration (WCt) and colorimetry (WCc)) and three Vis-NIR spectra preprocessing techniques (smoothing (SMO), continuum removal (CRR), and Savitzky-Golay first derivative (SGD)). The prediction performance varied among the models. DC and WCt provided a higher correlation between SOC and spectra than WCc, and thus, resulted in higher accuracy. The Cubist+CRR was ranked the best performing model, with an average of R2 = 0.81 and RMSE = 0.81% among analytical methods. Cubist+CRR also minimized the accuracy differences resulting from SOC analytical methods employed. The RF model had low accuracy and was unable to explain more than 46% of the variance. Overall, the analytical method significantly affects SOC predictions, and its impact may be larger than the preprocessing.
How to cite: Zborowski Horst-Heinen, T., Diniz Dalmolin, R. S., Samuel-Rosa, A., and Grunwald, S.: The interplay among analytical method, preprocessing, and modeling on soil organic carbon Vis-NIR-SWIR predictions, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7851, https://doi.org/10.5194/egusphere-egu21-7851, 2021.
Spectral data obtained from optical spaceborne sensors are being recognized as a valuable source of data that show promising results in assessing soil properties on medium and macro scale. Combining this technique with laboratory Visible-Near Infrared (VIS-NIR) spectroscopy methods can be an effective approach to perform robust research on plot scale to determine wildfire impact on soil organic matter (SOM) immediately after the fire. Therefore, the objective of this study was to assess the ability of Sentinel-2 superspectral data in estimating post-fire SOM content and comparison with the results acquired with laboratory VIS-NIR spectroscopy.
The study is performed in Mediterranean Croatia (44° 05’ N; 15° 22’ E; 72 m a.s.l.), on approximately 15 ha of fire affected mixed Quercus ssp. and Juniperus ssp. forest on Cambisols. A total of 80 soil samples (0-5 cm depth) were collected and geolocated on August 22nd 2019, two days after a medium to high severity wildfire. The samples were taken to the laboratory where soil organic carbon (SOC) content was determined via dry combustion method with a CHNS analyzer. SOM was subsequently calculated by using a conversion factor of 1.724. Laboratory soil spectral measurements were carried out using a portable spectroradiometer (350-1050 nm) on all collected soil samples. Two Sentinel-2 images were downloaded from ESAs Scientific Open Access Hub according to the closest dates of field sampling, namely August 31st and September 5th 2019, each containing eight VIS-NIR and two SWIR (Short-Wave Infrared) bands which were extracted from bare soil pixels using SNAP software. Partial least squares regression (PLSR) model based on the pre-processed spectral data was used for SOM estimation on both datasets. Spectral reflectance data were used as predictors and SOM content was used as a response variable. The accuracy of the models was determined via Root Mean Squared Error of Prediction (RMSEp) and Ratio of Performance to Deviation (RPD) after full cross-validation of the calibration datasets.
The average post-fire SOM content was 9.63%, ranging from 5.46% minimum to 23.89% maximum. Models obtained from both datasets showed low RMSEp (Spectroscopy dataset RMSEp = 1.91; Sentinel-2 dataset RMSEp = 0.99). RPD values indicated very good predictions for both datasets (Spectrospcopy dataset RPD = 2.72; Sentinel-2 dataset RPD = 2.22). Laboratory spectroscopy method with higher spectral resolution provided more accurate results. Nonetheless, spaceborne method also showed promising results in the analysis and monitoring of SOM in post-burn period.
Keywords: remote sensing, soil spectroscopy, wildfires, soil organic matter
Acknowledgment: This work was supported by the Croatian Science Foundation through the project "Soil erosion and degradation in Croatia" (UIP-2017-05-7834) (SEDCRO). Aleksandra Perčin is acknowledged for her cooperation during the laboratory work.
How to cite: Hrelja, I., Šestak, I., and Bogunović, I.: Estimation of soil organic matter using proximal and satellite sensors after a wildfire in Mediterranean Croatia, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-3896, https://doi.org/10.5194/egusphere-egu21-3896, 2021.
The Joint IAEA/FAO Division of Nuclear Techniques in Food and Agriculture, through its Soil and Water Management & Crop Nutrition Laboratory (SWMCNL), launched in October 2019, a new Coordinated Research Project (D15019) called “Monitoring and Predicting Radionuclide Uptake and Dynamics for Optimizing Remediation of Radioactive Contamination in Agriculture''. Within this context, the high-throughput characterization of soil properties in general and the estimation of soil-to-plant transfer factors of radionuclides are of critical importance.
For several decades, soil researchers have been successfully using near and mid-infrared spectroscopy (MIRS) techniques to estimate a wide range of soil physical, chemical and biological properties such as carbon (C), Cation Exchange Capacities (CEC), among others. However, models developed were often limited in scope as only small and region-specific MIR spectra libraries of soils were accessible.
This situation of data scarcity is changing radically today with the availability of large and growing library of MIR-scanned soil samples maintained by the National Soil Survey Center (NSSC) Kellogg Soil Survey Laboratory (KSSL) from the United States Department of Agriculture (USDA-NRCS) and the Global Soil Laboratory Network (GLOSOLAN) initiative of the Food Agency Organization (FAO). As a result, the unprecedented volume of data now available allows soil science researchers to increasingly shift their focus from traditional modeling techniques such as PLSR (Partial Least Squares Regression) to classes of modeling approaches, such as Ensemble Learning or Deep Learning, that have proven to outperform PLSR on most soil properties prediction in a large data regime.
As part of our research, the opportunity to train higher capacity models on the KSSL large dataset (all soil taxonomic orders included ~ 50K samples) makes it possible to reach a quality of prediction for exchangeable potassium so far unsurpassed with a Residual Prediction Deviation (RPD) around 3. Potassium is known for its difficulty of being predicted but remains extremely important in the context of remediation of radioactive contamination after a nuclear accident. Potassium can help reduce the uptake of radiocaesium by crops, as it competes with radiocaesium in soil-to-plant transfer.
To ensure informed decision making, we also guarantee that (i) individual predictions uncertainty is estimated (using Monte Carlo Dropout) and (ii) individual predictions can be interpreted (i.e. how much specific MIRS wavenumber regions contribute to the prediction) using methods such as Shapley Additive exPlanations (SHAP) values.
SWMCNL is now a member of the GLOSOLAN network, which helps enhance the usability of MIRS for soil monitoring worldwide. SWMCNL is further developing training packages on the use of traditional and advanced mathematical techniques to process MIRS data for predicting soil properties. This training package has been tested in October 2020 with thirteen staff members of the FAO/IAEA Laboratories in Seibersdorf, Austria.
How to cite: Albinet, F., Dercon, G., and Eguchi, T.: Deep Learning-based Soil Property Prediction for Remediation of Radioactive Contamination in Agriculture, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4416, https://doi.org/10.5194/egusphere-egu21-4416, 2021.
Due to short wavelengths, optical remote sensing data provides information about the properties of very thin soil surface layer. This is especially crucial for arable soils as their surface experiences intense impact of agricultural practices and natural conditions. In temperate zone atmospheric precipitation is one of the main natural factors affecting the surface state of arable soils. It causes the breakdown of soil surface aggregates and the redistribution of formed soil material resulting in surface sealing and the formation of soil crust.
We studied the properties of soil crust and its impact on the detection of soil properties on arable soils of European part of Russia.
Our research showed that the properties of soil surface crust (texture, mineralogical composition, organic matter content, content of microelements, spectral reflectance) differed from the properties of the rest of arable horizon. That discrepancy negatively impacted the performance and reproducibility of the models developed for the detection of arable soil properties and their monitoring on the basis of optical remote sensing data.
We found that the performance of the models for the detection of soil fertility indicators based on Sentinel-2 data varied depending on the acquisition date. Optimal dates were different for different fertility indicators. Introduction of information on soil surface state (% of crust and shadows/cracks) at different acquisition dates as predictors in the models developed based on Sentinel-2 data allowed improving their performance and stability.
Therefore, soil surface state is an important factor which should be considered when developing models for the detection and monitoring of arable soil properties based on optical remote sensing data or proximal sensing of soil surface. Usage of laboratory soil spectra libraries instead of field spectral data leads to less precise prediction models.
The research was supported by the Ministry of science and higher education of Russia (agreement No 075-15-2020-909), and RUDN University Strategic Academic Leadership Program.
How to cite: Prudnikova, E., Savin, I., and Vindeker, G.: Arable soil surface status as a factor affecting the quality of soil properties detection based on remote or proximal sensing technologies, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7532, https://doi.org/10.5194/egusphere-egu21-7532, 2021.
Soil organic carbon (SOC) is of particular interest in the study of agricultural systems as an indicator of soil quality and soil fertility. In the use of Vis-NIR spectroscopy for SOC detection, the interpretation of the spectral response with regards to the importance of individual wavelengths is challenging due to the soil’s composition of multiple organic and minerals compounds. Under field conditions, additional aspects affect the spectral data compared to lab conditions. This study compared the spectral wavelength importance in partial least square regression (PLSR) models for SOC between field and lab conditions. Surface soil samples were obtained from a long-term field experiment (LTE) with high SOC variability located in the state of Saxony-Anhalt, Germany. Data sets of Vis-NIR spectra were acquired in the lab and field using two spectrometers, respectively. Four different preprocessing methods were applied before building the models. Wavelength importance was observed using variable importance in projection. Differences in wavelength importance were observed depending on the measurement device, measurement condition, and preprocessing technique, although pattern matches were identifiable, especially in the NIR range. It is these pattern matches that aid model interpretation to effectively determine SOC under field conditions.
How to cite: Reyes, J. and Ließ, M.: Exploring the Vis-NIR wavelength importance in SOC models under field and lab conditions in a long-term field experiment, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12713, https://doi.org/10.5194/egusphere-egu21-12713, 2021.
Visible-Near Infrared (Vis-NIR) spectroscopy has proven its efficiency in predicting several soil properties such as soil organic carbon (SOC) content. In this preliminary study, we explored the ability of Vis-NIR to assess the temporal evolution of SOC content. Soil samples were collected in a watershed (ORE AgrHys), located in Brittany (Western France). Two sampling campaigns were carried out 5 years apart: in 2013, 198 soil samples were collected respectively at two depths (0-15 and 15-25 cm) over an area of 1200 ha including different land use and land cover; in 2018, 111 sampling points out of 198 of 2013 were selected and soil samples were collected from the same two depths. Whole samples were analyzed for their SOC content and were scanned for their reflectance spectrum. Spectral information was acquired from samples sieved at 2 mm fraction and oven dried at 40°C, 24h prior to spectra acquisition, with a full range Vis-NIR spectroradiometer ASD Fieldspec®3. Data set of 2013 was used to calibrate the SOC content prediction model by the mean of Partial Least Squares Regression (PLSR). Data set of 2018 was therefore used as test set. Our results showed that the variation ∆SOCobsobtained from observed values in 2013 and 2018 (∆SOCobs = Observed SOC (2018) - Observed SOC (2013)) is ranging from 0.1 to 25.9 g/kg. Moreover, our results showed that the prediction performance of the calibrated model was improved by including 11 spectra of 2018 in the 2013 calibration data set (R²= 0.87, RMSE = 5.1 g/kg and RPD = 1.92). Furthermore, the comparison of predicted and observed ∆SOC between 2018 and 2013 showed that 69% of the variations were of the same sign, either positive or negative. For the remaining 31%, the variations were of opposite signs but concerned mainly samples for which ∆SOCobs is less than 1,5 g/kg. These results reveal that Vis-NIR spectroscopy was potentially appropriate to detect variations of SOC content and are encouraging to further explore Vis-NIR spectroscopy to detect changes in soil carbon stocks.
How to cite: Zayani, H., Fouad, Y., Michot, D., Kassouk, Z., Lili-Chabaane, Z., and Walter, C.: Ability of Vis-PIR spectroscopy to monitor changes in organic carbon of loamy soils at two depths, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-15021, https://doi.org/10.5194/egusphere-egu21-15021, 2021.
Optical remote sensing and in particular hyperspectral or imaging spectroscopy remote sensing has been long proved to be an adequate method to predict topsoil organic carbon (Corg) content with good accuracy when the soils are well exposed and undisturbed. Several recent studies demonstrated further in science cases the potential of multispectral Copernicus Sentinel-2 data for bare soils Corg prediction, although challenges were reported related to the impact of disturbing factors. Disturbing factors that can affect the prediction and performances of soil surface properties from optical remote sensing are several and can be e.g. due to mixing in the field-of-view with partial vegetation cover depending on the landscape fragmentation. Most pixels at the remote sensing level are composites and in croplands, mixtures of soils with trees or green plants, or mixture with crop residues after harvest are likely. Another factor might be the presence of residual soil moisture or standing water after rain events. Soil reflectance decreases with increasing soil moisture and increasing soil roughness. Soil Surface roughness changes are observed due to variations in soil texture and to variable microtopography. Possible angular and solar illumination changes may affect the soil reflectance as well.
In the frame of the ESA WORLDSOILS Project (https://www.world-soils.com) aiming at developing a pre-operational Soil Monitoring System to provide yearly estimations of soil organic carbon at global scale based on space-based EO data, we are working on the development of a spatially upscaled soil spectral library (SUSSL). The SUSSL is based on a sub-selection of the European LUCAS soil database, and includes simulation of realistic scenarios of ‘landscape-like’ cropland reflectance data with effect of mixture with green and dry vegetation, effect of varying soil moisture content, and effect of variable soil roughness. This database is further convoluted to the different spectral response functions of several EO sensors to simulate EO view of surface reflectances in croplands. In a next step, the SUSSL shall be used for the test and validation of different correction, disaggregation and unmixing techniques to assess the capabilities of the retrieval of undisturbed surface reflectance, to which soil prediction models can be applied with increased accuracy. In this talk, we will present the database developed, including methodological choices and parameter selections for the simulation of the different disturbing effects. Further, preliminary assessments will be shown on the uncertainties of the undisturbed vs. disturbed signal and impact on soil properties prediction.
How to cite: Chabrillat, S., Milewski, R., Kuester, T., Dvorakova, K., and van Wesemael, B.: Development of a Spatially Upscaled Soil Spectral Library (SUSSL) of cropland signatures for EO sensors data simulation and calibration/validation of topsoil organic carbon prediction models, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-14365, https://doi.org/10.5194/egusphere-egu21-14365, 2021.
Pilot studies have demonstrated the potential for remote sensing techniques for soil organic carbon (SOC) mapping in exposed croplands. However, the use of remote sensing for SOC prediction is often hindered by disturbing factors at the soil surface such as photosynthetic active and nonphotosynthetic active vegetation, variation in soil moisture or surface roughness. With the increasing amount of freely available satellite data, many studies have focused on stabilizing the soil reflectance by building image composites that are generated using a set of criteria. These composites tend to minimize and cancel out the disturbing effects. Here we aim to develop a robust method that allows selecting Sentinel-2 (S-2) pixels that are not affected by the following disturbing factors: crop residues, surface roughness and soil moisture. We selected all S-2 cloud-free images covering the Loam Belt of Belgium from January 2019 to December 2020 (in total 38 images). We then built four exposed soil composites based on four sets of criteria: (1) NDVI < 0.25, (2) NDVI < 0.25 & Normalized Burn Ratio (NBR2) < 0.07, (3) the ‘greening-up’ period of a crop and (4) the ‘greening-up’ period of a crop & NBR2 < 0.07. The ‘greening-up’ period was selected based on the NDVI timeline, where ‘greening-up’ is considered as the last date of acquisition where the soil is bare (NDVI < 0.25) before the crop develops (NDVI > 0.6).,We then built a partial least square regression (PLSR) model with 10-fold cross-validation to estimate the SOC content based on 137 georeferenced calibration samples on the four above described composites. We obtained a non-satisfactory result for composites (1) to (3): R² = 0.22, RMSE = 3.46 g C kg-1 and RPD 1.12 for (1), R² = 0.19, RMSE = 3.43 g C kg-1 and RPD 1.10 for (2) and R² = 0.15, RMSE = 2.74 g C kg-1 and RPD 1.06 for (3). We, however, obtained a satisfactory result for composite (4): R² = 0.54, RMSE = 2.09 g C kg-1 and RPD 1.68. Hence, the ‘greening-up’ method combined with a strict NBR2 threshold allows selecting the purest exposed soil pixels suitable for SOC prediction. The limit of this method might be the surface coverage, which for a two-year period reached 47% of croplands, compared to 89% exposure if only the NDVI threshold is applied.
How to cite: Dvorakova, K. and van Wesemael, B.: Sentinel-2 exposed soil composite for soil organic carbon prediction: the ‘greening-up’ method for detecting suitable images, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11153, https://doi.org/10.5194/egusphere-egu21-11153, 2021.
In terms of agronomy, soil organic carbon (SOC) content is important for crop growth and development. From the environmental viewpoint, SOC sequestration is essential to mitigate the emission of greenhouse gases into the atmosphere. The use of sensors for carbon monitoring over croplands is a key issue in recent works. Sentinel-1/2 (S1, S2) satellites acquire data with regular frequency (weekly) and high spatial resolution (10 and 20 meters). Previous studies have demonstrated their potential for quantification of soil attributes including topsoil organic carbon content on single dates. Soil surface roughness and soil moisture influence the performance of spectral models according to acquisition date, particularly surface soil moisture (SM), as shown by multidate models of predicted SOC content (Vaudour et al., 2021). Still, the sensitivity of Sentinel-1/2 to SM must be better understood and exploited for a given single date. A possible solution to determine the influence of SM on single date model performance consists of including it as a covariate.
In order to predict the topsoil SOC content over croplands in the Pyrenees region, France (22177 km²), this study addresses: (i) the influence of the Sentinel image date and that of the soil sampling year; (ii) the contribution of SM products derived from the Sentinel-1/2 data (El Hajj et al., 2017) in the spectral models.
The influence of the image date and soil sampling date was analyzed for springs 2017 and 2018. Clouds, shadows and NDVI (> 0.35) values were excluded from the images. Best single performances (RPD ≥ 1.3) were scored for soil sampling sets collected in 2016-2018. The same dates were analyzed using either SM maps, or signal values of VV and VH polarizations from S1 images. SM or polarization values were extracted for each sample and integrated into the partial least squares regression (PLSR) models, respectively. The best performance (RPD = 1.57) was obtained with SM as a covariate in 2017, with lowest mean SM throughout the map.
El Hajj, M.; Baghdadi, N.; Zribi, M.; Bazzi, H. Synergic Use of Sentinel-1 and Sentinel-2 Images for Operational Soil Moisture Mapping at High Spatial Resolution over Agricultural Areas. Remote Sensing 2017, 9, 1292, doi:10.3390/rs9121292.
Vaudour, E.; Gomez, C.; Lagacherie, P.; Loiseau, T.; Baghdadi, N.; Urbina-Salazar, D.; Loubet, B.; Arrouays, D. Temporal Mosaicking Approaches of Sentinel-2 Images for Extending Topsoil Organic Carbon Content Mapping in Croplands. International Journal of Applied Earth Observation and Geoinformation 2021, 96, 102277, doi:10.1016/j.jag.2020.102277.
How to cite: Urbina Salazar, D., Vaudour, E., Baghdadi, N., Ceschia, E., and Arrouays, D.: Combined use of Sentinel-2 images and Sentinel-1-derived moisture maps for soil organic carbon content mapping in croplands, South-western France, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8836, https://doi.org/10.5194/egusphere-egu21-8836, 2021.
The success of any civil engineering structure's foundation design depends upon the accuracy of estimation of soil’s ultimate bearing capacity. Numerous numerical approaches have been proposed to estimate the foundation's bearing capacity value to avoid repetitive and expensive experimental work. All these models have their advantages and disadvantages. In this study, we compiled all the governing equations mentioned in Bureau of Indian standard IS:6403-1981 and modify the equation for Ultimate Bearing Capacity. The equation was modified by considering two new parameters, K1(for general shear) and K2 (for local shear) so that a common governing equation can be used for both general and local shear failure criteria. The program used for running the model was written in MATLAB language code and verified with the observed field data. Results indicate that the proposed model accurately characterized the ultimate, safe, and allowable bearing capacity of a shallow footing at different depths. The correlation coefficients between the observed and model-predicted bearing capacity values for a 2m foundation depth with footing size of 1.5 ×1.5, 2.0 × 2.0, and 2.5 × 2.5 m are 0.95, 0.94, and 0.96. A similar result was noted for the other foundation depth and footing size. Findings show that the model can be used as a reliable tool for predicting the bearing capacity of shallow foundations at any given depth. Moreover, the formulated model can also be used for the transition zone between general and local shear failure conditions.
How to cite: Hinge, G., Kumar Das, J., and Bharali, B.: A numerical approach to predict soil bearing potential for isolated footing, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-321, https://doi.org/10.5194/egusphere-egu21-321, 2021.
Earth observation is invaluable for the agricultural sector as well as the critical metals sector, providing cost-effective, spatially comprehensive information about Earth’s surface composition from the regional to paddock/mine-scale. A wide range of remote sensing instruments are used to monitor soils, to give information on properties such as moisture and mineralogy. At the same time, remote sensing data facilitate the discovery and mining of mineral deposits, including iron ore, copper and other metals critical for the transition of the fossil fuel-based energy sector to a sustainable, renewable energy future. One common factor of these two sectors is that all Earth observation systems require calibration sites that help to ensure the data being collected is of high accuracy. Another common factor is that both sectors require ground validation of the remotely sensed data, producing a plethora of publicly available Earth surface data distributed across numerous web portals and platforms. Both sectors aim, ultimately, towards characterising the composition of the subsurface - which starts in both sectors at Earth’s surface and reaches to 10s or even 100s of metres below. This can be achieved by developing conceptual models that describe the weathering of bedrock in the soil/regolith. In mineral resource exploration, specific weathering-resistant minerals (e.g. talc) can be traced at Earth’s surface by means of Earth observation to characterise the type of bedrock through cover (i.e. beneath the soil/regolith). Another example is the mapping of differences in kaolin crystallinity at Earth’s surface and in the subsurface (e.g. drilling, trenches) to infer the distribution of in-situ versus transported regolith, which is of key importance for raw materials exploration. Remote sensing is also commonly used for collecting baseline environmental data prior to mining and for monitoring its impact on the environment during and after the process. In soil science, infrared spectral measurements have been conducted on soil samples in laboratories for estimation of soil properties, such as soil carbon, pH, EC. These estimations require a training library as well as standardised preparation of the samples and measurement technique. The ultimate goal is the accurate measurement of these soil properties using remote sensing, where complex variance of the nature of the materials and illumination conditions exists.
This paper discusses opportunities for sharing facilities, data, workflows and methods for collecting, processing and interpreting remote and proximal multi- and hyperspectral sensing technologies. For this, publicly available mineralogical and geochemical data sets collected from the critical zone, such as in the frame of the National Geochemical Survey of Australia (NGSA; https://www.ga.gov.au/about/projects/resources/national-geochemical-survey) project and AuScope’s National Virtual Core Library Infrastructure Program (NVCL; https://www.auscope.org.au/nvcl), as well as publicly available Earth observation products, such as the Australian ASTER Geoscience Products, will be used to demonstrate the multidisciplinary applications of multi- and hyperspectral remote and proximal sensing data. For the benefit of meeting the United Nations’ Sustainable Development Goals, agriculture, resources and environment sectors should overcome unnecessary competition and work hand in hand.
How to cite: Laukamp, C. and Lau, I. C.: One’s soil is another one’s regolith – let’s combine our efforts in investigating the critical zone, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10474, https://doi.org/10.5194/egusphere-egu21-10474, 2021.
High spatial and temporal soil information is crucial to analyze soil developments and for monitoring long term changes to avoid soil degradation. A sufficient soil organic carbon (SOC) content is one of the key soil properties to achieve sustainable high productivity of soils, soil health and increased agroecosystem resiliency. For the usage of remote sensing approaches, naturally exposed soils in Germany occur rarely. Mainly agricultural regions can provide areas of exposed soils for short periods of time during a year. The Soil Composite Mapping Processor (SCMaP) is a fully automated approach to make use of per-pixel based bare-soil compositing to overcome the issue of limited soil exposure based on multispectral Landsat (TM 4, ETM 5, ETM+ 7 and OLI 8) imagery for individually determined time periods between 1984 and 2019.
Due to the high spatial and temporal resolution the SCMaP soil reflectance composites contain a considerable potential to derive detailed soil parameters as the SOC contents of exposed soils to add information to existing soil maps on field scale for areawide applications. Besides the soil reflectance composites several field soil samples provided by different federal authorities build the data base for the SOC modeling. Machine learning (ML) algorithms incl. Partial Least Squares and Random Forest regression with various inputs and set-ups are used and applied for several test areas in Germany. Furthermore, the capabilities of different compositing lengths (5-, 10- and 30-years) to derive spatial SOC contents are tested. The results and the validation of the different ML approaches and compositing lengths will be shown, providing insight into the benefits of this approach.
How to cite: Zepp, S., Bachmann, M., Möller, M., van Wesemael, B., Steininger, M., Wiesmeier, M., and Heiden, U.: Testing different remote sensing compositing periods for SOC content extraction in areas across Germany, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7371, https://doi.org/10.5194/egusphere-egu21-7371, 2021.
Organic matter added to agricultural soil determines the C balance and the nutrient cycling in these ecosystems. Organic fertilisation can result in the accumulation of C in soil but can also stimulate the decomposition of the existing soil C pool, as the incorporation of an easily accessible energy-rich substrate often trigger the growth and activity of decomposer. We monitored the fate of two types of organic material (wheat straw and green manure) during the first stages of their decomposition into the soil. For this, we incubated 1-m soil columns amended with the two organic fertilisers either into the topsoil or into the subsoil. We measured changes in C and N contents, and used 13C-NMR to resolve the structural group composition of the added organic material. We also scanned the incubated samples with a hyperspectral camera and developed predictive models for C to N and for alkyl to O-alkyl ratios at a very fine spatial resolution (53 x 53 µm2 per pixel) for organic particles in the whole soil cores.
The approach based on hyperspectral imaging was successful to follow the decomposition dynamics of POM during the incubation, and the associated decreases in C to N and increases in alkyl to O-alkyl ratios at a very fine spatial resolution, showing how different parts of the organic particles underwent distinct decomposition. We also observed contrasting decomposition dynamics between the wheat straw and the green manure. This method can bring new information about the first steps of fresh organic matter decomposition in soils and develop our general understanding of the soil organic matter decomposition continuum.
How to cite: Guigue, J., Just, C., Luo, S., Fogt, M., Schloter, M., Kögel-Knabner, I., and Hobley, E.: Mapping molecular information of organic amendments during their decomposition in soil derived from hyperspectral imaging, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-14340, https://doi.org/10.5194/egusphere-egu21-14340, 2021.
We are sorry, but presentations are only available for users who registered for the conference. Thank you.
We are sorry, but presentations are only available for users who registered for the conference. Thank you.