A Semi-Automated Object-Based Gully Networks Detection Using Different Machine Learning Models: A Case Study of Bowen Catchment, Queensland, Australia.

Gully erosion is a dominant source of sediment and particulates to the Great Barrier Reef (GBR) World Heritage area. We selected the Bowen catchment, a tributary of the Burdekin Basin, as our area of study; the region is associated with a high density of gully networks. We aimed to use a semi-automated object-based gully networks detection process using a combination of multi-source and multi-scale remote sensing and ground-based data. An advanced approach was employed by integrating geographic object-based image analysis (GEOBIA) with current machine learning (ML) models. These included artificial neural networks (ANN), support vector machines (SVM), and random forests (RF), and an ensemble ML model of stacking to deal with the spatial scaling problem in gully networks detection. Spectral indices such as the normalized difference vegetation index (NDVI) and topographic conditioning factors, such as elevation, slope, aspect, topographic wetness index (TWI), slope length (SL), and curvature, were generated from Sentinel 2A images and the ALOS 12-m digital elevation model (DEM), respectively. For image segmentation, the ESP2 tool was used to obtain three optimal scale factors. On using object pureness index (OPI), object matching index (OMI), and object fitness index (OFI), the accuracy of each scale in image segmentation was evaluated. The scale parameter of 45 with OFI of 0.94, which is a combination of OPI and OMI indices, proved to be the optimal scale parameter for image segmentation. Furthermore, segmented objects based on scale 45 were overlaid with 70% and 30% of a prepared gully inventory map to select the ML models' training and testing objects, respectively. The quantitative accuracy assessment methods of Precision, Recall, and an F1 measure were used to evaluate the model's performance. Integration of GEOBIA with the stacking model using a scale of 45 resulted in the highest accuracy in detection of gully networks with an F1 measure value of 0.89. Here, we conclude that the adoption of optimal scale object definition in the GEOBIA and application of the ensemble stacking of ML models resulted in higher accuracy in the detection of gully networks.

Collaborative involvement of woody plant roots and rhizosphere microorganisms in the formation of pedogenetic clays.

BACKGROUND AND AIMS: Previous studies have described the laying down of specific B horizons in south-western Australian ecosystems. This paper presents biomolecular, morphological and physicochemical analyses elucidating the roles of specific woody plant taxa and rhizosphere bacteria in producing these phenomena. METHODS: Clayey deposits within lateral root systems of eucalypts and appropriate background soil samples were collected aseptically at multiple locations on sand dunes flanking Lake Chillinup. Bacterial communities were profiled using tagged next-generation sequencing (Miseq) of the 16S rRNA gene and assigned to operational taxonomic units. Sedimentation, selective dissolution and X-ray diffraction analyses quantitatively identified clay mineral components. Comparisons were made of pedological features between the above eucalypt systems, giant podzols under proteaceous woodland on sand dunes at the study site of Jandakot and apparently similar systems observed elsewhere in the world. KEY RESULTS: Bacterial communities in clay pods are highly diverse, resolving into 569 operational taxonomic units dominated by Actinobacteria at 38.0-87.4 % of the total reads. Multivariate statistical analyses of community fingerprints demonstrated substrate specificity. Differently coloured pods on the same host taxon carry distinctive microfloras correlated to diversities and abundances of Actinobacteria, Acidobacteria, Firmicutes and Proteobacteria. A number of these microbes are known to form biominerals, such as phyllosilicates, carbonates and Fe-oxides. A biogenic origin is suggested for the dominant identified mineral precipitates, namely illite and kaolinite. Comparisons of morphogenetic features of B horizons under eucalypts, tree banksias and other vegetation types show remarkably similar developmental trajectories involving pods of precipitation surrounding specialized fine rootlets and their orderly growth to form a continuous B horizon. CONCLUSIONS: The paper strongly supports the hypothesis that B-horizon development is mediated by highly sophisticated interactions of host plant and rhizosphere organisms in which woody plant taxa govern overall morphogenesis and supply of mineral elements for precipitation, while rhizosphere microorganisms execute biomineralization processes.

Characterization of dissolved organic matter for prediction of trihalomethane formation potential in surface and sub-surface waters.

Dissolved organic matter (DOM) in surface waters used for drinking purposes can vary markedly in character dependent on their sources within catchments. The character of DOM further influences the formation of disinfection by products when precursor DOM present in drinking water reacts with chlorine during disinfection. Here we report the development of models that describe the formation potential of trihalomethanes (THMFP) dependent on the character of DOM in waters from discrete catchments with specific land-use and soil textures. DOM was characterized based on UV absorbance at 254 nm, apparent molecular weight and relative abundances of protein-like and humic-like compounds. DOM character and Br concentration (up to 0.5 mg/L) were used as variables in models (R(2)>0.93) of THMFP, which ranged from 19 to 649 µg/L. Chloroform concentration (12-594 µg/L) and relative abundance (27-99%) were first modeled (R(2)>0.85) and from these, the abundances of bromodichloromethane and chlorodibromomethane estimated using power and exponential functions, respectively (R(2)>0.98). From these, the abundance of bromoform is calculated. The proposed model may be used in risk assessment of catchment factors on formation of trihalomethanes in drinking water, in context of treatment efficiency for removal of organic matter.

The effect of vegetation and soil texture on the nature of organics in runoff from a catchment supplying water for domestic consumption.

The influence of vegetation and soil texture on the concentration and character of dissolved organic matter (DOM) present in runoff from the surface and sub-surface of zero order catchments of the Myponga Reservoir-catchment (South Australia) was investigated to determine the impacts of catchment characteristics and land management practices on the quality of waters used for domestic supply. Catchments selected have distinct vegetative cover (grass, native vegetation or pine) and contrasting texture of the surface soil horizon (sand or clay loam/clay). Water samples were collected from three slope positions (upper, middle, and lower) at soil depths of ~30 cm and ~60 cm in addition to overland flows. Filtered (0.45 µm) water samples were analyzed for dissolved organic carbon (DOC) and UV-visible absorbance and by F-EEM and HPSEC with UV and fluorescence detection to characterize the DOM. Surface and sub-surface runoff from catchments with clay soils and native vegetation or grass had lower DOC concentrations and lower relative abundances of aromatic, humic-like and high molecular weight organics than runoff from sandy soils with these vegetative types. Sub-surface flows from two catchments with Pinus radiata had similar DOC concentrations and DOM character, regardless of marked variation in surface soil texture. Runoff from catchments under native vegetation and grass on clay soils resulted in lower DOC concentrations and hence would be expected to have lower coagulant demand in conventional treatment for potable water supply than runoff from corresponding sandy soil catchments. However, organics in runoff from clay catchments would be more difficult to remove by coagulation. Surface waters from the native vegetation and grass catchments were generally found to have higher relative abundance of organic compounds amenable to removal by coagulation compared with sub-surface waters. Biophysical and land management practices combine to have a marked influence on the quality of source water used for domestic supply.

Sorption of dissolved organic matter in salt-affected soils: effect of salinity, sodicity and texture.

Loss of dissolved organic matter (DOM) from soils can have negative effects on soil fertility and water quality. It is known that sodicity increases DOM solubility, but the interactive effect of sodicity and salinity on DOM sorption and how this is affected by soil texture is not clear. We investigated the effect of salinity and sodicity on DOM sorption in soils with different clay contents. Four salt solutions with different EC and SAR were prepared using combinations of 1M NaCl and 1M CaCl(2) stock solutions. The soils differing in texture (4, 13, 24 and 40% clay, termed S-4, S-13, S-24 and S-40) were repeatedly leached with these solutions until the desired combination of EC and SAR (EC(1:5) 1 and 5dSm(-1) in combination with SAR <3 or >20) was reached. The sorption of DOC (derived from mature wheat straw) was more strongly affected by SAR than by EC. High SAR (>20) at EC1 significantly decreased sorption in all soils. However, at EC5, high SAR did not significantly reduce DOC sorption most likely because of the high electrolyte concentration of the soil solution. DOC sorption was greatest in S-24 (which had the highest CEC) at all concentrations of DOC added whereas DOC sorption did not differ greatly between S-40 and S-4 or S-13 (which had higher concentrations of Fe/Al than S-40). DOC sorption in salt-affected soil is more strongly controlled by CEC and Fe/Al concentration than by clay concentration per se except in sodic soils where DOC sorption is low due to the high sodium saturation of the exchange complex.

Selenate-enriched urea granules are a highly effective fertilizer for selenium biofortification of paddy rice grain.

This study examined the effects of applied selenium (Se) species, time of application, method of application, and soil water management regimen on the accumulation of Se in rice plants. Plants were grown to maturity in a temperature- and humidity-controlled growth chamber using three water management methods: field capacity (FC), submerged until harvest, and submerged and drained 2 weeks before harvest. Two Se species, selenate (SeO4(2-)) and selenite (SeO3(2-)), were applied at a rate equivalent to 30 g ha(-1). Four application methods were employed as follows: (i) Se applied at soil preparation, (ii) Se-enriched urea granules applied to floodwater at heading; (iii) foliar Se applied at heading; and (iv) fluid fertilizer Se applied to soil or floodwater at heading. Total Se concentrations in rice grains, husks, leaves, culms, and roots were measured, as well as Se speciation in grains from the Se-enriched urea granule treatment. Highest Se concentrations in the grain occurred with SeO4(2-) and with fertilizer applied at heading stage; SeO4(2-)-enriched urea granules applied at heading increased grain Se concentrations 5-6-fold (by 450-600 µg kg(-1)) compared to the control (no fertilizer Se applied) in all water treatments. Under paddy conditions other Se fertilization strategies were much less effective. Drainage before harvesting caused Se to accumulate in/on rice roots, possibly through adsorption onto iron plaque on roots. Rice grains contained Se mainly in the organic form as selenomethionine (SeM), which comprised >90% of the total grain Se in treatments fertilized with SeO4(2-)-enriched urea granules. The results of this study clearly show that of the fertilizer strategies tested biofortification of Se in rice grains can best be achieved in lowland rice by broadcast application of SeO4(2-)-enriched urea granules to floodwater at heading stage.

Spatial distribution of diuron sorption affinity as affected by soil, terrain and management practices in an intensively managed apple orchard.

We investigated how the sorption affinity of diuron (3'-(3,4-dichlorophenyl)-1,1-dimenthyl-urea), a moderately hydrophobic herbicide, is affected by soil properties, topography and management practices in an intensively managed orchard system. Soil-landscape analysis was carried out in an apple orchard which had a strong texture contrast soil and a landform with relief difference of 50 m. Diuron sorption (K(d)) affinity was successfully predicted (R(2)=0.79; p<0.001) using a mid-infrared - partial least squares model and calibrated against measured data using a conventional batch sorption technique. Soil and terrain properties explained 75% of the variance of diuron K(d) with TOC, pH(w), slope and WI as key variables. Mean diuron K(d) values were also significantly different (p<0.05) between alley and tree line and between the different management zones. Soil in the tree line generally had lower sorption capacity for diuron than soil in the alleys. Younger stands, which were found to have lower TOC than in the older stands, also had lower diuron K(d) values. In intensively managed orchards, sorption affinity of pesticides to soils was not only affected by soil properties and terrain attributes but also by management regime.

Introducing a decomposition rate modifier in the Rothamsted Carbon Model to predict soil organic carbon stocks in saline soils.

Soil organic carbon (SOC) models such as the Rothamsted Carbon Model (RothC) have been used to estimate SOC dynamics in soils over different time scales but, until recently, their ability to accurately predict SOC stocks/carbon dioxide (CO(2)) emissions from salt-affected soils has not been assessed. Given the large extent of salt-affected soils (19% of the 20.8 billion ha of arable land on Earth), this may lead to miss-estimation of CO(2) release. Using soils from two salt-affected regions (one in Punjab, India and one in South Australia), an incubation study was carried out measuring CO(2) release over 120 days. The soils varied both in salinity (measured as electrical conductivity (EC) and calculated as osmotic potential using EC and water content) and sodicity (measured as sodium adsorption ratio, SAR). For soils from both regions, the osmotic potential had a significant positive relationship with CO(2)-C release, but no significant relationship was found between SAR and CO(2)-C release. The monthly cumulative CO(2)-C was simulated using RothC. RothC was modified to take into account reductions in plant inputs due to salinity. A subset of non-salt-affected soils was used to derive an equation for a "lab-effect" modifier to account for changes in decomposition under lab conditions and this modifier was significantly related with pH. Using a subset of salt-affected soils, a decomposition rate modifier (as a function of osmotic potential) was developed to match measured and modelled CO(2)-C release after correcting for the lab effect. Using this decomposition rate modifier, we found an agreement (R(2) = 0.92) between modelled and independently measured data for a set of soils from the incubation experiment. RothC, modified by including reduced plant inputs due to salinity and the salinity decomposition rate modifier, was used to predict SOC stocks of soils in a field in South Australia. The predictions clearly showed that SOC stocks are reduced in saline soils. Therefore both the decomposition rate modifier and plant input modifier should be taken into account when accounting for SOC turnover in saline soils. Since modeling has previously not accounted for the impact of salinity, our results suggest that previous predictions may have overestimated SOC stocks.

Solubility and batch retention of CeO2 nanoparticles in soils.

There is a paucity of information on the environmental fate of cerium oxide nanoparticles (CeO2 NPs) for terrestrial systems that may be exposed to CeO2 NPs by the application of biosolids derived from wastewater treatment systems. Using ultrafiltration (UF), dissolution, and nonequilibrium retention (Kr) values of citrate-coated (8 nm diameter) CeO2 NPs and partitioning (Kd) values of dissolved Ce(III) and Ce(IV) were obtained in suspensions of 16 soils with a diversity of physicochemical properties. Dissolution of CeO2 NPs studied in solutions was only significant at pH 4 and was less than 3.1%, whereas no dissolved Ce was detected in soils spiked with CeO2 NPs. Kr values of CeO2 NP were low (median Kr=9.6 L kg(-1)) relative to Kd values of dissolved CeIII and CeIV (median Kd=3763 and 1808 L kg(-1), respectively), suggesting low CeO2 NP retention in soils. Surface adsorption of phosphate to CeO2 NP caused a negative zeta potential, which may explain the negative correlation of log Kr values with dissolved phosphate concentrations and the significant reduction of Kr values upon addition of phosphate to soils. The positive correlation of Kr values with clay content suggested heterocoagulation of CeO2 NPs with natural colloids in soils. Co-addition of CeO2 NPs with biosolids, on the other hand, did not affect retention.

Direct comparison between visible near- and mid-infrared spectroscopy for describing diuron sorption in soils.

Both visible near-infrared (VNIR) and mid-infrared (MIR) spectroscopy have been claimed to better predict pesticide sorption in soils than other methods. We compared the performances of VNIR and MIR spectroscopy for predicting both organic carbon content (foc) and the sorption affinity (Kd) of diuron in 112 surface soils from South Australia. Separate calibration models were developed between VNIR and MIR spectra, and foc and Kd using partial least-squares (PLS) regression. MIR clearly outperformed VNIR for predictions of both foc and Kd in soils. Correlation (R2) and accuracy (RPD) indices were 0.4 and 1.3 for the VNIR-PLS model versus 0.8 and 2.3 for the MIR-PLS model, respectively, for Kd prediction. PLS loadings for sorption prediction were compared in terms of the soil information they contained. While VNIR loading did not include any direct spectral information regarding soil minerals, MIR loading included peaks associated with sand, clays, and carbonates. Perhaps by better predicting foc and integrating the effects of OC as well as minerals, the MIR-PLS model provided a better prediction for diuron Kd values in our calibration set.

The effect of solvent-conditioning on soil organic matter sorption affinity for diuron and phenanthrene.

The effect of solvent-conditioning on the sorption of diuron and phenanthrene was investigated. The organic carbon-normalized sorption coefficients (K(OC)) for diuron and phenanthrene (determined from single initial concentrations of 0.8mgL(-1) and 1.5mgL(-1), respectively) were consistently higher following solvent-conditioning of a whole soil with five organic solvents (acetonitrile, acetone, methanol, chloroform and dichloromethane). The relative increase in K(OC) was inversely related to the polarity of the conditioning solvent (i.e. greater increases in K(OC) were observed for the least polar solvents: chloroform and dichloromethane). The effect of solvent-conditioning on the sorption properties of the same soil that had been lipid-extracted using accelerated solvent extraction (ASE) was also investigated. Since lipid extraction involves treatment with a non-polar solvent (95:5 dichloromethane:methanol) one may have expected no further increase in K(OC) on solvent-conditioning. On the contrary, the lipid-extracted soil exhibited very similar increases in K(OC) as the whole soil. This demonstrated that lipid removal and solvent-conditioning, which both increased K(OC) for this soil, are quite separate phenomena.

The effect of lipids on the sorption of diuron and phenanthrene in soils.

The influence of lipids on the sorption of diuron and phenanthrene to soils was investigated. Accelerated solvent extraction (ASE) was used to extract lipids from twelve soil horizons. Extractable lipids accounted for 3-13% of organic C. The organic carbon-normalized sorption coefficients (K(OC)) for diuron and phenanthrene were consistently higher for the lipid-extracted soils than for the whole soils (average of 31% for diuron and 29% for phenanthrene), indicating that lipids compete for or block sorption sites on the organic matter. Sorption experiments on one pair of HF-treated soils indicated that the blocking effects of minerals and lipids are independent, since lipid extraction and HF-treatment combined increased K(OC) by more than either treatment alone. Lipids extracted from whole and HF-treated soils were very similar in composition, consisting predominantly of long-chain polymethylene structures. K(OC) of the lipid itself was lower than for any of the whole soils and soil fractions (lipid extracted and HF-treated) for diuron, but higher for phenanthrene. Solid-state (13)C NMR spectra of the HF-treated soils before and after lipid extraction indicated that 15-20% of alkyl C was removed by ASE and that no other structures were affected.

Midinfrared spectroscopy and chemometrics to predict diuron sorption coefficients in soils.

The potential of mid-infrared (MIR) spectroscopy in combination with partial least-squares (PLS) regression was investigated to predict the soil sorption (distribution) coefficient (K(d)) of a nonionic pesticide (diuron). A calibration set of 101 surface soils collected from South Australia was utilized for reference sorption data and MIR spectra. Principal component analysis (PCA) was performed on the spectra to detect spectral outliers. The MIR-PLS model was developed and validated by dividing the initial data set into four validation sets. The model resulted in a coefficient of determination (R2) of 0.69, a standard error (SE) of 5.57, and a residual predictive deviation (RPD) of 1.63. The normalized sorption coefficient for the organic compound (K(oc)) approach, on the other hand, resulted in R2, SE, and RPD values of 0.42, 7.26, and 1.25, respectively. However, the significant statistical difference between the two models was mainly due to two outliers detected via PCA. Apart from spectral outliers, the performance of the two models was essentially similar for the rest of the calibration set. Outlier detection by the MIR-PLS model may gainfully be employed as a tool for improving prediction of K(d). The MIR-based model can provide a direct estimation of K(d) values based on the integrated properties of organic and mineral matter reflected in the infrared spectra.

Separating the effects of organic matter-mineral interactions and organic matter chemistry on the sorption of diuron and phenanthrene.

Even though it is well established that soil C content is the primary determinant of the sorption affinity of soils for non-ionic compounds, it is also clear that organic carbon-normalized sorption coefficients (K(OC)) vary considerably between soils. Two factors that may contribute to K(OC) variability are variations in organic matter chemistry between soils and interactions between organic matter and soil minerals. Here, we quantify these effects for two non-ionic sorbates-diuron and phenanthrene. The effect of organic matter-mineral interactions were evaluated by comparing K(OC) for demineralized (HF-treated) soils, with K(OC) for the corresponding whole soils. For diuron and phenanthrene, average ratios of K(OC) of the HF-treated soils to K(OC) of the whole soils were 2.5 and 2.3, respectively, indicating a substantial depression of K(OC) due to the presence of minerals in the whole soils. The effect of organic matter chemistry was determined by correlating K(OC) against distributions of C types determined using solid-state (13)C NMR spectroscopy. For diuron, K(OC) was positively correlated with aryl C and negatively correlated with O-alkyl C, for both whole and HF-treated soils, whereas for phenanthrene, these correlations were only present for the HF-treated soils. We suggest that the lack of a clear effect of organic matter chemistry on whole soil K(OC) for phenanthrene is due to an over-riding influence of organic matter-mineral interactions in this case. This hypothesis is supported by a correlation between the increase in K(OC) on HF-treatment and the soil clay content for phenanthrene, but not for diuron.

Clear effects of soil organic matter chemistry, as determined by NMR spectroscopy, on the sorption of diuron.

Organic matter has long been recognized as the main sorbent phase in soils for hydrophobic organic compounds (HOCs). In recent times, there has been an increasing realization that not only the amount, but also the chemical composition, of organic matter can influence the sorption properties of a soil. Here, we show that the organic carbon-normalized sorption coefficient (K(OC)) for diuron is 27-81% higher in 10 A11 horizons than in 10 matching A12 horizons for soils collected from a small (2ha) field. K(OC) was generally greater for the deeper (B) horizons, although these values may be inflated by sorption of diuron to clays. Organic matter chemistry of the A11 and A12 horizons was determined using solid-state 13C nuclear magnetic resonance (NMR) spectroscopy. K(OC) was positively correlated with aryl C (r2=0.59, significance level 0.001) and negatively correlated with O-alkyl C (r2=0.84, significance level <0.001). This is only the second report of correlations between whole soil K(OC) and NMR-derived measures of organic matter chemistry. We suggest that this success may be a consequence of limiting this study to a very small area (a single field). There is growing evidence that interactions between organic matter and clay minerals strongly affect K(OC). However, because the soil mineralogy varies little across the field, the influence of these interactions is greatly diminished, allowing the effect of organic matter chemistry on K(OC) to be seen clearly. This study in some way reconciles studies that show strong correlations between K(OC) and the chemistry of purified organic materials and the general lack of such correlations for whole soils.

Phosphorus transfer in surface runoff from intensive pasture systems at various scales: a review.

Phosphorus transfer in runoff from intensive pasture systems has been extensively researched at a range of scales. However, integration of data from the range of scales has been limited. This paper presents a conceptual model of P transfer that incorporates landscape effects and reviews the research relating to P transfer at a range of scales in light of this model. The contribution of inorganic P sources to P transfer is relatively well understood, but the contribution of organic P to P transfer is still relatively poorly defined. Phosphorus transfer has been studied at laboratory, profile, plot, field, and watershed scales. The majority of research investigating the processes of P transfer (as distinct from merely quantifying P transfer) has been undertaken at the plot scale. However, there is a growing need to integrate data gathered at a range of scales so that more effective strategies to reduce P transfer can be identified. This has been hindered by the lack of a clear conceptual framework to describe differences in the processes of P transfer at the various scales. The interaction of hydrological (transport) factors with P source factors, and their relationship to scale, require further examination. Runoff-generating areas are highly variable, both temporally and spatially. Improvement in the understanding and identification of these areas will contribute to increased effectiveness of strategies aimed at reducing P transfers in runoff. A thorough consideration of scale effects using the conceptual model of P transfer outlined in this paper will facilitate the development of improved strategies for reducing P losses in runoff.

Phosphorus release from phosphate rock and iron phosphate by low-molecular-weight organic acids.

Low-molecular-weight(LMW) organic acids widely exist in soils, particularly in the rhizosphere. A series of batch experiments were carried out to investigate the phosphorus release from rock phosphate and iron phosphate by low-molecular-weight organic acids. Results showed that citric acid had the highest capacity to solubilize P from both rock and iron phosphate. P solubilization from rock phosphate and iron phosphate resulted in net proton consumption. P release from rock phosphate was positively correlated with the pKa values. P release from iron phosphate was positively correlated with Fe-organic acid stability constants except for aromatic acids, but was notcorrelated with pKa. Increase in the concentrations of organic acids enhanced P solubilization from both rock and iron phosphate almost linearly. Addition of phenolic compounds further increased the P release from iron phosphate. Initial solution pH had much more substantial effect on P release from rock phosphate than from iron phosphate.