Towards the Improvement of Soil Salinity Mapping in a Data-Scarce Context Using Sentinel-2 Images in Machine-Learning Models.

Several recent studies have evidenced the relevance of machine-learning for soil salinity mapping using Sentinel-2 reflectance as input data and field soil salinity measurement (i.e., Electrical Conductivity-EC) as the target. As soil EC monitoring is costly and time consuming, most learning databases used for training/validation rely on a limited number of soil samples, which can affect the model consistency. Based on the low soil salinity variation at the Sentinel-2 pixel resolution, this study proposes to increase the learning database's number of observations by assigning the EC value obtained on the sampled pixel to the eight neighboring pixels. The method allowed extending the original learning database made up of 97 field EC measurements (OD) to an enhanced learning database made up of 691 observations (ED). Two classification machine-learning models (i.e., Random Forest-RF and Support Vector Machine-SVM) were trained with both OD and ED to assess the efficiency of the proposed method by comparing the models' outcomes with EC observations not used in the models´ training. The use of ED led to a significant increase in both models' consistency with the overall accuracy of the RF (SVM) model increasing from 0.25 (0.26) when using the OD to 0.77 (0.55) when using ED. This corresponds to an improvement of approximately 208% and 111%, respectively. Besides the improved accuracy reached with the ED database, the results showed that the RF model provided better soil salinity estimations than the SVM model and that feature selection (i.e., Variance Inflation Factor-VIF and/or Genetic Algorithm-GA) increase both models´ reliability, with GA being the most efficient. This study highlights the potential of machine-learning and Sentinel-2 image combination for soil salinity monitoring in a data-scarce context, and shows the importance of both model and features selection for an optimum machine-learning set-up.

Fate of antimony contamination generated by road traffic - A focus on Sb geochemistry and speciation in stormwater ponds.

Although antimony (Sb) contamination has been documented in urban areas, knowledge gaps remain concerning the contributions of the different sources to the Sb urban biogeochemical cycle, including non-exhaust road traffic emissions, urban materials leaching/erosion and waste incineration. Additionally, details are lacking about Sb chemical forms involved in urban soils, sediments and water bodies. Here, with the aim to document the fate of metallic contaminants emitted through non-exhaust traffic emissions in urban aquatic systems, we studied trace element contamination, with a particular focus on Sb geochemistry, in three highway stormwater pond systems, standing as models of surface environments receiving road-water runoff. In all systems, differentiated on the basis of lead isotopic signatures, Sb shows the higher enrichment factor with respect to the geochemical background, up to 130, compared to other traffic-related inorganic contaminants (Co, Cr, Ni, Cu, Zn, Cd, Pb). Measurements of Sb isotopic composition (δ123Sb) performed on solid samples, including air-exposed dusts and underwater sediments, show an average signature of 0.07 ±â€¯0.05‰ (n = 25, all sites), close to the δ123Sb value measured previously in certified reference material of road dust (BCR 723, δ123Sb = 0.03 ±â€¯0.05‰). Moreover, a fractionation of Sb isotopes is observed between solid and dissolved phases in one sample, which might result from Sb (bio)reduction and/or adsorption processes. SEM-EDXS investigations show the presence of discrete submicrometric particles concentrating Sb in all the systems, interpreted as friction residues of Sb-containing brake pads. Sb solid speciation determined by linear combination fitting of X-Ray Absorption Near Edge Structure (XANES) spectra at the Sb K-edge shows an important spatial variability in the ponds, with Sb chemical forms likely driven by local redox conditions: "dry" samples exposed to air exhibited contributions from Sb(V)-O (52% to 100%) and Sb(III)-O (<10% to 48%) species whereas only underwater samples, representative of suboxic/anoxic conditions, showed an additional contribution from Sb(III)-S (41% to 80%) species. Altogether, these results confirm the traffic emission as a specific source of Sb emission in surface environments. The spatial variations of Sb speciation observed along the road-to-pond continuum likely reflect a high geochemical reactivity, which could have important implications on Sb transfer properties in (sub)surface hydrosystems.

Biological attenuation of arsenic and iron in a continuous flow bioreactor treating acid mine drainage (AMD).

Passive water treatments based on biological attenuation can be effective for arsenic-rich acid mine drainage (AMD). However, the key factors driving the biological processes involved in this attenuation are not well-known. Here, the efficiency of arsenic (As) removal was investigated in a bench-scale continuous flow channel bioreactor treating As-rich AMD (∼30-40 mg L-1). In this bioreactor, As removal proceeds via the formation of biogenic precipitates consisting of iron- and arsenic-rich mineral phases encrusting a microbial biofilm. Ferrous iron (Fe(II)) oxidation and iron (Fe) and arsenic removal rates were monitored at two different water heights (4 and 25 mm) and with/without forced aeration. A maximum of 80% As removal was achieved within 500 min at the lowest water height. This operating condition promoted intense Fe(II) microbial oxidation and subsequent precipitation of As-bearing schwertmannite and amorphous ferric arsenate. Higher water height slowed down Fe(II) oxidation, Fe precipitation and As removal, in relation with limited oxygen transfer through the water column. The lower oxygen transfer at higher water height could be partly counteracted by aeration. The presence of an iridescent floating film that developed at the water surface was found to limit oxygen transfer to the water column and delayed Fe(II) oxidation, but did not affect As removal. The bacterial community structure in the biogenic precipitates in the bottom of the bioreactor differed from that of the inlet water and was influenced to some extent by water height and aeration. Although potential for microbial mediated As oxidation was revealed by the detection of aioA genes, removal of Fe and As was mainly attributable to microbial Fe oxidation activity. Increasing the proportion of dissolved As(V) in the inlet water improved As removal and favoured the formation of amorphous ferric arsenate over As-sorbed schwertmannite. This study proved the ability of this bioreactor-system to treat extreme As concentrations and may serve in the design of future in-situ bioremediation system able to treat As-rich AMD.