Search

Accurate estimation of precipitation at local to global scales can considerably enhance our understanding of climate system dynamics. While numerous precipitation products are available as indispensable tools for investigating precipitation and its associated processes, none can consistently provide the lowest estimation error across environmental conditions. The multiple source precipitation ensemble (MSPE) methods have been considered a vital solution. A new MSPE framework is proposed here, which simultaneously uses machine learning (ML) classification and regression techniques within an automatic workflow (MSPE aml). Six precipitation products and their ensembles based on different MSPE strategies were evaluated at 2365 gauged and 800 randomly selected ungauged sites over China. Results revealed significant precision inconsistencies among the products primarily due to their different data sources and retrieval algorithms; while MSPE aml can effectively reduce the random and classification errors of estimated precipitation according to the Kling-Gupta efficiency and Heidke skill score. The improvements demonstrated the unique features of MSPE aml , particularly the necessity of the joint use of ML classifiers and regressors and assigning spatiotemporal dynamic weights for merging precipitation data. Moreover, MSPE aml can substantially improve its generalizability through a simple binning procedure, making it applicable under more complex conditions. The varying contributions of predictor variables (indicated by Shapely values) in different ML models identified the complexity of the MSPE issue and further the importance of designing proper ML models according to specific targets. The proposed MSPE framework is expected to be a suitable solution for assembling multiple precipitation data sources with different time periods and scales. • A machine learning-based framework for merging multi-source daily precipitation datasets is proposed. • It is effective in producing high-precision precipitation estimates in both gauged and ungauged regions. • The new method provides enhanced generalizability through a simple binning procedure. • It is necessary to combine machine learning classification and regression for an ensemble. • Spatiotemporal dynamically weighted ensembles based on machine learning classification have great potential. [ABSTRACT FROM AUTHOR]

• Physicochemical soil properties nonlinearly explain the history of LUC in China. • An explainable machine-learning method provides both local and global explanations. • A new perspective on LUC change with reliance on soil properties as drivers. • Toward knowledge gap-filling regarding soil properties in explaining LUC history. Enhancing our comprehension of soil processes and their impact on Earth requires precise quantification of human-induced soil alterations, particularly those related to land use/cover (LUC) histories. Thoroughly validated LUC and soil maps specific to China, and an explainable machine-learning approach were applied to reveal how soil physicochemical properties, independently or in combination, explain LUC histories in the last sixty years. Validation using historical data showed that the proposed machine-learning models can adequately simulate the pattern of LUC histories. Nonlinear mappings of various soil properties in explaining LUC histories were demonstrated and critical thresholds for changes in the explanatory capacities of each soil property to specific LUC histories were identified. Specifically, physical soil properties, except for soil pH, particularly soil thickness, clay content, bulk density, and coarse fragments, play significant roles in explaining the historical trajectories of forests, croplands, and pasture/rangelands than chemical soil properties. Predicted LUC changes suggest more intensive dynamics that are characterized by expansions in forest and pasture/rangeland areas, coupled with a reduction in cropland areas. Overall, improving understanding of the bidirectional links between soil and LUC changes is a crucial and imperative step towards refining the representation of soils in Earth system models. [ABSTRACT FROM AUTHOR]

Sorption of pyrene and phenanthrene to model (illite and charcoal) and natural (Yangtze sediment) geosorbents were investigated by batch techniques using fluorescence spectroscopy. A higher adsorption of phenanthrene was observed with all sorbents, which is related to the better accessibility of smaller molecules to micropores in the molecular sieve sorbents. In addition, pyrene sorption in binary-solute systems with a constant initial concentration of phenanthrene (0.1 μmol L-1 or 2 μmol L-1) was studied. A 0.1 μmol L-1 concentration of phenanthrene causes no competitive effect on the pyrene sorption. A 2 μmol L-1 concentration of phenanthrene significantly suppresses the sorption of pyrene, especially in the low concentration range; nonlinearity of the pyrene sorption isotherms thus decreases. The competitive effect of 2 μmol L-1 phenanthrene on the pyrene sorption is overestimated by the ideal adsorbed solution theory (IAST) using the fitted single sorption results of both solutes. An adjustment of the IAST application by taking into account the molecular sieve effect is proposed, which notably improves the IAST prediction for the competitive effect. [ABSTRACT FROM AUTHOR]

Abstract: The effect of organic carbon (OC) and mineral surface on the sorption of polycyclic aromatic hydrocarbon (PAH) pyrene molecule to four Yangtze River sediments was investigated by sorption batch techniques using fluorescence spectroscopy. Pyrene sorption to the mineral fraction was estimated with model sorbent illite, the main clay mineral in Yangtze sediment. The Freundlich model fitted sorption to illite and to sediments was normalized to the specific surface area (SSA). Comparison of the SSA-normalized sorption capacities of illite and sediments suggests a negligible contribution of the pyrene sorption to the mineral fraction. In addition, composite models, such as the linear Langmuir model (LLM) and the linear Polanyi–Dubinin–Manes model (LPDMM) were applied for fitting the sorption of pyrene to the pristine sediments. The application of composite models allows assessing the partition of pyrene into amorphous organic carbon (AOC) and the adsorption in the porous structure of black carbon (BC). The modelling results indicate that the pyrene adsorption to the minor BC components (<0.2%) is more effective than the partition to AOC (0.5–1.3%). Besides the pristine sediments, sediments preheated at 375°C were also studied, in which the AOC fraction was removed during the preheating treatment. The modelling results with LPDMM and Polanyi–Dubinin–Manes model (PDMM) indicate a similar adsorption capacity of BC in pristine and preheated sediments, respectively. The low AOC concentrations in sediments do not diminish the BC micropore filling with pyrene. Simulation of pyrene distribution in the investigated Yangtze River sediments support the importance of the BC fraction in the PAH immobilization under environmental conditions. [Copyright &y& Elsevier]