•A RF-based SMAP soil moisture product downscaling method is proposed.•The method provides a good approach to connect soil moisture with other surface variables.•Field validation shows higher ...accuracy from the downscaled SSM than the SMAP SSM itself.•The method has good potential to get high resolution (1 km) and reliable soil moisture.
The low-resolution characteristic of passive microwave surface soil moisture (SSM) products greatly limits their application in many fields at regional or local scale. Aiming to overcome this limitation, a random forest (RF)-based downscaling approach was proposed in this study to disaggregate the Soil Moisture Active and Passive (SMAP) SSM product with the synergistic use of the Optical/Thermal infrared (TIR) observations from the Moderate-Resolution Imaging Spectro-radiometer (MODIS) onboard the Terra and Aqua satellites. The Iberian Peninsula was selected as the study area during the period from 2015 to 2016.
First, the performance of the RF-based approach in building the SSM relationship model with surface variables (surface temperature, vegetation index, leaf area index, albedo, water index, solar factor, and elevation) was compared with that resulting from a widely used polynomial-based relationship model. Good agreement was achieved for the RF-based method with a correlation coefficient (R) above 0.95 and a mean root-mean-square deviation (RMSD) of 0.009 m3/m3.
Next, four data combinations (AM + Terra, AM + Aqua, PM + Terra, and PM + Aqua) were generated according to the different overpass times of the SMAP and MODIS observations, and they were separately used to derive the spatially downscaled SSM with the proposed RF-based downscaling method. Validation was performed with the in situ measurements from the REMEDHUS network of the University of Salamanca in Spain. The results indicated that all combinations have similar good performances with an unbiased root-mean-square difference (ubRMSD) of 0.022 m3/m3, and the downscaled SSM at 1-km spatial resolution presented better accuracy while showing higher spatial heterogeneity and more detailed temporal pattern.
Finally, the temporal changing pattern of the downscaled SSM was assessed with the precipitation time series from eight meteorological stations in the study area, and the rainfall effect on the variation of SSM was well tracked from its temporal changes.
Overall, this study suggests that the proposed RF-based downscaling method is able to capture the variation of SSM well, and it should be helpful to improve the resolution of passive microwave soil moisture data and facilitate their applications at small scales.
Soil moisture (SM) is an essential component of the environmental and the agricultural system. Continuous monitoring and forecasting of soil moisture is a desirable strategy to understand the soil ...dynamics for proactive planning and decision-making measures for agriculture and related fields. In this study hybrid data-intelligent, extreme learning machine (ELM) models are designed and explored for monthly SM forecasting. The chaotic, complex and dynamical behavior of SM can compound the accuracy of data-driven models. Consequently, two versatile, computationally efficient and self-adaptive multi-resolution utilities namely, complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and the ensemble empirical mode decomposition (EEMD) algorithms are utilized to address these data non-stationarity issues, which if not resolved can lead to model prediction inaccuracies. The difference in these approaches is that, during the EEMD process, a Gaussian white noise is added to the intact (i.e., unresolved) time series only, while, the CEEMDAN requires sequential additions at each decomposition phase. Integration of these multi-resolution tools with the ELM model led to the hybrid CEEMDAN-ELM and the EEMD-ELM models, that were benchmarked with random forest (RF) equivalent models. Using WaterDyn model's hind-simulated SM data, these models were applied (without any climate inputs) to forecast the upper (0.2 m) and the lower layer (0.2–1.5 m depth) soil moisture in Australia's agricultural-hub, the Murray-Darling Basin. The standalone ELM and RF model has similar computation efficiency and model performances. However, despite the implementation of computationally expensive ensemble techniques (i.e., EEMD and CEEMDAN, the hybrid ensembles EEMD-ELM and CEEMDAN-ELM were highly efficient with improved performances. The research outcomes showed that the CEEMDAN-ELM model outperformed the alternative models at three (out of the seven) sites applied for upper layer SM forecasts, while the EEMD-ELM hybrid model was superior at all seven sites for the lower layer soil moisture forecasts. The study signifies the important role of the self-adaptive multi-resolution utility (CEEMDAN) hybridized with the ELM algorithm to potentially develop automated prediction systems for forecasting soil moisture, with potential applications in agriculture.
•Hybrid moisture (SM) data-intelligent model (ELM integrated with EEMDAN) is designed.•Hindcasted SM from WaterDyn hydrological are used to construct ELM-EEMDAN model.•Hybrid-Random Forest (RF) and standalone RF/ELM are used as benchmarks.•Evaluation is performed at six sites in drought-prone, Australia's Murray Darling basin.•Hybrid ELM-EEMDAN outperforms standalone ELM and RF-EEMDAN/RF models.
Carbon dioxide (CO2) emissions is a major greenhouse gas that causes global warming. Many researchers in the fields of architecture, engineering, and construction try to measured CO2 emissions during ...a building's lifecycle. However, research on the CO2 emissions during construction stage are less studied than those during other stages because they are considered to be lower than the emissions from the building's materials' production or operational stage. In addition, research has been hindered by a complicated calculation process and a lack of data, and thus few methods are available for forecasting construction-stage carbon emissions, especially at the early design stage. In order to estimate the environmental effects of the emissions from the vast number of construction activities, this study applies a random forest (RF) based predictive method to predict construction-stage carbon emissions. The RF-based model uses data from 38 buildings in the Pearl River Delta region of China for the initial training set to find the relation between construction-stage carbon emissions and design parameters. Compared with the multilinear regression method, the RF-based model has a higher coefficient of determination and lower mean square error. The model developed in this study facilitates the prediction of construction-stage carbon emissions at the early design stage of a building. This opens up novel opportunities to reduce carbon emissions from buildings, which had previously been possible only at the latter stages of a building's life cycle. It will also help policymakers account for the probable distribution and amount of CO2 emissions in a city when multiple construction projects are proceeding simultaneously, so that measures can be implemented to avoid excessive emissions.
•A random forest-based model is developed for the prediction of construction-stage carbon emissions at the early design stage.•The relationships between design parameters and predicted construction-stage carbon emissions are quantitatively determined.•The building's foundation area, underground area, and height are found to have the greatest effect on its construction-stage carbon emissions.•The random forest-based model facilitates the prediction of construction-stage carbon emissions at the early design stage of a building project.
Hypertensive disorders of pregnancy are one of the leading causes of maternal morbidity and mortality worldwide. Management of these conditions can pose many clinical dilemmas and can be particularly ...challenging during the immediate postpartum period. Models for predicting and managing postpartum hypertension are necessary to help address this clinical challenge.
This study aimed to evaluate predictive models of blood pressure spikes in the postpartum period and to investigate clinical management strategies to optimize care.
This was a retrospective cohort study of postpartum women who participated in remote blood pressure monitoring. A postpartum blood pressure spike was defined as a blood pressure measurement of ≥140/90 mm Hg while on an antihypertensive medication and a blood pressure measurement of ≥150/100 mm Hg if not on an antihypertensive medication. We identified 3 risk level patient clusters (low, medium, and high) when predicting patient risk for a blood pressure spike on postpartum days 3 to 7. The variables used in defining these clusters were peak systolic blood pressure before discharge, body mass index, patient systolic blood pressure per trimester, heart rate, gestational age, maternal age, chronic hypertension, and gestational hypertension. For each risk cluster, we focused on 2 treatments, namely (1) postpartum length of stay (<3 days or ≥3 days) and (2) discharge with or without blood pressure medications. We evaluated the effectiveness of the treatments in different subgroups of patients by estimating the conditional average treatment effect values in each cluster using a causal forest. Moreover, for all patients, we considered discharge with medication policies depending on different discharge blood pressure thresholds. We used a doubly robust policy evaluation method to compare the effectiveness of the policies.
A total of 413 patients were included, and among those, 267 (64.6%) had a postpartum blood pressure spike. The treatments for patients at medium and high risk were considered beneficial. The 95% confidence intervals for constant marginal average treatment effect for antihypertensive use at discharge were −3.482 to 4.840 and − 5.539 to 4.315, respectively; and for a longer stay they were −5.544 to 3.866 and −7.200 to 4.302, respectively. For patients at low risk, the treatments were not critical in preventing a blood pressure spike with 95% confidence intervals for constant marginal average treatment effect of 1.074 to 15.784 and −2.913 to 9.021 for the different treatments. We considered the option to discharge patients with antihypertensive use at different blood pressure thresholds, namely (1) ≥130 mm Hg and/or ≥80 mm Hg, (2) ≥140 mm Hg and/or ≥90 mm Hg, (3) ≥150 mm Hg and/or ≥ 100 mm Hg, or (4) ≥160 mm Hg and/or ≥ 110 mm Hg. We found that policy (2) was the best option with P<.05.
We identified 3 possible strategies to prevent outpatient blood pressure spikes during the postpartum period, namely (1) medium- and high-risk patients should be considered for a longer postpartum hospital stay or should participate in daily home monitoring, (2) medium- and high-risk patients should be prescribed antihypertensives at discharge, and (3) antihypertensive treatment should be prescribed if patients are discharged with a blood pressure of ≥140/90 mm Hg.
The wear behavior of AZ91 alloy was investigated by considering different parameters, such as load (10−50 N), sliding speed (160−220 mm/s) and sliding distance (250−1000 m). It was found that wear ...volume loss increased as load increased for all sliding distances and some sliding speeds. For sliding speed of 220 mm/s and sliding distance of 1000 m, the wear volume losses under loads of 10, 20, 30, 40 and 50 N were calculated to be 15.0, 19.0, 24.3, 33.9 and 37.4 mm3 respectively. Worn surfaces show that abrasion and oxidation were present at a load of 10 N, which changes into delamination at a load of 50 N. ANOVA results show that the contributions of load, sliding distance and sliding speed were 12.99%, 83.04% and 3.97%, respectively. The artificial neural networks (ANN), support vector regressor (SVR) and random forest (RF) methods were applied for the prediction of wear volume loss of AZ91 alloy. The correlation coefficient (R2) values of SVR, RF and ANN for the test were 0.9245, 0.9800 and 0.9845, respectively. Thus, the ANN model has promising results for the prediction of wear performance of AZ91 alloy.
Knowledge of long-term changes in vegetation cover is essential for paleoenvironmental reconstruction and Earth system modeling. The vegetation on the northeastern Qinghai-Tibet Plateau (QTP) is ...highly sensitive to climatic changes, but studies of long-term changes in the vegetation cover of this region are lacking. To better understand the changes in the regional vegetation cover since late Marine Isotope Stage (MIS) 3, we obtained a pollen record from Luanhaizi Lake and used it to quantitatively reconstruct changes in the cover of trees and grasses using the random forest method. The pollen spectra show that the vegetation of the Luanhaizi Lake area was probably alpine tundra from late MIS3 to the Last Glacial Maximum, after which it changed to alpine steppe with alpine shrub and sparse forest at lower elevations during the last deglaciation. Alpine steppe and alpine meadow dominated the vegetation during the Holocene, with sparse forest in the surrounding low-elevation areas. The quantitative vegetation cover reconstruction suggests that the vegetation cover of the Luanhaizi Lake area was low (20–30%) during 47.0–20.0 ka. The tree cover then increased from ∼3% to ∼10%, and the grass cover increased from ∼20% to ∼45% during the last deglaciation (20.0–11.9 ka). The increasing Northern Hemisphere summer insolation caused increases in the regional temperature, meltwater supply, and monsoon precipitation, which promoted the development of steppe vegetation. The continued increases in insolation and monsoonal precipitation during the Holocene further increased the tree cover, which reached a maximum of 17.4% at ∼7.2 ka. Vegetation cover reconstructions from two other sites on the eastern QTP for the last deglaciation indicate contrasting patterns of changes, likely due to contrasts in elevation, climate, and environment.
•Vegetation cover reconstruction for the northeastern QTP since ∼47 ka.•The vegetation cover increased since ∼20 ka on the northeastern QTP.•Grass cover and tree cover peaked during different intervals.•Spatial differences existed in the evolution of vegetation cover on the eastern QTP since ∼20 ka.
Multi-attribute decision-making (MADM) is the most commonly used methodology for green supplier evaluation and performance improvement. Previous MADM models have mainly relied upon the knowledge and ...opinions of domain-experts as the starting point for making decisions. However, the results are affected by the subjectivity of these judgments and knowledge limitations. This study develops a data-driven MADM model that utilizes potential rules/patterns derived from a large amount of historical data to help decision-makers objectively select suitable green suppliers and provide systemic improvement strategies to help reach the aspiration level. First, the random forest (RF) algorithm is applied to explore the pairwise influential strength relations among attributes derived from real audit data. The influence matrix derived using the RF algorithm is used as input for decision-making trial and evaluation laboratory (DEMATEL)-based analytical network process analysis which is carried out to obtain the influential strength weights of the attributes. Then, multi-objective optimization on the basis of ratio analysis to the aspiration level (MOORA-AS) is utilized to evaluate the gap between the current and aspiration levels for each green supplier. The developed critical influence strength route (CISR) can help managers derive various strategies for improving green supplier performance. The functioning of the proposed model is illustrated using data obtained from the green supplier management department of a Taiwanese electronics company. The results reveal that the proposed model can effectively help decision-makers to solve the problem of green supplier selection and devise strategies for improvement.
•A new data-driven method is developed for the big data era.•The model combines data mining with MADM for green supplier problems.•The critical influence route can provide a systematic method for improvement.•The model eliminates the shortcomings of depending upon expert opinions for the input data.
The trend of secondary soil salinization has been increasingly concerned in the arid region under drip irrigation in recent decades in China. It remains unclear whether the potential increasing ...nonuniformity of irrigation water aroused by the increased acreage of subunit leads to salt harmfulness and yield reduction. Cotton field experiments were conducted in 2018 and 2019 to evaluate the effects of lateral length on the distribution of water and salt in soil as well as lint yield under mulched drip irrigation to identify the dominating factors using the random forest regression. Three lateral lengths of 40-, 80-, and 120-m were used. Along with a trigger point of 60–70% of field capacity (FC), irrigation amount was determined using three irrigation upper limits of 90%, 100%, and 110% of FC in the 2018 and of 90%, 110%, and 130% of FC in the 2019. The random forest analysis revealed that the nonuniform water applied by drip irrigation greatly affected the distribution of soil water content with an affecting weight exceeding 0.4. However, it imposed little impact on soil salt content along the lateral with the affecting weights less than 0.1. The relative importance of spatial variation of soil properties on soil water content decreased with increasing lateral length, while the nonuniformly applied water became the dominant factor affecting the distribution of soil water content for a field with lateral length of 120-m. The initial soil salt content dominated the spatial and temporal variation of soil salt content for treatments with all the three lengths of lateral. The long lateral did not produce obvious soil salt accumulation at the 0–60 cm soil depth along the lateral. The cumulative irrigation water and nitrogen applied imposed greater influence on the cotton lint yield than other factors. A medium lateral length is a promising selection to maintain sustainable production balancing possible salt accumulation and crop production in arid lands with relatively low initial soil salt content.
•Nonuniform water applied greatly affected soil water content with 120-m laterals.•Initial salt content dominated spatial and temporal variation of soil salt content.•120-m lateral did not produce obvious soil salt accumulation in the 0–60 cm profile.•80-m drip lateral could be used to balance salt accumulation and crop yield.