Self-organizing Map Method Research Articles

Exploring the interrelations and driving factors among typical ecosystem services in the Yangtze river economic Belt, China

Exploring the spatiotemporal characteristics of ecosystem services (ESs) and their drivers is crucial for managers to develop significant scientific policies that further sustainable development. We used the Yangtze River Economic Belt (YREB) to explore the trends, hotspots, and drivers of water yield (WY), soil conservation (SC), carbon sequestration (CS), and food supply (FS) between 2000 and 2020. Similarly, we analyzed relationships among ESs and drivers of the multiple ecosystem services landscape index (MESLI). We used the self-organizing map method to obtain the types and distribution of the ES bundles, revealing the bundles, trade-offs, and synergies among ESs. The four ESs had an increasing trend, with CS having the highest increase; ES hotspot analysis showed differences among upper, middle, and lower reaches. Constraint lines among ESs and drivers were diverse; the corresponding SC and WY reached thresholds when CS values were 1477.81 and 460.5 t km−2, respectively. When FS values were 67.34 and 86.17 × 104 Yuan·km−2, CS and WY reached their thresholds. All critical drivers of the four ESs were natural factors. The thresholds that the MESLI reached with driver status were 1000 mm (evapotranspiration), 2121 mm (precipitation), 2.42° (slope), 1.46% (soil organic matter), 36.08% (sand), 30.75% (proportion of non-agricultural population), 18.57% (cropland proportion), 1.05 × 104 persons·km−2 (population density), and 84.84% (proportion of non-agricultural industries in total gross domestic product), respectively. FS, water supply, and ecological conservation bundles changed over the 20 years, and trade-offs and synergies among ESs within bundles differed. We revealed the complexity of ESs from multiple perspectives, which will enable the development of ecosystem management and conservation recommendations for the YREB and large-scale economic zones worldwide.

Combination of multivariate data analysis and mixing modelling to assess tracer potential of contaminants of emerging concern in aquifers

Collected evidence has shown that contaminants of emerging concern (CECs) in conjunction with more conventional tracers (major ions, nutrients, isotopes etc.) can be used to trace pollution origin in aquatic systems. However, in highly mixed aquifer systems signals obtained from conventional tracers overlap diminishing their potential to be used as tracers. In this study, we present an approach that incorporates multivariate statistical analysis (principal component analysis (PCA) and Kohonen's Self-Organizing Map method (SOM)) and mixing modelling to identify the most suitable CECs to be employed as anthropogenic tracers. The study area is located in the Besòs River Delta (Barcelona, NE Spain) and represents the highly mixed aquifer system. A one-year monthly based monitoring campaign was performed to collect the information about the concentrations of 105 CECs as well as major and minor ions in the river and along the groundwater flow. The dimensionality of the obtained dataset was reduced to 25 CECs, based on their estimated health risk effects, for multivariate data analysis. The obtained results showed the overlap of conventional tracers' signals obtained from PCA. In case of CECs, PCA revealed differences in their distributions allowing the differentiation of the roles of natural attenuation processes, local and regional flows on their occurrence in different parts of the aquifer. This was not possible to do using solely CECs' distribution profiles. SOMs provided the lacking information about the modality of the distribution of each CECs, revealing their ability to represent factors controlling the groundwater hydrochemistry, which assist in defining their tracer potential. Based on the obtained results four identified persistent CECs, two with unimodal (lamotrigine and 5-Desamino-5-oxo-lamotrigine) and two with bimodal (carbamazepine and diazepam (higher modality was not revealed)) distributions, were selected to run a mixing model to compare their tracer performance.

UJI HOMOGENITAS DAN HETEROGENITAS CLUSTER DATA INTERKOMPARASI ANAK TIMBANGAN

Intercomparison of mass quantities is an activity to compare performance between laboratories ofmass quantities and reference laboratories which produces intercomparative data in the form ofconventional mass values and uncertainty values. Clustering of weights was carried out to obtain ananalysis of the laboratory’s participants management. The analysis carried out is very dependent on theaccuracy of the cluster. Clustering using self-organizing maps algorithm, and the results are tested for thelevel of homogeneity within the cluster and the level of heterogeneity between clusters using the icdrate value.The clustering of intercomparison’s data using the self-organizing maps method resulted in an icdratevalue of 0,0713 for an artifact of 1000 g and an icdrate of 0,2889 for an artifact of 200g

Identification of Fresh and Unfresh Fish Based on Eye Image Using The Self-Organizing Maps (SOM) Method

Fish is a common protein source and easy to obtain in Indonesia, but because of its high water content, fish quickly spoils. Fish freshness can be detected using several conventional methods, such as chemical analysis, biochemistry, microbiological analysis, and sensory examination. Another identification method involves observing the color of the fish's eyes. Fish identification is crucial before any further processing, ensuring that the fish's quality delivered to consumers remains high. To tackle the problem of differentiating between fresh and non-fresh fish, this research employs Self-Organizing Maps (SOM) as the primary methodology. This research focuses on identifying fresh and non-fresh fish using the SOM method, using actual data involving tilapia as the research object. The data includes eye images of new and non-fresh fish, and various procedures are required to obtain the desired data. This data is then used as training and testing data. The process continues with the preprocessing stage, which is a data modification process to improve performance in subsequent steps and feature extraction using HSV color histograms. Classification of processed data is carried out using the SOM method. Once completed, the identification results are displayed. This research produces a system for identifying fresh and non-fresh fish based on eye images using SOM, which achieves a good accuracy of 85.71%.

Revealing the Environmental Characteristics of Towns in the Middle Himalayas Using a Geographic Information System and Self-Organizing Map

Through long-term interactions with the natural environment, the ethnic groups in the mid-Himalayas have formed unique urban environmental characteristics. Effectively identifying urban environmental characteristics is a prerequisite for implementing sustainable urban management strategies. This study took 194 towns in the mid-Himalayan as the research objects. GIS was used to statistically analyze the terrain, climate, soil, and other environmental characteristics of the towns. The SOM (Self-organizing map) method was used to classify the comprehensive environmental characteristics of the towns. The results show that the main urban environmental characteristics in this area are low-altitude towns account for a large proportion, gentle-slope towns account for a small proportion, rainfall is mainly 125–265 mm, vegetation cover is dominated by high-coverage towns, mainly distributed in central and southern parts, the soil is dominated by embryonic soil and alluvial soil. The SOM method overcomes the subjectivity and low degree of automation in traditional research on urban environmental characteristics using threshold indicator methods or feature interpretation methods. Based on environmental characteristics, the towns were divided into six categories, and the classification results showed a distinct north–south zonal distribution pattern. There were significant differences in the environmental characteristics of towns in different clusters, such as the towns in cluster 5 had high altitudes, low rainfall, and low vegetation coverage, while the towns in cluster 2 had low altitudes, high rainfall, and high vegetation coverage. Finally, based on the SOM clustering results, governance strategies were proposed for towns in different clusters to cope with climate and environmental changes and promote sustainable development in the mid-Himalayan.

Subseasonal Features of the Indian Monsoon

Abstract The Indian monsoon is of utmost concern to agriculture, the economy, and the livelihoods of billions in South Asia. However, little attention has been paid to the possibility of distinct subseasonal episodes phase-locked in the Indian monsoon annual cycle. This study addresses this gap by utilizing the self-organizing map (SOM) method to objectively classify six distinct subseasonal stages based on the 850-hPa wind fields. Each subseasonal stage ranges from 23 to 90 days. The Indian summer monsoon (ISM) consists of three substages, the ISM-onset, ISM-peak, and ISM-withdrawal, altogether contributing to 82% of the annual precipitation. The three substages signify the rapid northward advance, dominance, and gradual southward retreat of southwesterlies from mid-May to early October. The winter monsoon also comprises three substages (fall, winter, and spring), distinguishable by the latitude of the Arabian Sea high pressure ridge and hydrological conditions. This study proposes two compact indices based on zonal winds in the northern and southern Arabian Sea to measure the winter and summer monsoons, respectively. These indices capture the development and turnabouts of the six SOM-derived stages and can be used for subseasonal monsoon monitoring and forecasts. The spring and the ISM-onset episodes are highly susceptible to compound hazards of droughts and heatwaves, while the greatest flood risk occurs during the ISM-peak stage. The fall stage heralds the peak season for tropical storms over the Arabian Sea and the Bay of Bengal. The annual start and end dates of the ISM-peak are highly correlated (0.6–0.8) with the criteria-based dates proposed previously, supporting the delineation of the Indian monsoon subseasonal features. Significance Statement This research explores the existence of subseasonal features in the Indian monsoon annual cycle. Through the use of machine learning, we discover that the Indian summer monsoon and winter monsoon each consist of three substages. These substages’ evolution can be measured by two compact indices proposed herein, which can aid in subseasonal monsoon monitoring and forecasts in South Asia. Pertaining to hazard adaptations, this work pinpoints the subseasonal episodes most susceptible to droughts, heatwaves, floods, and tropical storms. High correlations are obtained when validating the substages’ yearly start and end dates against those documented in the existing literature, offering credibility to the subseasonal features of the Indian monsoon.

Atmospheric Latent Energy Transport Pathways into the Arctic and Their Connections to Sea Ice Loss during Winter over the Observational Period

Abstract To investigate patterns of horizontal atmospheric latent energy (LE) transport toward the Arctic, we applied the self-organizing maps (SOM) method to the daily vertically integrated horizontal LE flux from ERA5 in winter (January–March) during 1979–2021. A clear picture depicting the LE transport to the Arctic at a synoptic scale then emerged, with four primary transport pathways identified: the northern Europe, the Davis Strait, the Greenland Sea, and the Bering Strait pathways. The four primary pathways occurred at a comparable frequency, and noticeable interannual variability was observed in their time series of frequency during 1979–2021. Further analysis suggested that the northward LE transport through all these pathways is significantly modulated by cyclones, with the northern Europe and the Greenland Sea pathways being mostly affected. Generally, more frequent and stronger cyclones were observed near the entry regions of LE transport compared to other regions. Moreover, this study provides a comprehensive picture of how atmospheric LE transport is related to air temperature, moisture, surface heat flux, and sea ice anomalies over the Arctic Ocean in winter. Through a thermodynamic perspective, we argue that the deleterious impacts of poleward LE transport on Arctic sea ice are to a large extent attributable to the enhanced local atmosphere–ice interactions, which increase downward longwave radiation (DLR) plus turbulent fluxes, consequently warming the surface and promoting the loss of sea ice. According to the quantitative results, among the four primary pathways, LE transport through the Davis Strait and the Greenland Sea could cause the loss of Arctic sea ice most efficiently.

Using Self-Organizing Map and Multivariate Statistical Methods for Groundwater Quality Assessment in the Urban Area of Linyi City, China

Groundwater holds an important role in the water supply in Linyi city, China. Investigating the hydrochemical characteristics of groundwater, and revealing the factors governing groundwater geochemistry, is a primary step for ensuring the safe and rational exploitation of groundwater resources. This study used a self-organizing map (SOM) and multivariate statistical methods to assess groundwater quality in the urban area of Linyi city. Based on the hydrochemical dataset consisting of nine parameters (i.e., pH, Ca2+, Mg2+, Na+, K+, HCO3−, Cl−, SO42−, and NO3−) from 89 groundwater samples, the SOM was first applied to obtain the weight vectors of the output nodes. Hierarchical cluster analysis (HCA) was used for organizing the nodes into four clusters. The node cluster indices were then remapped to the groundwater samples according to the winner node for each sample. The hydrochemical characteristics and factors controlling the groundwater geochemistry of the four clusters were analyzed using principal component analysis (PCA) and graphical methods including Piper and Gibbs diagrams, as well as binary plots of the major ions in groundwater. Results indicated that groundwater geochemistry in this area is primarily governed by water–rock interactions, such as the dissolution of halite, calcite, and gypsum, along with the influence of municipal sewage and the degradation of organic matter. This study demonstrates that the integration of an SOM and multivariate statistical methods improves the understanding of groundwater geochemistry and hydrochemical evolution in complex groundwater flow systems impacted by utilization.

Multidisciplinary robust optimization approach of fan rotors under structural constraints with blade curvature

The robust optimization approach is an important method to solve the problem of fans/compressors performance disparities caused by geometric tolerances. Nonetheless, since the optimum design of turbomachinery blades is multidisciplinary in nature, traditional robust optimization approaches that only consider aerodynamic performance may result in the blade strength exceeding their material limit. Therefore, it is necessary to investigate the multidisciplinary robust optimization approach (MROA) considering both aerodynamic and structural performance. For MROA, there is a big challenge of huge computation cost and runtime for the time-consuming high-fidelity CFD and FEM. In this paper, a new method to reduce the computation cost and runtime of MROA is proposed by using the blade curvature to approximate the evaluation of blade stress. Firstly, data mining by self-organizing mapping method is used to analyze and extract the constraint value of the blade curvature. And the blade curvature constraint value is used as a penalty function instead of the time-consuming high-fidelity FEM, which greatly reduces the computational cost and runtime of the multidisciplinary optimization. Then, Polynomial chaos Kriging (PCK) is then used as a surrogate model for uncertainty quantification (UQ) of aerodynamic performances, and one MROA based on the PCK model-based UQ method and the curvature constraints is developed. The proposed method is verified using a fan rotor as a case study. The results show that the maximum stress of the traditional optimization method was 453 MPa, which exceeded the yield limit of aluminum alloy (420 MPa). In contrast, the maximum stress of the method proposed in this paper was 343 MPa. In terms of aerodynamic performance, compared with the baseline, the mean efficiency of the traditional and the proposed method increased by 2.6% and 2.1%, with a variance reduced by 45.5% and 48.5%, respectively. Therefore, compared with the traditional robust optimization method, the proposed method reduces the maximum stress of the blade by 24.3% with almost the same improvement for aerodynamic performance means and disparities.

Analysis Of Unemployment Clusters In Indonesia Using The Self Organizing MAP Method

Unemployment is a situation where someone is not working or is trying to find a job but unable to find work. The spread of unemployment in Indonesia has different characteristics in each region, so it is necessary to classify the unemployed so that each government policy program can be carried out in a more focused and directed manner. This study discusses cluster analysis of unemployment using the Self Organizing Map (SOM) method in classifying the unemployed in Indonesia in 2020. The SOM method is able to show dominant patterns and variables in clusters. The variables used in this study consisted of school enrollment rates, average length of schooling, labor force participation rates, and the percentage of the population using computers. The results of this study formed 3 unemployment rate clusters with cluster 1 being a low unemployment group consisting of 144 districts/cities, cluster 2 with a medium level consisting of 287 districts/cities, and cluster 3 with a high level consisting of 83 districts/cities. The grouping using the SOM method on district/city unemployment data in Indonesia is good because it has a minimum standard deviation ratio of 0.529.

Application of the Self-Organizing Map Method in February Temperature and Precipitation Pattern over China: Comparison between 2021 and 2022

In this study, we compared two anomalous wet February periods in 2021 and 2022 in China. The same anomalies appeared in the spatial distribution of precipitation, with anomalous precipitation centered over the southeast coast. However, temperature discrepancies appeared in most of China, with anomalously high temperatures in 2021 and lower temperatures in 2022. Both instances of increased precipitation were attributed to warm and moist advection from the south, with transport in 2021 being partly enhanced by the South China Sea cyclone, whereas transport in 2022 was mainly due to the subtropical western North Pacific anticyclone. Therefore, in this study, we aimed to compare and analyze temperature and precipitation anomalies in February 2021 and 2022 using the self-organizing map method. Warm events in East Asia and cold events in Siberia and the Tibetan Plateau types were obtained by mode 1, which contained 2021. Mode 6 exhibited opposite warm types in Siberia and cold types in southern Asia, including February temperature and precipitation anomalies in 2022. Based on the results of this study, we can conclude that precipitation anomalies in February 2021 and 2022 occurred under different temperature and circulation anomalies, and both were influenced by La Niña events. Autumn sea ice loss in the Barents Sea contributed significantly to warm and rainy events in February 2021. However, the cold and rainy events of February 2022 were closely related to the strengthening of the Siberian High.

Stochastic regional-based profit-maximizing hub location problem: A sustainable overview

Climate change is one of the most important concerns globally, and countries need to improve their transportation infrastructure to handle flows with the least air pollution. In specific cities and regions, carbon emissions may cause in future many issues for people and societies. Although transportation hubs are very helpful in the reduction of transportation costs, they may cause an increase in carbon emissions in some regions, leading to long-term environmental problems. In this study, a model is developed for the profit-maximizing hub location problem that incorporates three pillars of sustainability. The proposed model is not only designed to maximize profit but also to control carbon emissions using a carbon cap policy and to reduce differences in emissions between regions with consideration of population density. This scheme will lead to achieving a sustainable transportation network. A two-stage stochastic programming approach is employed to cope with demand uncertainty. An enhanced sampling based on the self-organizing map method was utilized to cluster scenarios that lead to dealing with small-sized problems. Furthermore, classical Benders decomposition, Pareto-optimal cut Benders decomposition, and L-shaped algorithms are employed to solve the proposed models more efficiently. The proposed models are analyzed using the well-known Turkish Network (TR) data set. Computational results demonstrate that the proposed models help to achieve sustainability. Factors of sustainability such as environmental and social can be achieved by slightly reducing the amount of profit in the economic factor. Furthermore, results show that the L-shaped algorithm with a multi-cut scheme outperforms the commercial solver, classical Benders decomposition, and Pareto-optimal cut Benders decomposition algorithms in large-size instances.

Long-Term Changes, Synoptic Behaviors, and Future Projections of Large-Scale Anomalous Precipitation Events in China Detected by a Deep Learning Autoencoder

Abstract The frequency of large-scale anomalous precipitation events over China has increased during 1961–2018. However, it remains challenging to understand the mechanisms associated with these anomalous events showing different spatial patterns. Here, we applied an autoencoder technique to identify large-scale anomalous precipitation events for both observations and model simulations, which were then classified into several patterns through a self-organizing map method. The synoptic behavior and atmospheric circulation background of different anomalous patterns were also analyzed using simultaneous composite analyses. Results show that occurrences of different anomalous precipitation patterns have increased significantly, except those centered in North China, Northeast China, and the Yangtze River basin. The anomalous precipitation patterns manifest various intraseasonal distributions, which are linked to zonal oscillations of the western North Pacific subtropical high (WNPSH) and meridional displacements of East Asia westerly jet (EAJ). Accompanied by the westward movement of WNPSH, anticyclonic systems transport warm moist air from the Indian Ocean and the South China Sea to converge with the cold air caused by anomalous cyclones over the northwest flank of the WNPSH, leading to large-scale anomalous precipitation in these regions. Besides WNPSH, the northward and southward displacements of EAJ also favor the occurrence of anomalous precipitation events in northern and southern China, respectively. Our study also illustrates that the occurrence frequency of anomalous precipitation events is projected to increase remarkably under the Shared Socioeconomic Pathway 5–8.5 (SSP585) scenario by rates of twofold to fourfold.

Using machine learning to predict processes and morphometric features of watershed

The research aims to classify alluvial fans’ morphometric properties using the SOM algorithm. It also determines the relationship between morphometric characteristics and erosion rate and lithology using the GMDH algorithm. For this purpose, alluvial fans of 4 watersheds in Iran are extracted semi-automatically using GIS and digital elevation model (DEM) analysis. The relationships between 25 morphometric features of these watersheds, the amount of erosion, and formation material are investigated using the self-organizing map (SOM) method. Principal component analysis (PCA), Greedy, Best first, Genetic search, Random search as feature selection algorithms are used to select the most important parameters affecting erosion and formation material. The group method of data handling (GMDH) algorithm is employed to predict erosion and formation material based on morphometries. The results indicated that the semi-automatic method in GIS could detect alluvial fans. The SOM algorithm determined that the morphometric factors affecting the formation material were fan length, minimum height of fan, and minimum fan slope. The main factors affecting erosion were fan area (Af) and minimum fan height (Hmin-f). The feature selection algorithm identified (Hmin-f), maximum fan height (Hmax-f), minimum fan slope, and fan length (Lf) to be the morphometries most important for determining formation material, and basin area, fan area, (Hmax-f) and compactness coefficient (Cirb) were the most important characteristics for determining erosion rates. The GMDH algorithm predicted the fan formation materials and rates of erosion with high accuracy (R2 = 0.94, R2 = 0.87).

Fractional gradient descent algorithm for switching models using self-organizing maps: One set data or all the collected data

This paper proposes two fractional gradient descent algorithms for switching models. Each submodel is assigned a weight which can determine the identity of the submodel in each sampling instant. By using the fractional gradient descent algorithms, the parameters of each submodel can be obtained, and then the weights of all the submodels can be estimated based on the self-organizing maps method. These two algorithms can deal with different kinds of switching models on a case by case basis. In addition, compared with the traditional identification algorithms, the proposed methods have two advantages: (1) has faster convergence rates; (2) has less computational efforts. Simulation example demonstrates the effectiveness of the proposed methods.

Analysis of Ecosystem Service Contribution and Identification of Trade-Off/Synergy Relationship for Ecosystem Regulation in the Dabie Mountains of Western Anhui Province, China

The study of tradeoffs/synergies among ecosystem services (ESs) is highly significant for land-use planning and regional ecosystem optimization. Land-use change and topographic factors have important implications for ESs. Strengthening the comparative analysis of the capacity of ESs provided by different land-use types in specific regions, studying the topographic gradient effects of ecosystem service trade-offs/synergies with slope changes, and identifying the dominate trade-off/synergy relationship among ESs will help us to carry out ecosystem regulation according to local conditions through land-use layout optimization at a fine scale. Our research site was located in the Dabie Mountains of western Anhui Province, China (DBM), where, based on the InVEST software, R language, self-organizing maps (SOM), and GeoDA, the temporal and spatial variations of five typical ESs, including food supply, soil retention, water yield, carbon storage, and biodiversity maintenance from 2005 to 2020, were analyzed, and spatial distributions of the different ESs clusters were also recognized by using the SOM method. Moreover, the impacts of land-use type and slope on ESs, and the characteristics of trade-offs/synergies among the five ESs, were discussed. Results showed, firstly, that the total values of ESs showed a changing trend of “three increases and two decreases” from 2005 to 2020. Among the ESs, food supply, soil retention and water yield showed upward trends, with annual growth rates of 2.83%, 6.50% and 2.98%, respectively, whereas carbon storage and biodiversity maintenance showed downward trends, with annual decline rates of 0.03% and 0.07%, respectively. Second, the results showed that the Moran’s I index of the total ESs was 0.3995 in 2005 and 0.4305 in 2020, respectively, indicating that they had significant spatial clustering characteristics. The Low-Low clustering regions with reduced changes were mainly in the central and northern parts of the study area, whereas the High-High clustering regions with increased changes were found distributed mainly in the south of the study area. Thirdly, it was found that cropland and woodland were the main contributors to the total amounts of ESs, but the supply capacity of ESs per unit area of woodland was the largest, constituting nearly 1/3 of the total supply capacity of the ESs. Last, the slope effect on trade-offs and synergies was significant between typical ecosystem service pairs in the study area; trade-offs were the main relationships between the pairs of ESs in the study area, which accounted for nearly 60% of all types of trade-offs/synergies during the 15 years. In addition, the spatial distributions of the trade-offs/synergies between ESs pairs were visualized clearly, and the six ES bundles were identified by using the SOM method at the township administrative scale. The identification of ecosystem service bundles is of great significance for the division of ecological functional zones and ecological regulation in the DBM.

Personalized prediction of the secondary oocytes number after ovarian stimulation: A machine learning model based on clinical and genetic data.

Controlled ovarian stimulation is tailored to the patient based on clinical parameters but estimating the number of retrieved metaphase II (MII) oocytes is a challenge. Here, we have developed a model that takes advantage of the patient's genetic and clinical characteristics simultaneously for predicting the stimulation outcome. Sequence variants in reproduction-related genes identified by next-generation sequencing were matched to groups of various MII oocyte counts using ranking, correspondence analysis, and self-organizing map methods. The gradient boosting machine technique was used to train models on a clinical dataset of 8,574 or a clinical-genetic dataset of 516 ovarian stimulations. The clinical-genetic model predicted the number of MII oocytes better than that based on clinical data. Anti-Müllerian hormone level and antral follicle count were the two most important predictors while a genetic feature consisting of sequence variants in the GDF9, LHCGR, FSHB, ESR1, and ESR2 genes was the third. The combined contribution of genetic features important for the prediction was over one-third of that revealed for anti-Müllerian hormone. Predictions of our clinical-genetic model accurately matched individuals' actual outcomes preventing over- or underestimation. The genetic data upgrades the personalized prediction of ovarian stimulation outcomes, thus improving the in vitro fertilization procedure.

Spatial heterogeneity and partitioning of soil health indicators in the Northern Great Plains using self-organizing map and change point methods

Spatial heterogeneity and partitioning of soil health indicators in the Northern Great Plains using self-organizing map and change point methods

Reconstructing subsurface temperature profiles with sea surface data worldwide through deep evidential regression methods

In this paper, the combination of the deep evidential regression network (DERN) and empirical orthogonal function (EOF) methods, namely, the DERN-EOF method, was proposed to reconstruct the subsurface temperature profiles. Firstly, the DERN method was compared with the previous machine learning methods used in the subsurface temperature field reconstruction, for example the Self-organizing map method. The new network was shown to have certain advantages of higher accuracy and the capability to provide uncertainty estimation, which enables the reconstruction of the subsurface field with higher belief. Analysis of the method performance suggests that the DERN could improve reconstruction accuracy by almost 20% and that the uncertainty statistics agrees well with the data distribution. Then the DERN method combined with the EOF method was used to reconstruct subsurface temperature profiles at a global scale. Results suggest that the reconstruction error in highly dynamic regions, like the Kuroshio Extension and Gulf Stream regions, is significantly higher than that in other regions, and the uncertainties increase with the reconstruction error, suggesting that the proposed method can provide a relatively believable uncertainty estimation.

Uncertain role of clouds in shaping summertime atmosphere-sea ice connections in reanalyses and CMIP6 models

Downwelling longwave radiation (DLR) driven by the atmospheric and cloud conditions in the troposphere is suggested to be a dominant factor to determine the summertime net surface energy budget over the Arctic Ocean and thus plays a key role to shape the September sea ice. We use reanalyses and the self-organizing map (SOM) method to distinguish CMIP6 model performance in replicating the observed strong atmosphere-DLR connection. We find all models can reasonably simulate the linkage between key atmosphere variables and the clear sky DLR but behave differently in replicating the atmosphere-DLR connection due to cloud forcing. In ERA5 and strongly coupled models, tropospheric high pressure is associated with decreased clouds in the mid- and high-levels and increased clouds near the surface. This out-of-phase structure indicates that DLR cloud forcing is nearly neutral, making the clear sky DLR more important to bridge JJA circulation to late-summer sea ice. In MERRA-2 and weakly coupled models, tropospheric clouds display a vertically homogeneous reduction; the cloud DLR is thus strongly reduced due to the cooling effect, which partially cancels out the clear sky DLR and makes the total DLR less efficient to translate circulation forcing to sea ice. The differences of cloud vertical distribution in CMIP6 appear to be differentiated by circulation related relative humidity. Therefore, a better understanding of the discrepancy of different reanalyses and remote sensing products is critical to comprehensively evaluate simulated interactions among circulation, clouds, sea ice and energy budget at the surface in summer.