Temporal Datasets Research Articles

Morphological Stasis in Time? A Triatoma brasiliensis brasiliensis Study Using Geometric Morphometrics in the Long Run.

Simple SummaryTriatomines are vector insects capable of transmitting the protozoan that causes Chagas disease, thus representing a health risk in several countries, especially in Central and South America. Triatoma brasiliensis brasiliensis, the main triatomine vector in northeastern Brazil, needs frequent monitoring as it is able to colonize various natural and artificial ecotopes as well as to infest domiciles. This research uses geometric morphometrics as a tool to evaluate changes in the morphology and analyze a large temporal dataset of 102 years of collections of T. b. brasiliensis. Finding novelty results which indicate a morphological stasis in the morphology.Triatoma brasiliensis brasiliensis Neiva, 1911 is one of the most important vectors of Chagas disease in the Brazilian semiarid regions in the north-east. The risk imposed by T. b. brasiliensis to the human populations, due to frequent invasions and/or colonization of the domiciles, demands constant monitoring and control actions as well as an understanding of its evolutionary process. In this context, the following research studies the pattern of shape adaptation over time using a large dataset from 102 years of specimen collections in order to identify the morphological plasticity of this vector in Brazil. This dataset was analyzed using geometric morphometrics tools and the timescale was divided into eight different groups, containing specimens from 1912 to 2014. Geometric morphometrics analysis showed an interesting morphological stasis in the wing shape of T. b. brasiliensis, which allowed us to understand the high capacity of adaptation to changes in climate condition through time, and the invasive status which Triatoma species have around the world. Moreover, these results showed novel findings as an interesting phenotypic pattern, with no modifications in more than 100 years, leading us to understand the shape evolution in Triatominae as a vector species of diseases.

Consistency and local adaptation in use of ecological and eco-cultural indicators: assessing risk from contamination

The United States and other countries have radiological and chemical legacy wastes remaining from World War II and the Cold War. Assessing risk to human health and the environment from contaminated sites requires inventorying wastes, and examining risks. We use five large U.S. Department of Energy sites to assess the kinds and temporal patterns of indicators used to evaluate ecological resources. Our objective was to determine if there is consistency in types of indicators monitored, whether there are temporal data sets, and how eco-cultural indicators are used. For our assessment, we examined the sites’ Annual Environmental Reports that are meant to inform regulators, stakeholders, resource trustees and the public of their environmental performance in reducing risk and protecting humans and the environment. We present tables of ecological and eco-cultural indicators (and temporal trends) for each site, including contaminant levels in listed species, contaminant levels in consumed species, population levels of endangered/threatened species, biodiversity in communities, and information on invasive species. There is consistency in types of ecological and eco-cultural indicators selected among sites, but the specific indicators differ and are often site-specific. There are temporal patterns for species that serve an ecological and eco-cultural function, and that provide information on risk to eco-receptors and humans. There are fewer cultural indicators, and no temporal trends data for them. The data can be used to improve indicator use and monitoring across the DOE complex, and provide models for assessment of risk to ecological and eco-cultural resources at other contaminated sites. Being able to assess relative risk among sites provides managers, regulators, and the public with information to aid in prioritization of remediation tasks, as well as assessing whether remediation and restoration have reduced risks to ecological receptors and human consumers, and achieved the continued protection of ecological and eco-cultural resources on these sites. It also provides a model to prioritize funds and projects among preserves, national forests and wildlife refuges, and other protected lands. The first step is determining current indicators and commonalities among sites, which will allow managers, public policymakers, and the public to make science-based, adaptive management decisions.

Impact of Expanding Coal Industry on the Regional Environment in an Indian Coal-Belt - A Case Study from Jharkhand, India

Environmental sustainability of coal industry is a major concern considering the trends of its exponential effect on the region and towards climate change. The study examines the land use/land cover changes using spatio temporal datasets of the years 2000 and 2020 in Dhanbad district which comes under the major coal-belt of India. The study shows the impact of expanding industries on the environment depicting a remarkable decrease in vegetation (18.89%), increase in built-up land (14.32%) and coalfields (0.9%) exhibiting massive deforestation activities. The study focuses on the increase in emissions from coal mining activities including GHGs, SOx, PMs, etc. over the years and their harmful impacts on the region. The sustainability evaluation of the coalfields of the study area is done through 15 predetermined indicators, suggesting the urgency and need of actions for sustainable functioning of coal mining industry in order to limit global warming to below 2 °C as suggested in the Paris Agreement-2016. The study also lays stress on the changing economic pattern of the region as an effect of increasing industrialization and changes in the land use.

Remote sensing of atmospheric HDO/H2O in southern California from CLARS-FTS

Atmospheric isotopologues of water vapor (e.g., HDO) are important tracers for understanding Earth's hydrological cycles. Most remote sensing and in-situ measurements of these isotopologues, however, are either column averaged values or sparse in space and time. Measurements targeting the planetary boundary layer (PBL), the part of the atmosphere that has high sensitivity to surface sources of water vapor isotopologues, are much rarer. In this study, we retrieved HDO and H2O columns from observations by the California Laboratory for Atmospheric Remote Sensing Fourier Transform Spectrometer (CLARS-FTS), a mountaintop observatory on Mt. Wilson (1.67 km a.s.l.) overlooking the Los Angeles (LA) basin in southern California. CLARS-FTS observations are highly sensitive to the lower atmosphere due to the long light path along the PBL. Retrievals were conducted using spectral windows between 6000 and 7000 cm−1 from CLARS-FTS observations (2011–2019). The isotopological abundance δD, which represents the relative difference of the HDO/H2O ratio to a standard abundance ratio, is also calculated. The averaged δD retrievals are (−156.1 ± 60.0)‰ with an uncertainty of (6.1 ± 10.2)‰ for LA Basin Survey mode and (−344.7 ± 95.0)‰ with an uncertainty of (42.4 ± 31.6)‰ for Spectralon Viewing Observation mode. In LA, the δD shows a seasonal cycle that is primarily driven by the change of atmospheric humidity. A comparison analysis shows that the δD measurements by CLARS-FTS, a collocated Total Carbon Column Observing Network (TCCON), and the TROPOspheric Monitoring Instrument (TROPOMI) are in good agreement. The difference between CLARS and TCCON δD retrievals can primarily be attributed to the difference in their observation geometries. We envision that the HDO and δD measurements from CLARS-FTS provide high spatial and temporal resolution datasets for further study of hydrological processes, such as the partitioning of the water flux into soil evaporation and transpiration, standing water evaporation, or transport and mixing from the oceans, in the LA megacity.

The Family Meal Framework: A grounded theory study conceptualising the work that underpins the family meal

The family meal has been recognised as an integral part of family life. With the positive health outcomes associated with the family meal, it has been proposed as a strategy for encouraging health-promoting behaviours. However, a detailed understanding of the physical and mental work required to execute the family meal is lacking. The aim of this research was to conduct a grounded theory study to understand the components required to successfully execute the family meal. Two temporal data sets (1993–4/2020) in which diverse participants were sampled were used for this study. Methods used to conduct qualitative interviews with parents in the 1990s were mirrored in the conduct of qualitative interviews with parents in 2020. The interview data was analysed drawing on grounded theory methodology and methods. The entire sample included 54 parents from 28 families. A conceptual framework, ‘The Family Meal Framework’, was developed from the analyses. The five main components of The Framework are the cognitions (invisible work considering the needs of the family), actions (physical tasks required for the family meal), outcomes (the event of the family meal), the beliefs and feelings (expectations and attitudes toward the family meal), and the person(s) responsible (who undertakes the work). This framework provides a novel theory describing the reactive, cyclical nature of the work required to execute the family meal. This new understanding provides discrete opportunities for intervention in family meal research, practice, policy and promotion.

Four millennia of long-term individual foraging site fidelity in a highly migratory marine predator

Theory and field studies suggest that long-term individual foraging site fidelity (IFSF) may be an important adaptation to competition from increasing population. However, the driving mechanisms and extent of long-term IFSF in wild populations of long-lived, migratory animals has been logistically difficult to study, with only a few confirmed instances. Temporal isotopic datasets can reveal long-term patterns in geographical foraging behaviour. We investigate the isotopic compositions of endangered short-tailed albatross (Phoebastria albatrus) over four millennia leading up to their near-extinction. Although not exhibited by short-tailed albatross today, we show past sub-populations displayed a high-degree of long-term IFSF, focusing on the same locations for hundreds of generations. This is the first large-scale evidence for the deep antiquity of long-term IFSF and suggests that it’s density-driven. Globally, as populations of species like short-tailed albatross continue to recover from overexploitation, potential for resurgence of geographic specialization may increase exposure to localized hazards, requiring closer conservation monitoring.

Distinct body-size responses to warming climate in three rodent species.

In mammals, body-size responses to warming climates are diverse, and the mechanisms underlying these different responses have been little investigated. Using temporal and spatial datasets of three rodent species distributed across different climatic zones in China, we investigated temporal and spatial trends of body size (length and mass), identified the critical drivers of these trends, and inferred the potential causes underlying the distinct body-size responses to the critical drivers. We found that body mass of all species remained stable over time and across space. Body length, however, increased in one species over time and in two species across space. Generally, body-length variation was predicted best by minimum ambient temperature. Moreover, in two species, body length changed linearly with temperature differences between ancestral and colonization areas. These distinct temperature-length patterns may jointly be caused by species-specific temperature sensitivities and experienced magnitudes of warming. We hypothesize that species or populations distributed across distinct temperature gradients evolved different intrinsic temperature sensitivities, which affect how their body sizes respond to warming climates. Our results suggest that size trends associated with climate change should be explored at higher temporal and spatial resolutions, and include clades of species with similar distributions.

Rewiring of the Liver Transcriptome across Multiple Time-Scales Is Associated with the Weight Loss-Independent Resolution of NAFLD Following RYGB.

Roux-en-Y gastric bypass (RYGB) surgery potently improves obesity and a myriad of obesity-associated co-morbidities including type 2 diabetes and non-alcoholic fatty liver disease (NAFLD). Time-series omics data are increasingly being utilized to provide insight into the mechanistic underpinnings that correspond to metabolic adaptations in RYGB. However, the conventional computational biology methods used to interpret these temporal multi-dimensional datasets have been generally limited to pathway enrichment analysis (PEA) of isolated pair-wise comparisons based on either experimental condition or time point, neither of which adequately capture responses to perturbations that span multiple time scales. To address this, we have developed a novel graph network-based analysis workflow designed to identify modules enriched with biomolecules that share common dynamic profiles, where the network is constructed from all known biological interactions available through the Kyoto Encyclopedia of Genes and Genomes (KEGG) resource. This methodology was applied to time-series RNAseq transcriptomics data collected on rodent liver samples following RYGB, and those of sham-operated and weight-matched control groups, to elucidate the molecular pathways involved in the improvement of as NAFLD. We report several network modules exhibiting a statistically significant enrichment of genes whose expression trends capture acute-phase as well as long term physiological responses to RYGB in a single analysis. Of note, we found the HIF1 and P53 signaling cascades to be associated with the immediate and the long-term response to RYGB, respectively. The discovery of less intuitive network modules that may have gone overlooked with conventional PEA techniques provides a framework for identifying novel drug targets for NAFLD and other metabolic syndrome co-morbidities.

3D Radiative Transfer for Exoplanet Atmospheres. gCMCRT: A GPU-accelerated MCRT Code

Radiative transfer (RT) is a key component for investigating atmospheres of planetary bodies. With the 3D nature of exoplanet atmospheres being important in giving rise to their observable properties, accurate and fast 3D methods are required to be developed to meet future multidimensional and temporal data sets. We develop an open-source GPU RT code, gCMCRT, a Monte Carlo RT forward model for general use in planetary atmosphere RT problems. We aim to automate the post-processing pipeline, starting from direct global circulation model (GCM) output to synthetic spectra. We develop albedo, emission, and transmission spectra modes for 3D and 1D input structures. We include capability to use correlated-k and high-resolution opacity tables, the latter of which can be Doppler-shifted inside the model. We post-process results from several GCM groups, including ExoRad, SPARC/MITgcm THOR, UK Met Office UM, Exo-FMS, and the Rauscher model. Users can therefore take advantage of desktop and HPC GPU computing solutions. gCMCRT is well suited for post-processing large GCM model grids produced by members of the community and for high-resolution 3D investigations.

Development and validation of early prediction models for new-onset functional impairment at hospital discharge of ICU admission.

We aimed to develop and validate models for predicting new-onset functional impairment after intensive care unit (ICU) admission with predictors routinely collected within 2days of admission. In this multi-center retrospective cohort study of acute care hospitals in Japan, we identified adult patients who were admitted to the ICU with independent activities of daily living before hospitalization and survived for at least 2days from April 2014 to October 2020. The primary outcome was functional impairment defined as Barthel Index ≤ 60 at hospital discharge. In the internal validation dataset (April 2014 to March 2019), using routinely collected 94 candidate predictors within 2 days of ICU admission, we trained and tuned the six conventional and machine-learning models with repeated random sub-sampling cross-validation. We computed the variable importance of each predictor to the models. In the temporal validation dataset (April 2019 to October 2020), we measured the performance of these models. We identified 19,846 eligible patients. Functional impairment at discharge was developed in 33% of patients (n = 6488/19,846). In the temporal validation dataset, all six models showed good discrimination ability with areas under the curve above 0.86, and the differences among the six models were negligible. Variable importance revealed newly detected early predictors, including worsened neurologic conditions and catabolism biomarkers such as decreased serum albumin and increased blood urea nitrogen. We successfully developed early prediction models of new-onset functional impairment after ICU admission that achieved high performance using only data routinely collected within 2days of ICU admission.

Design of an architecture for storing a time connected model of a digital twin

Recently, in connection with the development of information technology, digital twin technology has become widespread, which allows replacing physical and analog resources with digital ones. The digital twin is also a relevant and effective method of real-time interaction. However, with a wealth of real-time data and information from the digital twin, it is difficult to present them intuitively to users and operators. A digital twin for the study of an arbitrary object can formally be represented as mathematical abstractions designed for efficient processing of temporal multimodal da-ta of the object under study. Based on the above abstractions and their application for the formal specification of the digital twin, the so-called multiimage is built, which ac-tually provides a formal description of the sequences of multimodal data about the ob-ject under study, obtained taking into account time as a result of measurement, genera-tion and other related data acquisition operations. As a result of the study, a sequence of discrete states of the object under study is formed, each of which can be characterized by a set of temporal multimodal data associated with a specific moment in time, which actually serves as a temporal key for a certain state of the object. The purpose of the study is to design the architecture of a software system for stor-ing and then reproducing a temporal connected model using a database designed to store time series data. TimescaleDB, a relational database for time series data, is pro-posed as the main data storage engine. The main challenges for a database that must process a large amount of temporary data is its scalability - to process and store a large amount of data without performance degradation. The proposed method for building a temporal connected model of a digital twin based on TimescaleDB using the concept of hypertables will achieve efficient use of re-sources by parallelizing the load when searching for data, and a three-level architec-ture designed to distribute the responsibilities of a software system can greatly simplify the implementation and subsequent processing of a temporal model of a digital twin.

Training concepts in Noise Clustering Classifier -A case study for Pigeon Pea crop mapping

In this research work, a novel approach of ‘Individual Sample as Mean’ (ISM) based Noise Clustering has been proposed, where each sample is taken as an input parameter in training the model. When compared to the existing method which uses statistical mean for training, this proposed technique is far superior and robust, the efficiency of which has been demonstrated using Pigeon Pea crop mapping. The issue of spectral overlap of Pigeon Pea with other crops in vicinity is overcome by adopting a temporal approach. In order to fill the void in multi-temporal data availability, dual sensor approach was implemented. Different vegetation indices have been tested in reducing data dimensionality and the temporal datasets were optimized so as to capture the unique growth phenology of the Pigeon Pea crop. A new technique of ‘Class Based Sensor Independent – Modified Soil-Adjusted Vegetation Index 2 (CBSI-MSAVI2)’ has been proposed in this research work for the purpose of dimensionality reduction which ensures the maximum enhancement of target crop class in temporal domain. Based on an evaluation of classification results using Mean Membership Difference (MMD) method, the proposed ISM-Based NC appears to have produced superior results for Pigeon Pea since the test fields had membership values that were closer to those of the training fields. The MMD for ISM-Based NC was around 0.0013 whereas for the usual Mean-Based approach the value was at around 0.0071 respectively for Pigeon Pea crop fields. It was also noted that variance was zero in outputs generated by the ISM-based NC approach thereby proving the efficiency of the same in suppressing heterogeneity within the field.

Machine-Learning Approaches for Predicting the Need of Oxygen Therapy in Early-Stage COVID-19 in Japan: Multicenter Retrospective Observational Study.

BackgroundEarly prediction of oxygen therapy in patients with coronavirus disease 2019 (COVID-19) is vital for triage. Several machine-learning prognostic models for COVID-19 are currently available. However, external validation of these models has rarely been performed. Therefore, most reported predictive performance is optimistic and has a high risk of bias. This study aimed to develop and validate a model that predicts oxygen therapy needs in the early stages of COVID-19 using a sizable multicenter dataset.MethodsThis multicenter retrospective study included consecutive COVID-19 hospitalized patients confirmed by a reverse transcription chain reaction in 11 medical institutions in Fukui, Japan. We developed and validated seven machine-learning models (e.g., penalized logistic regression model) using routinely collected data (e.g., demographics, simple blood test). The primary outcome was the need for oxygen therapy (≥1 L/min or SpO2 ≤ 94%) during hospitalization. C-statistics, calibration slope, and association measures (e.g., sensitivity) evaluated the performance of the model using the test set (randomly selected 20% of data for internal validation). Among these seven models, the machine-learning model that showed the best performance was re-evaluated using an external dataset. We compared the model performances using the A-DROP criteria (modified version of CURB-65) as a conventional method.ResultsOf the 396 patients with COVID-19 for the model development, 102 patients (26%) required oxygen therapy during hospitalization. For internal validation, machine-learning models, except for the k-point nearest neighbor, had a higher discrimination ability than the A-DORP criteria (P < 0.01). The XGboost had the highest c-statistic in the internal validation (0.92 vs. 0.69 in A-DROP criteria; P < 0.001). For the external validation with 728 temporal independent datasets (106 patients [15%] required oxygen therapy), the XG boost model had a higher c-statistic (0.88 vs. 0.69 in A-DROP criteria; P < 0.001).ConclusionsMachine-learning models demonstrated a more significant performance in predicting the need for oxygen therapy in the early stages of COVID-19.

Spatiotemporal mapping of inundation area at Lake Limboto in Gorontalo, Indonesia, using cloud computing technology

Monitoring activities on the dynamics of water shrinkage at Lake Limboto are essential to the lake’s ecosystem’s recovery. A remote sensing technology functions to monitor the dynamics of lake inundation area; this allows one to produce a comprehensive set of spatial and temporal data. Such complex satellite dataset demands extra time, greater storage resources, and greater computing capacity. The Google Earth Engine platform emerges as the alternative to tackle such problems. The present study aims to explore the capability of Google Earth Engine in formulating spatial and temporal maps of the inundation area at Lake Limboto. A total of 345 scenes of Landsat image on the study area (available during the period of 1989–2019) were involved in generating a quick inundation area map of the lake. The whole processes (pre-processing, processing, analysing, and evaluating) were automatized by using the Google Earth Engine interface. The evaluation of mapping result accuracy indicated that the average score of F1-score and Intersection over Union (IoU) was at 0.88 and 0.91, respectively. Moreover, the mapping results of the lake’s inundation area from 1989 to 2019 showed that the inundation area tended to decrease significantly in size over time. During the period, the lake’s area also shrank from 3023.8 ha in 1989 to 1275.0 ha in 2019. All in all, the spatiotemporal information about the changes in lake area may be treated as a reference for decision-making processes of lake management in the future.

Temporal Trends of Persistent Organic Pollutants in Sarasota Bay Common Bottlenose Dolphins (Tursiops truncatus)

Legacy persistent organic pollutants (POPs) were measured in blubber samples collected from 196 common bottlenose dolphins (Tursiops truncatus) from 2000 to 2016 in the Sarasota Bay, Florida, ecosystem. Legacy POPs included polychlorinated biphenyl congeners (ΣPCB; sum of 37 congeners or congener groups), dichloro-diphenyl-trichloroethane (DDT) and transformation products (ΣDDT), chlordane and related compounds (ΣChlor), mirex, polybrominated diphenyl ether congeners (ΣPBDE, sum of five congeners), dieldrin and hexachlorobenzene (HCB). All POP compounds or compound groups declined in dolphin blubber over the study period. POP classes declined at different rates within classifications (sex, adult or calf). For example, dieldrin decline in male dolphins was greatest (−13% per year) while HCB decline was lower (−6.3% per year). For individual POP classes, rates of decline depended on classification. For example, ΣPCB, the POP class present at the highest concentration relative to other POPs, declined at −8.4, −7.8, and −5.6% per year in adult females, adult males and calves, respectively. Overall POP declines were highest in adult males and lowest in calves. Declines in POP concentration with time based on individual dolphins resampled at different time points were consistent with declines calculated based on yearly or biennial cross-sections taken from 2000 through 2016. Overall rates of decline in Sarasota Bay bottlenose dolphins were like declines seen for Great Lakes fish and about twice the rate of decline observed in arctic temporal trend data sets. POP trends appear to have leveled off since 2009 in Sarasota Bay dolphins. For example, ΣPCBs in calves appear to have stabilized at 21 mg/kg lipid. Recent published work on delphinids with blubber concentrations spanning those observed in the present work suggest that levels of legacy POPs currently observed in Sarasota dolphins may suppress population growth. Results from this study confirm the need for continued monitoring of POPs in the Sarasota Bay bottlenose dolphin population to see if concentrations are continuing to fall or have indeed plateaued.

Evaluation of domain generalization and adaptation on improving model robustness to temporal dataset shift in clinical medicine

Temporal dataset shift associated with changes in healthcare over time is a barrier to deploying machine learning-based clinical decision support systems. Algorithms that learn robust models by estimating invariant properties across time periods for domain generalization (DG) and unsupervised domain adaptation (UDA) might be suitable to proactively mitigate dataset shift. The objective was to characterize the impact of temporal dataset shift on clinical prediction models and benchmark DG and UDA algorithms on improving model robustness. In this cohort study, intensive care unit patients from the MIMIC-IV database were categorized by year groups (2008–2010, 2011–2013, 2014–2016 and 2017–2019). Tasks were predicting mortality, long length of stay, sepsis and invasive ventilation. Feedforward neural networks were used as prediction models. The baseline experiment trained models using empirical risk minimization (ERM) on 2008–2010 (ERM[08–10]) and evaluated them on subsequent year groups. DG experiment trained models using algorithms that estimated invariant properties using 2008–2016 and evaluated them on 2017–2019. UDA experiment leveraged unlabelled samples from 2017 to 2019 for unsupervised distribution matching. DG and UDA models were compared to ERM[08–16] models trained using 2008–2016. Main performance measures were area-under-the-receiver-operating-characteristic curve (AUROC), area-under-the-precision-recall curve and absolute calibration error. Threshold-based metrics including false-positives and false-negatives were used to assess the clinical impact of temporal dataset shift and its mitigation strategies. In the baseline experiments, dataset shift was most evident for sepsis prediction (maximum AUROC drop, 0.090; 95% confidence interval (CI), 0.080–0.101). Considering a scenario of 100 consecutively admitted patients showed that ERM[08–10] applied to 2017–2019 was associated with one additional false-negative among 11 patients with sepsis, when compared to the model applied to 2008–2010. When compared with ERM[08–16], DG and UDA experiments failed to produce more robust models (range of AUROC difference, − 0.003 to 0.050). In conclusion, DG and UDA failed to produce more robust models compared to ERM in the setting of temporal dataset shift. Alternate approaches are required to preserve model performance over time in clinical medicine.

Intercomparison of Sentinel-2 and modelled snow cover maps in a high-elevation Alpine catchment

Modelling runoff generation in high-elevation Alpine catchments requires detailed knowledge on the spatio-temporal distribution of snow storage. With Sentinel-2 MultiSpectral Instrument (MSI), it is possible to map snow cover with a high temporal and spatial resolution. In contrast to the coarse MODIS data, Sentinel-2 MSI enables the investigation of small-scale differences in snow cover duration in complex terrains due to gravitational redistribution (slope), energy balance and wind-driven redistribution (aspect). In this study, we describe the generation of high-resolution spatial and temporal snow cover data sets from Sentinel-2 images for a high-elevation Alpine catchment and discuss how the data contribute to our understanding of the spatio-temporal snow cover distribution. The quality of snow and cloud detection is evaluated against in-situ snow observations and against other snow and cloud products. The main problem was in the false detection of snow in the presence of clouds and in topographically shaded areas. We then seek to explore the potential of the generated high-resolution snow cover maps in calibrating the gravitational snow redistribution module of a physically based snow model, especially for an area with a very data-scarce point snow observation network. Generally, the calibrated snow model is able to simulate both the mean snow cover duration with a high F1 accuracy score of > 0.9 and the fractional snow-covered area with a correlation coefficient of 0.98. The snow model is also able to reproduce spatio-temporal variability in snow cover duration due to surface energy balance dynamics, wind and gravitational redistribution.

A preliminary assessment of the space‐for‐time substitution method in soil carbon change prediction

AbstractPrediction of the soil C response to environmental change is often limited by the absence of temporal soil monitoring data from which applicable models can be developed. This constrains the understanding of the processes governing soil C cycling in data‐poor areas. Here, we demonstrate a method to predict soil organic C (SOC) change over time based on the space‐for‐time (SFT) substitution method by constructing spatial and temporal datasets of SOC change from soil datasets pertaining to four periods over the past 40 yr in China. We empirically test the assumption that environmental drivers of the spatial gradients of SOC variation also drive SOC change over time. We find that the accuracy of SOC change predictions obtained with the SFT method is comparable to that of predictions directly obtained with temporal models, with the prediction performance ranging from 86 to 118%. Our results generally support the application of the SFT method to model SOC dynamics over time and highlight the importance of this method to better understand soil C cycling in ecosystems associated with insufficient soil monitoring data. Our findings provide a priori guidance for the suitability of the SFT method as an approach to infer the soil C response to environmental change.

Generating pre-harvest crop maps by applying convolutional neural network on multi-temporal Sentinel-1 data

ABSTRACT Pre-harvest crop mapping, the fundamental requirement for many of the crop management decisions, continues to be challenging either due to cloud cover in satellite images or due to spectral separability issues. These limitations are overcome in this study by using Synthetic Aperture Radar (SAR) data and deep learning technique. Two-dimensional convolutional neural network (2D-CNN) architecture is applied on multi-temporal SAR data of Sentinel-1 to classify soybean, jowar, cotton and sugarcane crops in a large geographic area located in a Central Indian state. Classification experiments are conducted with full season data from nine overpasses in three modes namely VH alone, VH and VH/VV ratio, and VH and VV, and these experimental results reveal that VH and VV combination performed better with an overall accuracy of 91.75% as compared to 84.96% and 88.75% respectively by others. Classification performances with three sets of temporal data, covering part of the season, reveal that crop map can be generated as early as 27th August (i.e. roughly 45 days prior harvesting) with an accuracy of 89.15% slightly less than the mapping accuracy achieved with full season data (i.e. till 14th October). 2D-CNN algorithm has performed better than SVM and RF techniques. This methodology can be extended to similar agro-ecological regions of the country. Richly available SAR data from Sentinel-1 and the potential of deep learning techniques for recognising complex phenological patterns offer immense opportunities for early-season crop mapping even during monsoon season in tropical countries like India.

Force-Directed Timelines: Visualizing &amp; Exploring Temporal Patterns

Visualization has shown to be a valuable tool in the analysis of large and complex temporal datasets, aided by the emergence of new models such as Time Curves, which distorts timelines to position time points based on their similarity with each other, reflecting changes in the data over time. In this paper, we further explore time-series functionally and aesthetically by presenting an interactive and parameter-based implementation of the Time Curves model, complemented with the addition of supporting visualizations and data analysis methods. In our implementation we introduce Time Paths, a force-directed layout that can dynamically transform the original model to not only smoothen the transitions between time points, but also reduce visual noise in favor of portraying overall patterns. The proposed addition of visual elements to the model includes temporal glyphs and a supporting timeline graph which help discover and better understand temporal patterns across complex datasets. Through interactive exploration, we demonstrate how these methods can be used to analyze and identify the main agents at the source of significant instances in three biological datasets. These methods are presented within CroP, a data visualization tool with coordinated multiple views aimed at the analysis of biological datasets.