scholarly journals Isoscape of amount-weighted annual mean precipitation tritium (<sup>3</sup>H) activity from 1976 to 2017 for the Adriatic–Pannonian region – AP<sup>3</sup>H_v1 database

2020 ◽  
Vol 12 (3) ◽  
pp. 2061-2073 ◽  
Author(s):  
Zoltán Kern ◽  
Dániel Erdélyi ◽  
Polona Vreča ◽  
Ines Krajcar Bronić ◽  
István Fórizs ◽  
...  
Keyword(s):  
Water Molecule ◽  
Time Scale ◽  
Model Performance ◽  
Special Focus ◽  
Global Models ◽  
Tritium Activity ◽  
Out Of Sample ◽  
Natural Tracer

Abstract. Tritium (3H) as a constituent of the water molecule is an important natural tracer in hydrological sciences. The anthropogenic tritium introduced into the atmosphere unintentionally became an excellent tracer of processes on a time scale of up to 100 years. A prerequisite for tritium applications is to know the distribution of tritium activity in precipitation. Here we present a database of isoscapes derived from 41 stations for amount-weighted annual mean tritium activity in precipitation for the period 1976 to 2017 on spatially continuous interpolated 1 km×1 km grids for the Adriatic–Pannonian region (called the AP3H_v1 database), with a special focus on post-2010 years, which are not represented by existing global models. Five stations were used for out-of-sample evaluation of the model performance, independently confirming its capability of reproducing the spatiotemporal tritium variability in the region. The AP3H database is capable of providing reliable spatiotemporal input for hydrogeological application at any place within Slovenia, Hungary, and their surroundings. Results also show a decrease in the average spatial representativity of the stations regarding tritium activity in precipitation from ∼440 km in 1970s, when bomb tritium still prevailed in precipitation, to ∼235 km in the 2010s. The post-2010 isoscapes can serve as benchmarks for background tritium activity for the region, helping to determine potential future local increases in technogenic tritium from these backgrounds. The gridded tritium isoscape is available in NetCDF-4 at https://doi.org/10.1594/PANGAEA.896938 (Kern et al., 2019).

scholarly journals Isoscape of precipitation amount-weighted annual mean tritium (<sup>3</sup>H) activity from 1976 to 2017 for the Adriatic-Pannonian region

2020 ◽  
Author(s):  
Zoltán Kern ◽  
István Gábor Hatvani ◽  
Dániel Erdélyi ◽  
Polona Vreča ◽  
Ines Krajcar Bronić ◽  
...  
Keyword(s):  
Water Molecule ◽  
Time Scale ◽  
Input Data ◽  
Precipitation Amount ◽  
Model Performance ◽  
Regional Model ◽  
Special Focus ◽  
Global Models ◽  
Tritium Activity ◽  
Natural Tracer

Abstract. Tritium (3H) as a constituent of the water molecule is an important natural tracer in hydrological sciences. The anthropogenic tritium introduced into the atmosphere became unintentionally an excellent tracer of processes on the time scale of up to a 100 years. A prerequisite for tritium applications is to know the distribution of tritium activity in precipitation. Here we present the spatially continuous gridded database (isoscapes) for amount-weighted annual mean tritium activity in precipitation for the period 1976 to 2017 on 1 × 1 km grids for the Adriatic-Pannonian Region (using 39 stations), with a special focus on post-2010 years which are not represented by existing global models. Three stations were used to check the model performance independently confirming its capability to reproducing the spatiotemporal tritium variability in the region. This Regional model is capable of providing reliable spatiotemporal input data for hydrogeological application at any place within Slovenia, Hungary and its surroundings. Results also show a decrease in the average spatial representativity of the stations regarding tritium activity in precipitation from ~ 600 km in 1970s when bomb-tritium was still prevailing in precipitation, to ~ 300 km in the 2010s. The post-2010 isoscapes can serve as benchmarks for background tritium activity for the region, helping to determine local increases of technogenic tritium from these backgrounds. The gridded tritium isoscape is available in NetCDF-4 at doi:10.1594/PANGAEA.896938 (Kern et al., 2019).

Predicting the trend of stock index based on feature engineering and CatBoost model

2021 ◽  
pp. 2150027
Author(s):  
Renzhe Xu ◽  
Yudong Chen ◽  
Tenglong Xiao ◽  
Jingli Wang ◽  
Xiong Wang
Keyword(s):  
Time Scale ◽  
Stock Index ◽  
Feature Engineering ◽  
Multiple Time ◽  
Multiple Time Scale ◽  
Back Testing ◽  
Price Series ◽  
Out Of Sample ◽  
Out Of Sample Prediction ◽  
Sharp Ratio

As an important tool to measure the current situation of the whole stock market, the stock index has always been the focus of researchers, especially for its prediction. This paper uses trend types, which are received by clustering price series under multiple time scale, combined with the day-of-the-week effect to construct a categorical feature combination. Based on the historical data of six kinds of Chinese stock indexes, the CatBoost model is used for training and predicting. Experimental results show that the out-of-sample prediction accuracy is 0.55, and the long–short trading strategy can obtain average annualized return of 34.43%, which is a great improvement compared with other classical classification algorithms. Under the rolling back-testing, the model can always obtain stable returns in each period of time from 2012 to 2020. Among them, the SSESC’s long–short strategy has the best performance with an annualized return of 40.85% and a sharp ratio of 1.53. Therefore, the trend information on multiple time-scale features based on feature engineering can be learned by the CatBoost model well, which has a guiding effect on predicting stock index trends.

scholarly journals Vocal markers of Autism Spectrum Disorder: assessing the generalizability of machine learning models

2021 ◽  
Author(s):  
Astrid Rybner ◽  
Emil Trenckner Jessen ◽  
Marie Damsgaard Mortensen ◽  
Stine Nyhus Larsen ◽  
Ruth Grossman ◽  
...  
Keyword(s):  
Machine Learning ◽  
Autism Spectrum Disorder ◽  
Model Performance ◽  
Autism Spectrum ◽  
Spectrum Disorder ◽  
Future Studies ◽  
Out Of Sample ◽  
Biobehavioral Markers ◽  
Similar Task ◽  
Machine Learning Models

Background: Machine learning (ML) approaches show increasing promise to identify vocal markers of Autism Spectrum Disorder (ASD). Nonetheless, it is unclear to what extent such markers generalize to new speech samples collected in diverse settings such as using a different speech task or a different language. Aim: In this paper, we systematically assess the generalizability of ML findings across a variety of contexts. Methods: We re-train a promising published ML model of vocal markers of ASD on novel cross-linguistic datasets following a rigorous pipeline to minimize overfitting, including cross-validated training and ensemble models. We test the generalizability of the models by testing them on i) different participants from the same study, performing the same task; ii) the same participants, performing a different (but similar) task; iii) a different study with participants speaking a different language, performing the same type of task. Results: While model performance is similar to previously published findings when trained and tested on data from the same study (out-of-sample performance), there is considerable variance between studies. Crucially, the models do not generalize well to new similar tasks and not at all to new languages. The ML pipeline is openly shared. Conclusion: Generalizability of ML models of vocal markers - and more generally biobehavioral markers - of ASD is an issue. We outline three recommendations researchers could take in order to be more explicit about generalizability and improve it in future studies.

scholarly journals The sub-annual calibration of hydrological models considering climatic intra-annual variations

2017 ◽  
Author(s):  
Binru Zhao ◽  
Huichao Dai ◽  
Dawei Han ◽  
Guiwen Rong
Keyword(s):  
Time Scale ◽  
Temporal Dynamics ◽  
Model Performance ◽  
Optimal Time ◽  
Temperature Pattern ◽  
Climate Conditions ◽  
Fcm Algorithm ◽  
Calibration Scheme ◽  
Annual Variations ◽  
Improve Model

Abstract. Changing climate leads to change of temporal dynamics of hydrological systems by affecting the catchment conditions. Considering climatic variations when calibrating a hydrological model can improve model performance, which allows parameter sets to vary according to sub-periods with different climate conditions. This study has explored climatic intra-annual variations by using two classification approaches to recognize the sub-periods with similar climatic patterns, Calendar-Based Grouping (CBG) method and Fuzzy C-Means (FCM) algorithm. The model performances of the sub-annual calibration schemes based on these two approaches are compared using the conceptual model IHACRES. The effect of time scales on sub-annual calibration schemes was also studied. Results indicate that the sub-annual calibration scheme based on CBG method performs better than that based on Rainfall-dominated FCM algorithm, since the CBG method has a better performance in recognizing temperature pattern, and the main source of catchment change is from the change of vegetation, which is mainly affected by temperature in the study site. The optimal time scale is dependent on the sub-annual calibration scheme, with bimonthly for CBG method and Temperature-dominated FCM algorithm and seasonal for Rainfall-dominated FCM algorithm. Overall, when using sub-annual calibration schemes, the selection of the partitioning method and time scale is very important to model performances

scholarly journals Combining multiple statistical methods to evaluate the performance of process-based vegetation models across three forest stands

2017 ◽  
Vol 63 (4) ◽  
pp. 153-172 ◽  
Author(s):  
Joanna A. Horemans ◽  
Alexandra Henrot ◽  
Christine Delire ◽  
Chris Kollas ◽  
Petra Lasch-Born ◽  
...  
Keyword(s):  
Time Scale ◽  
Singular Spectrum Analysis ◽  
Model Performance ◽  
Model Errors ◽  
Forest Stands ◽  
Climate Conditions ◽  
Management Actions ◽  
Statistical Measures ◽  
Net Ecosystem Carbon Exchange ◽  
Vegetation Models

AbstractProcess-based vegetation models are crucial tools to better understand biosphere-atmosphere exchanges and ecophysiological responses to climate change. In this contribution the performance of two global dynamic vegetation models, i.e. CARAIB and ISBACC, and one stand-scale forest model, i.e. 4C, was compared to long-term observed net ecosystem carbon exchange (NEE) time series from eddy covariance monitoring stations at three old-grown European beech (Fagus sylvatica L.) forest stands. Residual analysis, wavelet analysis and singular spectrum analysis were used beside conventional scalar statistical measures to assess model performance with the aim of defining future targets for model improvement. We found that the most important errors for all three models occurred at the edges of the observed NEE distribution and the model errors were correlated with environmental variables on a daily scale. These observations point to possible projection issues under more extreme future climate conditions. Recurrent patterns in the residuals over the course of the year were linked to the approach to simulate phenology and physiological evolution during leaf development and senescence. Substantial model errors occurred on the multi-annual time scale, possibly caused by the lack of inclusion of management actions and disturbances. Other crucial processes defined were the forest structure and the vertical light partitioning through the canopy. Further, model errors were shown not to be transmitted from one time scale to another. We proved that models should be evaluated across multiple sites, preferably using multiple evaluation methods, to identify processes that request reconsideration.

scholarly journals The representation of solar cycle signals in stratospheric ozone – Part 2: Analysis of global models

2018 ◽  
Vol 18 (15) ◽  
pp. 11323-11343 ◽  
Author(s):  
Amanda C. Maycock ◽  
Katja Matthes ◽  
Susann Tegtmeier ◽  
Hauke Schmidt ◽  
Rémi Thiéblemont ◽  
...  
Keyword(s):  
Solar Cycle ◽  
Atmospheric Chemistry ◽  
Climate Models ◽  
Stratospheric Ozone ◽  
Temperature Response ◽  
Climate Impacts ◽  
Spectral Irradiance ◽  
Special Focus ◽  
Global Models ◽  
The Impact

Abstract. The impact of changes in incoming solar irradiance on stratospheric ozone abundances should be included in climate simulations to aid in capturing the atmospheric response to solar cycle variability. This study presents the first systematic comparison of the representation of the 11-year solar cycle ozone response (SOR) in chemistry–climate models (CCMs) and in pre-calculated ozone databases specified in climate models that do not include chemistry, with a special focus on comparing the recommended protocols for the Coupled Model Intercomparison Project Phase 5 and Phase 6 (CMIP5 and CMIP6). We analyse the SOR in eight CCMs from the Chemistry–Climate Model Initiative (CCMI-1) and compare these with results from three ozone databases for climate models: the Bodeker Scientific ozone database, the SPARC/Atmospheric Chemistry and Climate (AC&amp;C) ozone database for CMIP5 and the SPARC/CCMI ozone database for CMIP6. The peak amplitude of the annual mean SOR in the tropical upper stratosphere (1–5 hPa) decreases by more than a factor of 2, from around 5 to 2 %, between the CMIP5 and CMIP6 ozone databases. This substantial decrease can be traced to the CMIP5 ozone database being constructed from a regression model fit to satellite and ozonesonde measurements, while the CMIP6 database is constructed from CCM simulations. The SOR in the CMIP6 ozone database therefore implicitly resembles the SOR in the CCMI-1 models. The structure in latitude of the SOR in the CMIP6 ozone database and CCMI-1 models is considerably smoother than in the CMIP5 database, which shows unrealistic sharp gradients in the SOR across the middle latitudes owing to the paucity of long-term ozone measurements in polar regions. The SORs in the CMIP6 ozone database and the CCMI-1 models show a seasonal dependence with enhanced meridional gradients at mid- to high latitudes in the winter hemisphere. The CMIP5 ozone database does not account for seasonal variations in the SOR, which is unrealistic. Sensitivity experiments with a global atmospheric model without chemistry (ECHAM6.3) are performed to assess the atmospheric impacts of changes in the representation of the SOR and solar spectral irradiance (SSI) forcing between CMIP5 and CMIP6. The larger amplitude of the SOR in the CMIP5 ozone database compared to CMIP6 causes a likely overestimation of the modelled tropical stratospheric temperature response between 11-year solar cycle minimum and maximum by up to 0.55 K, or around 80 % of the total amplitude. This effect is substantially larger than the change in temperature response due to differences in SSI forcing between CMIP5 and CMIP6. The results emphasize the importance of adequately representing the SOR in global models to capture the impact of the 11-year solar cycle on the atmosphere. Since a number of limitations in the representation of the SOR in the CMIP5 ozone database have been identified, we recommend that CMIP6 models without chemistry use the CMIP6 ozone database and the CMIP6 SSI dataset to better capture the climate impacts of solar variability. The SOR coefficients from the CMIP6 ozone database are published with this paper.

scholarly journals Multiple isotopic models using various kinetic fractionation coefficients to estimateδ18O of leaf water

2018 ◽  
Author(s):  
Yonge Zhang ◽  
Xinxiao Yu ◽  
Lihua Chen ◽  
Guodong Jia
Keyword(s):  
Steady State ◽  
Time Scale ◽  
Model Performance ◽  
Special Care ◽  
Leaf Water ◽  
Data Sets ◽  
Kinetic Fractionation ◽  
Improved Model ◽  
Material Exchange ◽  
Good Agreement

AbstractInvestigation ofδ18O of leaf water may improve our understanding of the evapotranspiration partitioning and material exchange between the inside and outside of leaves. In this study,δ18O of bulk leaf water (δL,b) was estimated by both isotopic–steady–state (ISS) and non–steady–state (NSS) assumptions considering the Péclet effect. Specifically, we carefully modified kinetic fractionation coefficients (αk). The results showed that the Péclet effect is required to predictδL,b. On the diel time scale, both NSS assumption + Péclet effect (NSS + P) and ISS assumption + Péclet effect (ISS + P) using modifiedαk(αk–modified) forδL,bshowed a good agreement with observedδL,b(p> 0.05). When using previously proposedαk, however, both NSS + P and ISS + P were not reliable estimators ofδL,b(p< 0.05). On a longer time scale (days), estimates of daily meanδL,bfrom ISS + P outperformed the estimates from NSS + P when using the sameαkvalues. Also, the employment ofαk–modifiedimproved model performance in predicting daily meanδL,bcompared to the use of previously proposedαk. Clearly, special care must be taken concerningαkwhen using isotopic models to estimateδL,b.HighlightFor hourly and daily mean data sets, the employment of modified kinetic fractionation coefficients significantly improved model performance forδ18O of bulk leaf water.

scholarly journals A comparison of attentional neural network architectures for modeling with electronic medical records

JAMIA Open ◽  
2021 ◽  
Vol 4 (3) ◽  
Author(s):  
Anthony Finch ◽  
Alexander Crowell ◽  
Yung-Chieh Chang ◽  
Pooja Parameshwarappa ◽  
Jose Martinez ◽  
...  
Keyword(s):  
Transfer Learning ◽  
Model Performance ◽  
Mortality Prediction ◽  
Weighted Averaging ◽  
Data Sets ◽  
Kaiser Permanente ◽  
Out Of Sample ◽  
Medical Modeling ◽  
Using Data ◽  
Demographic Features

Abstract Objective Attention networks learn an intelligent weighted averaging mechanism over a series of entities, providing increases to both performance and interpretability. In this article, we propose a novel time-aware transformer-based network and compare it to another leading model with similar characteristics. We also decompose model performance along several critical axes and examine which features contribute most to our model’s performance. Materials and methods Using data sets representing patient records obtained between 2017 and 2019 by the Kaiser Permanente Mid-Atlantic States medical system, we construct four attentional models with varying levels of complexity on two targets (patient mortality and hospitalization). We examine how incorporating transfer learning and demographic features contribute to model success. We also test the performance of a model proposed in recent medical modeling literature. We compare these models with out-of-sample data using the area under the receiver-operator characteristic (AUROC) curve and average precision as measures of performance. We also analyze the attentional weights assigned by these models to patient diagnoses. Results We found that our model significantly outperformed the alternative on a mortality prediction task (91.96% AUROC against 73.82% AUROC). Our model also outperformed on the hospitalization task, although the models were significantly more competitive in that space (82.41% AUROC against 80.33% AUROC). Furthermore, we found that demographic features and transfer learning features which are frequently omitted from new models proposed in the EMR modeling space contributed significantly to the success of our model. Discussion We proposed an original construction of deep learning electronic medical record models which achieved very strong performance. We found that our unique model construction outperformed on several tasks in comparison to a leading literature alternative, even when input data was held constant between them. We obtained further improvements by incorporating several methods that are frequently overlooked in new model proposals, suggesting that it will be useful to explore these options further in the future.

scholarly journals Model dependence in multi-model climate ensembles: weighting, sub-selection and out-of-sample testing

2018 ◽  
Author(s):  
Gab Abramowitz ◽  
Nadja Herger ◽  
Ethan Gutmann ◽  
Dorit Hammerling ◽  
Reto Knutti ◽  
...  
Keyword(s):  
Ad Hoc ◽  
Model Performance ◽  
Coupled Model ◽  
Data Sets ◽  
Model Code ◽  
Model Dependence ◽  
Out Of Sample ◽  
Large Ensembles ◽  
Sample Testing ◽  
Model Ensembles

Abstract. The rationale for using multi-model ensembles in climate change projections and impacts research is often based on the expectation that different models constitute independent estimates, so that a range of models allows a better characterisation of the uncertainties in the representation of the climate system than a single model. However, it is known that research groups share literature, ideas for representations of processes, parameterisations, evaluation data sets and even sections of model code. Thus, nominally different models might have similar biases because of similarities in the way they represent a subset of processes, or even be near duplicates of others, weakening the assumption that they constitute independent estimates. If there are near-replicates of some models, then treating all models equally is likely to bias the inferences made using these ensembles. The challenge is to establish the degree to which this might be true for any given application. While this issue is recognized by many in the community, quantifying and accounting for model dependence in anything other than an ad-hoc way is challenging. Here we present a synthesis of the range of disparate attempts to define, quantify and address model dependence in multi-model climate ensembles in a common conceptual framework, and provide guidance on how users can test the efficacy of approaches that move beyond the equally weighted ensemble. In the upcoming Coupled Model Intercomparison Project phase 6 (CMIP6), several new models that are closely related to existing models are anticipated, as well as large ensembles from some models. We argue that quantitatively accounting for dependence in addition to model performance, and thoroughly testing the effectiveness of the approach used will be key to a sound interpretation of the CMIP ensembles in future scientific studies.

Sign in / Sign up

Export Citation Format

Share Document