Find research datasets worth reusing

opencc-by-4.0Nov 2022View details →

Ecological Niche Models, in 2019 and across RCP 2.6, 4.5, and 8.5 scenarios in 2050 and 2100, of 1508 European Marine Species, developed with Artificial Neural Networks, Maximum Entropy, Support Vector Machines, and AquaMaps at 0.5° Resolution

Native ecological niche models of 1508 European species (894 fish and 614 non fish) developed with AquaMaps, Artificial Neural Networks, Maximum Entropy, and Support Vector Machines, for 2019 and under RCP 2.6, 4.5, and 8.5 scenarios in 2050 and 2100, at 0.5° spatial resolution.

opencc-by-4.0Dec 2022View details →

Biodiversity Index, in 2019 and across RCP 4.5, and 8.5 scenarios in 2050 and 2100 of 1508 European Marine Species based on ensemble Ecological Niche Models developed with Artificial Neural Networks, Maximum Entropy, Support Vector Machines, and AquaMaps at 0.5° Resolution

Biodiversity Index in 2019 and across RCP 4.5, and 8.5 scenarios in 2050 and 2100 of 1508 European marine species based on ensemble Ecological Niche Models developed with Artificial Neural Networks, Maximum Entropy, Support Vector Machines, and AquaMaps at 0.5° Resolution. The Index counts the number of species (among the 1508) potentially present in each 0.5° cell according to the ensemble models. For each ensemble model, a threshold of at least 3 models agreeing on species presence in the cell was used to indicate species presence.

opencc-by-4.0Nov 2022View details →

Ecological Niche Models of 96 European Marine Species, for 2019, developed with AquaMaps, Artificial Neural Networks, Maximum Entropy, and Support Vector Machines at 0.1° Resolution

Native ecological niche models of 96 European marine species of particular commercial and conservation interest developed with AquaMaps, Artificial Neural Networks, Maximum Entropy, and Support Vector Machines, for 2019 at 0.1° spatial resolution.

opencc-by-4.0Dec 2022View details →

Ensemble Ecological Niche Models and Biodiversity Index for 2019 of 96 European Marine Species based on Ecological Niche Models developed with Artificial Neural Networks, Maximum Entropy, AquaMaps, and Support Vector Machines at 0.1° Resolution

Ensemble Ecological Niche Models for 2019 of 96 European marine species of particular commercial and conservation interest, based on Ecological Niche Models developed with (i) Artificial Neural Networks, (ii) Maximum Entropy, (iii) Support Vector Machines, and (iv) AquaMaps at 0.1° Resolution. The data report, for each 0.1° cell, how many models (from 0 to 4) overcome a model-specific decision threshold to assess species presence in the cell. A Biodiversity Index is also provided as the count of the number of species (among the 96) potentially present in each 0.1° cell according to the ensemble models. For each ensemble model, a threshold of at least 3 models agreeing on species presence in the cell was used to indicate species presence.

edi48/100

Stream nitrate concentrations and discharge, stream nitrate uptake, and results of stream network nitrate model to determine lateral nitrate load from land to stream in Oak Creek, Arizona, USA

Data package associated with Handler et al. (2024) "Nitrate loads from land to stream are balanced by in-stream nitrate uptake across season in a dryland stream". The study describes the nitrate dynamics in Oak Creek watershed. Data include measurements from four seasonal synoptic sampling campaigns, nine seasonal stream nitrate uptake experiments on the main stem and tributaries, and the results of a network model that estimates the lateral load of nitrate from surrounding landscape to the stream network as well as network-level stream nitrate uptake and retention.

openCC0Oct 2024View details →

opencc-zeroAug 2014View details →

Supplement for "Using Phylogenetic Networks to Model Chinese Dialect History"

This is the supplementary material accompanying the paper "Using Phylogenetic Networks to Model Chinese Dialect History", which appeared in 2014 in "Language Dynamics and Change" (volume 4, issue 2).

openmit-licenseApr 2024View details →

Dataset and neural network weights to the paper: "Generative diffusion for regional surrogate models from sea-ice simulations"

All the needed code and data to reproduce the results from the paper: "Generative diffusion for regional surrogate models from sea-ice simulations". While most of the code is a frozen clone of the original <a href="https://github.com/cerea-daml/diffusion-nextsim-regional">Repository</a>, this capsule also includes the dataset and neural network weights to train and apply the surrogate models. The dataset for training and evaluation can be found at data/nextsim, which includes three different Zarr folders for training/validation/testing. The dataset is based on neXtSIM simulation data and ERA5 forcing data and extracted from the <a href="https://ige-meom-opendap.univ-grenoble-alpes.fr/thredds/catalog/meomopendap/extract/catalog.html">SASIP shared data OpenDAP server</a>: <ul> <li>The neXtSIM simulations were performed by Gauillaume Boutin and published in the paper "<a href="https://doi.org/10.5194/tc-17-617-2023">Arctic sea ice mass balance in a new coupled ice–ocean model using a brittle rheology framework</a>" (Boutin et al., 2023) and available as Zenodo <a href="../records/7277523">dataset</a> (Boutin et al., 2022).</li> <li>The forcing data is based on the ERA5 reanalysis dataset published in the paper: "<a href="https://doi.org/10.1002/qj.3803">The ERA5 global reanalysis</a>" (Hersbach et al., 2020) and available as dataset from the Copernicus Climate Change Service (C3S, Copernicus Climate Change Service, 2023). The here used forcing data is based on the <a href="https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels">hourly reanalysis data on single levels</a> and interpolated with nearest neighbors to the curvilinear grid as used in the output from the neXtSIM simulations. Disclaimer: The results contain modified Copernicus Climate Change Service information, 2023. Neither the European Commission nor ECMWF is responsible for any use that may be made of the Copernicus information or data it contains.</li> </ul> The neural network weights are included under data/models and split into weights for the deterministic models and the diffusion models. These neural network weights have been used to generate the results presented in the paper. In this capsule, the notebooks folder includes also the figures used within the paper and additional trajectory data used in the qualitative analysis of the paper. Generally, we recommend to just download the data.tar.gz file and use otherwise the original <a href="https://github.com/cerea-daml/diffusion-nextsim-regional">Repository</a>, since the here included code can be outdated. We further refer to the repository for additional information.   Contained in this capsule: <ul> <li>configs.tar.gz: The configuration files for the experiments.</li> <li>data.tar.gz: The dataset and neural network weights.</li> <li>diffusion_nextsim.tar.gz: The main code for the neural network etc.</li> <li>environment.yaml: The anaconda environment file, can be used to install the needed packages.</li> <li>notebooks.tar.gz: The notebooks that were used to create the figures in the paper. The figures from the paper and the data from the qualitative analysis are included as well.</li> <li>readme.md: The readme file from the repository.</li> <li>scripts.tar.gz: The scripts used for the experiments.</li> <li>setup.py: the file to install the diffusion_nextsim package in a python environment.</li> </ul> References: Guillaume Boutin, Heather Regan, Einar Ólason, Laurent Brodeau, Claude Talandier, Camille Lique, & Pierre Rampal. (2022). Data accompanying the article "Arctic sea ice mass balance in a new coupled ice-ocean model using a brittle rheology framework" (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7277523 Boutin, G., Ólason, E., Rampal, P., Regan, H., Lique, C., Talandier, C., Brodeau, L., and Ricker, R.: Arctic sea ice mass balance in a new coupled ice–ocean model using a brittle rheology framework, The Cryosphere, 17, 617–638, https://doi.org/10.5194/tc-17-617-2023, 2023. Copernicus Climate Change Service (2023): ERA5 hourly data on single levels from 1940 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), DOI: <a href="https://doi.org/10.24381/cds.adbb2d47">10.24381/cds.adbb2d47</a>. Hersbach H, Bell B, Berrisford P, et al. The ERA5 global reanalysis. Q J R Meteorol Soc. 2020; 146: 1999–2049. <a href="https://doi.org/10.1002/qj.3803">https://doi.org/10.1002/qj.3803</a>

opencc-zeroJun 2024View details →

scGraph2Vec: a deep generative model for gene embedding augmented by Graph Neural Network and single-cell omics data

This repository contains the training data and source code to reproduce the results of our paper: scGraph2Vec: a deep generative model for gene embedding augmented by Graph Neural Network and single-cell omics data More description can be also found in GitHub (https://github.com/LPH-BIG/scGraph2Vec).

opencc-by-4.0Nov 2024View details →

Deep learning models predicting gene functions and pathways using public DRKG knowledge graph and graph neural network

The attached dataset contains pretrained link prediction models, as described in our paper 'Morphological Map of Under- and Over-Expression of Genes in Human Cells'.

opencc-by-4.0Nov 2021View details →

Self-Supervised Representation Learning on Neural Network Weights for Model Characteristic Prediction - Datasets

Datasets to NeurIPS 2021 accepted paper "Self-Supervised Representation Learning on Neural Network Weights for Model Characteristic Prediction". Datasets are pytorch files containing a dictionary with training, validation and test sets. Train, validation and test sets are custom dataset classes which inherit from the standard torch dataset class. Corresponding code an be found at https://github.com/HSG-AIML/NeurIPS_2021-Weight_Space_Learning. Datasets 41, 42, 43 and 44 are our dataset format wrapped around the zoos from Unterthiner et al, 2020 (https://github.com/google-research/google-research/tree/master/dnn_predict_accuracy) Abstract: Self-Supervised Learning (SSL) has been shown to learn useful and information-preserving representations. Neural Networks (NNs) are widely applied, yet their weight space is still not fully understood. Therefore, we propose to use SSL to learn neural representations of the weights of populations of NNs. To that end, we introduce domain specific data augmentations and an adapted attention architecture. Our empirical evaluation demonstrates that self-supervised representation learning in this domain is able to recover diverse NN model characteristics. Further, we show that the proposed learned representations outperform prior work for predicting hyper-parameters, test accuracy, and generalization gap as well as transfer to out-of-distribution settings.

opencc-by-4.0Nov 2021View details →

Surrogate-based optimization using an artificial neural network for a parameter identification in a 3D marine ecosystem model

Abstract: Parameter identification for marine ecosystem models is important for the assessment and validation of marine ecosystem models against observational data. The surrogate-based optimization (SBO) is a computationally efficient method to optimize complex models. SBO replaces the computationally expensive (high-fidelity) model by a surrogate constructed from a less accurate but computationally cheaper (low-fidelity) model in combination with an appropriate correction approach, which improves the accuracy of the low-fidelity model. To construct a computationally cheap low-fidelity model, we tested three different approaches to compute an approximation of the annually periodic solution (i.e., a steady annual cycle) of a marine ecosystem model: firstly, a reduced number of spin-up iterations (several decades instead of millennia), secondly, an artificial neural network (ANN) approximating the steady annual cycle and, finally, a combination of the both approaches. Except for the low-fidelity model using only the ANN, the SBO yielded a solution close to the target and reduced the computational effort significantly. If an ANN approximating appropriately a marine ecosystem model is available, the SBO using this ANN as low-fidelity model presents a promising and computational efficient method for the validation.   Content: <ul> <li>SQLite database including the data of the different optimization runs</li> <li>Structure and weights of the used artificial neural network</li> <li>Tracer concentrations obtain from the high-fidelity model for the different optimization runs</li> </ul>

opencc-by-4.0May 2022View details →

Neural network dataset, script and model

This data was collected from several sources and compiled into a single text file called NN.txt Ten quiet and five disturbed days in each month from 2009 to 2019 were used to develop the dataset. These days were got from the list of International Q and D days accessed on the website of the World Data Centre of Geomagnetism, Kyoto (<a href="https://wdc.kugi.kyoto-u.ac.jp/qddays/index.html">https://wdc.kugi.kyoto-u.ac.jp/qddays/index.html</a>). The dataset has seventeen columns defined as Date, Year (Y), Day of the year (DOY), hour of the day (HH), cosine and sine components of the day of the year for annual variation (DAC and DAS), cosine and sine components of the day of the year for semi-annual variation (DSC and DSS), cosine and sine components of the hour for daily variation (HRC and HRS), geographical coordinates (Lat, Lon), geomagnetic coordinates (Glat, Glon),  Dst Index, F10.7 Index, TEC and sunspot number (SSN). The geographical coordinates were converted to geomagnetic coordinates using quasi dipole coordinates (Emmert et al., 2010; Richmond, 1995). Dst index, F10.7 index, and SSN were downloaded from the OmniWeb database (https://omniweb.gsfc.nasa.gov/form/dx1.html). Jason TEC data was downloaded from the FTP access of the CEDAR Madrigal database (http://cedar.openmadrigal.org/ftp/).  The data resolution was an 18-second interval.  It should be noted that any day that had a missing value was eliminated from the database. The data downloaded from Omni web was on a resolution of 1 hour and it was put at an 18-second interval by repeating the same value. The TEC data from the CEDAR Madrigal database is in a second interval and therefore it was averaged at an 18-second interval.  The dataset was trained with MATLAB software, the MATLAB script (NN script) is attached. The output was a Neural network model also attached

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - CIFAR10

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on CIFAR10. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "cifar_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained with small and large CNN models, in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - STL10 - Raw Datasets

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from STL10. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains the raw model zoos as collections of models (file names beginning with "cifar_"). Zoos are trained with small and large CNN models, in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). Due to the large filesize, the preprocessed datasets are hosted in a separate repository. The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - SVHN

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from SVHN. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "svhn_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - Fashion-MNIST

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from Fashion-MNIST. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "fmnist_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - MNIST

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from MNIST. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "mnist_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

opencc-by-4.0Mar 2020View details →

RRR/RAPID input and output files corresponding to "Underlying Fundamentals of Kalman Filtering for River Network Modeling"

Corresponding peer-reviewed publication This dataset corresponds to all the RRR/RAPID input and output files that were used in the study reported in: <ul> <li> Emery, C. M., C. H. David, K. M. Andreadis, M. J. Turmon, J. T. Reager, and J. M. Hobbs (2020), Underlying Fundamentals of Kalman Filtering for River Network Modeling, Journal of Hydrometeorology, 21, 453-474, DOI: 10.1175/JHM-D-19-0084.1. </li> </ul> When making use of any of the files in this dataset, please cite both the aforementioned article and the dataset herein.  Known bugs and limitations in this dataset or the associated manuscript. In the final version of the published manuscript, Figure 5a, Figure 5b, Figure 5c, and Figure SF1 are inaccurate.  The issue in these figures is that they were all prepared with an incorrect indexing relating observed and simulated discharge, hence observations at any one location were consistently being compared to simulations at another different location.  As a result, all values of "measured" discharge errors (i.e. Bias, STDE, and RMSE) are incorrect.  This issue did not affect the values of "estimated" errors, nor did it affect all values of the Nash-Sutcliffe effeciency that are presented.  The figures published in the manuscript can all be recreated using the files in which "BUG_DO_NOT_USE" was appended to the name.  Correct figures can also be created using corresponding file names that were not so appended.  Note that corrected versions of Figure 5a, Figure 5b, Figure 5c, and Figure SF1 all retain the same strong linear relationships that are discussed in the paper.  The slope of the daily discharge STDE trend initially reported as \(\alpha = 0.3876\) in Figure 5c changes to \(\alpha = 0.4507\) after correction.  The resulting value of the ideal inflation factor hence changes from \(I = {1 \over 0.3876} \approx 2.58\) to \(I = {1 \over 0.4507} \approx 2.22\). This updated ideal inflation factor has no impact on the conclusions reached in the manuscript because it remains closer to \(I = 2.58\) than to \(I = 1\) or \(I = 5\), i.e. the three values that were evaluated. Additionally, a faulty version 1.3.1 of the Python toolbox netCDF4 led to incorrect interpretation of _FillValue in which every data point of value greater than _FillValue was interpreted as masked. This created discrepancies in the following three files, which were updated between V1 and V2 of this dataset: "timeseries_rap_exp01.csv", "timeseries_rap_exp18.csv", and "stats_rap_exp18.csv". Faulty versions of the same files have "BUG_NETCDF4" appended to their names. Correct files have been recreated with file names that were not so appended.