Find research datasets worth reusing

opencc-by-4.0Jan 2021View details →

Bias-corrected monthly precipitation data over South Siberia for 1979-2019

Bias-Corrected Precipitation data over South Siberia (CPSS 1.2) contains monthly precipitation data for the area within the coordinates 50–65 N, 60–120 E for the period from January 1979 to December 2019. CPSS data were combined from monthly total precipitation data from ERA5 reanalysis European Centre for Medium-Range Weather Forecasts (Copernicus Climate Change…, 2017) and precipitation data records from ground weather stations (Il’in et al., 2013). The ERA5 data were scaled according to the derived scale coefficient. The linear scaling coefficient for each month and weather station were calculated and extrapolated to the study area using the ordinary kriging method. Data spatial resolution is 0.25° in the latitude and 0.25° in the longitude.  CPSS reproduces the spatial variability of precipitation more precisely than can be done from the weather station observation network. The CPSS dataset will be useful for the study of extreme precipitation events and allow for more accurate hydrologic risk assessment at a regional level based on climate model results. Data provided in NetCDF (Network Common Data Form) format. Copernicus Climate Change Service (C3S), 2017. ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate. Copernicus Climate Change Service Climate Data Store (CDS), Available at <a href="https://cds.climate.copernicus.eu/cdsapp#!/home">https://cds.climate.copernicus.eu/cdsapp#!/home</a> Il’yin, B.M., Bulygina, O.N., Bogdanova, E.G, Veselov, V.M. and Gavrilova, S.Y., 2013. Dataset of monthly precipitation totals, with the elimination of systematic errors of precipitation gauges. Available at  <a href="http://meteo.ru/data/506-mesyachnye-summy-osadkov-s-ustraneniem-sistematicheskikh-pogreshnostej-osadkomernykh-priborov">http://meteo.ru/data/506-mesyachnye-summy-osadkov-s-ustraneniem-sistematicheskikh-pogreshnostej-osadkomernykh-priborov</a>

opencc-by-4.0Mar 2024View details →

Collider Bias Correction for Multiple Covariates in GWAS Using Robust Multivariable Mendelian Randomization

This repository contains the data underlying the figures in paper "Collider Bias Correction for Multiple Covariates in GWAS Using Robust Multivariable Mendelian Randomization".     The file names and sheet names in the xlsx file indicate the corresponding figures of data.  The underlying data of manhattan plots and QQ plots are in text file. For other figures, the underlying data are in the spreadsheet. In each file, column names indicate the MVMR method used to obtain the result.  For example:  In text files: The abbreviation "mPC" refers to metabolomic principle components. beta_no_correction: the SNP effect estimate without bias correction. beta_cml or beta_MVMR_cml: the standard error of SNP effect estimate after the bias correction of MVMR-cML. SE_UVMR_cml: the standard error of SNP effect estimate after the bias correction of UVMR-cML. p_value_Egger or p_value_MVMR_Egger: the p-value of SNP effect estimate after the bias correction of MVMR-Egger regression. In the spreadsheet, column names follow the same style.  The GWAS data is also available. The column names follows the plink output file. The detailed explanations are available at https://www.cog-genomics.org/plink/2.0/formats#glm_linear

opencc-by-4.0Jan 2024View details →

Global Surface Ozone Concentration Dataset 1990-2017 Generated by Bayesian Maximum Entropy Data Fusion With RAMP Bias Correction

This dataset reports estimates of surface ozone concentration at fine spatial resolution for 1990 to 2017, at 0.5 degree horizontal resolution.  Also reported is the variance.  Estimates correspond to this paper: Becker, J. S., DeLang, M. N., K.-L. Chang, M. L. Serre, O. R. Cooper, H. Wang, M. G. Schultz, S. Schroder, X. Lu, L. Zhang, M. Deushi, B. Josse, C. A. Keller, J.-F. Lamarque, M. Lin, J. Liu, V. Marecal, S. A. Strode, K. Sudo, S. Tilmes, L. Zhang, M. Brauer, and J. J. West (2023) Using Regionalized Air Quality Model Performance and Bayesian Maximum Entropy data fusion to map global surface ozone concentration, Elementa Science of the Anthropocene, 11: 1, doi: 10.1525/elementa.2022.00025. The dataset reports estimates of surface ozone for the OSDMA8 metric (the 6-month ozone-season average of the daily maximum 8-hr concentration), estimated through a data fusion of ozone observations from the Tropospheric Ozone Assessment Report (TOAR) database, and output from multiple global atmospheric models.  Estimates are created in each year by a combination of M3Fusion to create a multi-model composite, Regional Air Quality Model Performance (RAMP) regional and nonlinear bias correction, and Bayesian Maximum Entropy (BME) data fusion in space and time.  The estimates here are the final results using a weighted RAMP bias correction.

opencc-by-4.0Feb 2022View details →

Bias-corrected EURO-CORDEX RCM simulations for the OPTAIN case studies

Bias-corrected EURO-CORDEX RCM simulations are available on a daily timescale for: -period 1981-2099/2100, -6 RCM, -3 scenarios (RCPs 2.6, 4.5 and 8.5), -7 variables (mean, minimum and maximum temperature, precipitation, solar radiation, wind speed at 2 m and relative humidity) and -18 domains and 23 locations within these domains. Bias correction and further downscaling to 0.1° was done using <a href="https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-land?tab=overview">ERA5-Land</a> reanalysis data with non-parametric empirical quantile mapping. Moreover, the interpolation of gridded bias-corrected climate model simulations to the locations was made using universal kriging. Organization of the data The name of the files are domain-type.zip, where type is gridded (NetCDF) or point (csv). Each zip file contains multiple files, organized in subfolders: experiment/modelNumber/variable.nc for gridded and experiment/modelNumber/variable-pilotFieldNumber.txt for point data, where experiment is rcp26, rcp45 or rcp85. domain and pilotFieldNumber <table> <tbody> <tr> <td> domain </td> <td> domain location (min and max. Longitude, min and max latitude) </td> <td> pilotFieldNumber </td> <td> pilot field location (longitude, latitude) </td> <td> case study number </td> <td> country </td> <td> Name (OPTAIN case study) </td> </tr> <tr> <td> 01 </td> <td> 50.95 51.45 14.55 15.05 </td> <td> </td> <td>   </td> <td> 1 </td> <td> DEU </td> <td> Schoeps </td> </tr> <tr> <td> 02 </td> <td> 46.35 47.05 6.55 7.15 </td> <td> 2 </td> <td> 46.816667 6.95 </td> <td> 2 </td> <td> CHE </td> <td> Petite Glane </td> </tr> <tr> <td> 02_1 </td> <td> 46.75 47.25 7.25 7.75 </td> <td> 1 </td> <td> 46.983333 7.466667   </td> </tr> <tr> <td> 02_34 </td> <td> 47.35 47.85 8.35 </td> <td> 3 4 </td> <td> 47.433333 8.516667 47.683333 8.616667 </td> </tr> <tr> <td> 02_5 </td> <td> 46.15 46.65 5.95 6.45 </td> <td> 5 </td> <td> 46.4 6.233333 </td> </tr> <tr> <td> 03a </td> <td> 46.65 47.15 17.45 17.95 </td> <td> 1 2 3 4 </td> <td> 46.92649 17.68246 46.9166 17.68976 46.91283 17.69754 46.91283 17.69723 </td> <td> 3a </td> <td> HUN </td> <td> Csorsza </td> </tr> <tr> <td> 03b </td> <td> 46.45 46.95 16.65 17.15 </td> <td> </td> <td>   </td> <td> 3b </td> <td> HUN </td> <td> Felso Valicka </td> </tr> <tr> <td> 04 </td> <td> 52.35 52.85 18.45 18.95 </td> <td> 1 </td> <td> 52.597469 18.728617 </td> <td> 4 </td> <td> POL </td> <td> Upper Zglowiaczka </td> </tr> <tr> <td> 05 </td> <td> 46.35 46.85 15.35 15.85 </td> <td> </td> <td>   </td> <td> 5 </td> <td> SVN </td> <td> Pesnica </td> </tr> <tr> <td> 06 </td> <td> 46.45 46.95 16.15 16.65 </td> <td> </td> <td>   </td> <td> 6 </td> <td> HUN/SVN </td> <td> Kebele/Kobiljski </td> </tr> <tr> <td> 07 </td> <td> 49.85 50.35 4.75 5.25 </td> <td> </td> <td>   </td> <td> 7 </td> <td> BEL </td> <td> La Wimbe </td> </tr> <tr> <td> 08 </td> <td> 55.15 55.75 23.55 24.05 </td> <td> 1 2 </td> <td> 55.522057 23.799235 55.42233194 23.82580339 </td> <td> 8 </td> <td> LTU </td> <td> Dotnuvele </td> </tr> <tr> <td> 09 </td> <td> 45.45 45.95 9.65 10.15 </td> <td>   </td> <td>   </td> <td> 9 </td> <td> ITA </td> <td> Cherio </td> </tr> <tr> <td> 10 </td> <td> 59.45 59.95 10.75 11.25 </td> <td> 1 2 3 4 5 6 7 8 </td> <td> 59.71949 10.83576 59.6833306 10.8833298 59.6833306 10.8833298 59.665 10.9475 59.665 10.9475 59.841012 10.903597 59.757631 11.072031 59.539623 10.856447 </td> <td> 10 </td> <td> NOR </td> <td> Krogstad </td> </tr> <tr> <td> 11 </td> <td> 46.45 46.95 17.55 18.05 </td> <td> 1 2 </td> <td> 46.658333 17.75583 46.656944 17.75833 </td> <td> 11 </td> <td> HUN </td> <td> Tetves </td> </tr> <tr> <td> 12 </td> <td> 49.35 49.85 14.75 15.25 </td> <td> 1 </td> <td> 49.616837 15.078266 </td> <td> 12 </td> <td> CZE </td> <td> Cechticky </td> </tr> <tr> <td> 13 </td> <td> 55.85 56.35 25.85 26.45 </td> <td>   </td> <td>   </td> <td> 13 </td> <td> LVA </td> <td> Dviete </td> </tr> <tr> <td> 14 </td> <td> 59.75 60.25 17.55 18.05 </td> <td>   </td> <td>   </td> <td> 14 </td> <td> SWE </td> <td> Ingvastaan Lehstaan </td> </tr> </tbody> </table>   modelNumber <table> <tbody> <tr> <td> modelNumber </td> <td> Driving Model (GCM) </td> <td> Ensemble </td> <td> RCM </td> <td> End date </td> </tr> <tr> <td> 1 </td> <td> EC-EARTH </td> <td> r12i1p1 </td> <td> CCLM4-8-17 </td> <td> 31.12.2100 </td> </tr> <tr> <td> 2 </td> <td> EC-EARTH </td> <td> r3i1p1 </td> <td> HIRHAM5 </td> <td> 31.12.2100 </td> </tr> <tr> <td> 3 </td> <td> HadGEM2-ES </td> <td> r1i1p1 </td> <td> HIRHAM5 </td> <td> 30.12.2099 </td> </tr> <tr> <td> 4 </td> <td> HadGEM2-ES </td> <td> r1i1p1 </td> <td> RACMO22E </td> <td> 30.12.2099 </td> </tr> <tr> <td> 5 </td> <td> HadGEM2-ES </td> <td> r1i1p1 </td> <td> RCA4 </td> <td> 30.12.2099 </td> </tr> <tr> <td> 6 </td> <td> MPI-ESM-LR </td> <td> r2i1p1 </td> <td> REMO2009 </td> <td> 31.12.2100 </td> </tr> </tbody> </table>   variable <table> <tbody> <tr> <td> variable </td> <td> description </td> <td> Unit </td> </tr> <tr> <td> Tmean </td> <td> Mean temperature </td> <td> °C </td> </tr> <tr> <td> Tmin </td> <td> Min temperature </td> <td> °C </td> </tr> <tr> <td> Tmax </td> <td> Max temperature </td> <td> °C </td> </tr> <tr> <td> prec </td> <td> Precipitation </td> <td> mm </td> </tr> <tr> <td> solarRad </td> <td> Solar radiation </td> <td> MJ/m2 </td> </tr> <tr> <td> windSpeed </td> <td> Wind speed at 2m </td> <td> m/s </td> </tr> <tr> <td> relHum </td> <td> Relative humidity </td> <td> % </td> </tr> </tbody> </table>   Methodolody Bias correction was done using non-parametric empirical quantile mapping with modified method from R package <a href="https://cran.r-project.org/web/packages/qmap/index.html">qmap</a>. Parameters selected were: corrections for each day of the year using a moving windows for a 31 days; 100 quantiles; wet days corrections for precipitation. The reference period is 1981-2010. The interpolation of gridded bias-corrected climate model simulations to the location was made using universal kriging  with R packages <a href="https://cran.r-project.org/web/packages/automap/index.html">automap</a> and <a href="https://cran.r-project.org/web/packages/gstat/index.html">gstat</a> with (external) variables x, y, x2, y2, x*y, z, where x is latitude, y is longitude, and z is elevation. For Digital Elevation Model <a href="https://webmap.ornl.gov/wcsdown/dataset.jsp?dg_id=10008_1">Shuttle Radar Topography Mission</a> was used. If there was an error using above mentioned variables, the number of variables was reduced to x, y, x*y, z and if there was still an error to x, y, z.   Funding This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 862756.

opencc-by-4.0Jul 2024View details →

RemoTeC full-physics retrieval GOSAT/TANSO-FTS Level 2 bias-corrected XCO2 2009-2023, version 2.4.1 operated at Heidelberg University

The data set contains bias-corrected column averaged dry air mole fractions (XCO2) retrieved with the RemoTeCv2.4.1 full-physics algorithm (Butz et al. 2011, Guerlet et al. 2013) applied on GOSAT TANSO-FTS Level 1B (L1B) data from 2009-04-18 to 2023-08-29. The GOSAT TANSO-FTS L1B data product is produced by JAXA/NOIES/MOE and provided by ESA. The XCO2 data together with related variables are aggregated as daily files, only good quality retrievals are included. If the data is used for publications, please contact andre.butz@uni-heidelberg.de to discuss potential co-authorship and technical details.   Summary: Shortname: REMOTEC_L2_CO2_GOSAT Longname: RemoTeC full-physics retrieval GOSAT/TANSO-FTS Level 2 bias-corrected XCO2 version 2.4.1 DOI: 10.5281/zenodo.12773070 Version: 2.4.1 Format: netCDF Spatial Coverage: -180.0,-90.0,180.0,90.0 Temporal Coverage: 2009-04-18 to 2023-08-29

opencc-by-4.0Aug 2024View details →

Phantom measurement data for 'Fast bias-corrected conductivity mapping using stimulated echoes', Iyyakkunnel et al. (2024)

This dataset contains the phantom measurement data used in the article by Iyyakkunnel et al., titled "Fast Bias-Corrected Conductivity Mapping Using Stimulated Echoes," published in MAGMA, 2024 (doi: 10.1007/s10334-024-01194-3). In this study, the feasibility of using a stimulated echo sequence for electrical properties tomography (EPT) is demonstrated. The data were acquired with a 3T MRI system (Magnetom Prisma; Siemens Healthcare, Erlangen, Germany) using a dual-tuned 1H/23Na quadrature head coil for transmission and reception (Rapid Biomedical, Rimpar, Germany). The dataset includes magnitude and phase measurements for the proposed Double-Angle Stimulated Echo (DA-STE) sequence, as well as reference measurements, including Double Angle measurements using a Gradient Echo sequence (GRE-DAM) for the B1+ magnitude, and a Single Echo Spin Echo sequence (SE) for the transceive phase. For both the DA-STE and SE sequences, each measurement was repeated with inverted readout gradient polarities, denoted as LR (left-right) and RL (right-left) in the respective measurement folders. For each measurement, magnitude and phase data are provided in separate folders (in dicom (.dcm) format). Please note that for DA-STE, the two echo acquisitions are sequentially stored in the same measurement folder. For further measurement details, please refer to the mentioned original article.

opencc-by-4.0Dec 2021View details →

Precipitation oxygen isoscape for mainland China from 1870 to 2017 generated based on data fusion and bias correction of iGCMs simulations

The dataset includes the stable oxygen isotope of precipitation for the mainland of China over the 1870-2017 period, at a spatial resolution of 50-60 km and a monthly temporal resolution. In order to make full use of observations to integrate the advantages of various iGCMs, the combination of data fusion and bias correction methods are used. Some physical-based ancillary data are introduced in the fusion methods, including elevation and meteorological data, to enrich the climate and terrain information in the process of data fusion. Specifically,(1) for the 1979-2001 period, nine simulations from six iGCMs (CAM2, GISS E, HadAM3, IsoGSM2, LMDZ4, and MIROC32) and ancillary data are fused with observations by using the CNN fusion method;(2) for the 2002-2007 period, seven simulations from four iGCMs (GISS E, IsoGSM2, LMDZ4, and MIROC32) and ancillary data are fused by using the CNN fusion method;(3) for the 1969-1978 period, four simulations from three iGCMs (CAM2, GISS E, and HadAM3) and ancillary data are fused by using the CNN fusion method;(4) for the 1958-1968 and 2008-2017 periods, two iGCM simulations (CAM2 and HadAM3 for 1958-1968 and IsoGSM2 and LMDZ4 zoomed for 2008-2017) are corrected by using two BCMs, and ensemble mean (mean of four simulations) is then calculated;(5) for the 1870-1957 period, one iGCM simulation (HadAM3) is corrected by using two BCMs, and the ensemble mean (mean of two simulations) is then calculated.Compared with the existing iGCMs, the isoscape has high quality and stability for a large region in China at the monthly scale. However, it should be noted that the isoscape may be more reliable for the common periods of most iGCMs (1969-2007), but mediocre for other periods.

opencc-by-sa-4.0Aug 2024View details →

Bias Correction of CRCM5-LE for Hydrological Bavaria

The frequency and intensity of extreme hydrometeorological events are anticipated to rise as a result of climate change. For precise analysis, especially in low-flow assessments, it is crucial to have data on precipitation and temperature with high spatial and sub-daily resolution. However, such data is often lacking in both density and duration. The <a href="https://www.climex-project.org/">ClimEx-II </a>project (Climate Change and Hydrological Extreme Events 2nd Phase) is dedicated to enhancing our understanding of these shifts in hydrological extremes. The Canadian Regional Climate Model version 5 Large Ensemble (CRCM5-LE; Leduc et al. (2019)) under RCP 8.5 builds the climatic boundary conditions for the hydrological modelling. The ensemble covers a European and a North American domain, each comprising 50 members from 1951 to 2100. The SDCLIREF v2 (Lehr- und Forschungseinheit für physische Geographie und komplexe Umweltsysteme 2024),  a sub-daily (3h), high-resolution (500m) data set for the domain of Bavaria and hydrologically important neighbouring catchments marks the reference data set.  In ClimEx-II, bias correction is a crucial step before downscaling (regional) climate model simulations to higher resolutions as it adjusts local inconsistencies in the climate model. The corrected and downscaled meteorological inputs can be used to drive a hydrological model (Emami and Koch 2018,  Fang et al. 2015). The quality of the corrected data depends on the method used. Therefore, this data set comprises a comparison of the input and output data from three different bias correction methods, UBC (Cannon et al., 2015), MBCn (Cannon 2018) and VBC (Funk et al., 2024) for three diverse climate regions in Bavaria: <ul> <li>Fränkische Saale Salz is a franconian catchment</li> <li>Iller Kempten is a pre-alpine catchment</li> <li>Hart an der Ziller is an alpine catchment</li> </ul> Each catchment comprises six to seven grid cells of a 12 km resolution. Five climate variables of hydrological importance are corrected in a 3-hourly temporal resolution per grid cell: <ul> <li>Near-Surface Dewpoint Temperature in °C (dew)</li> <li>Precipitation in kg/m2 (pr)</li> <li>Surface Downwelling Shortwave Radiation in W/m2 (rsds)</li> <li>Near-Surface Wind Speed in m/s (sfcWind)</li> <li>Near-Surface Air Temperature in °C (tas)</li> </ul> The environment in each file comprises the inputs to the bias correction <ul> <li>mp_dts: CRCM5-LE model data during the projection period (2011-2030) before correction</li> <li>mc_dts: CRCM5-LE model data during calibration period (1991-2010)</li> <li>oc_dts: SDCLIREF v2 reference data during calibration period (1991-2010)</li> </ul> ,the outputs from the bias correction comparison during the projection period <ul> <li>vbc: CRCM5-LE model data during the projection period (2011-2030) after correction by VBC</li> <li>mbcn: CRCM5-LE model data during the projection period (2011-2030) after correction by MBCn</li> <li>ubc: CRCM5-LE model data during the projection period (2011-2030) after correction by UBC</li> </ul> and the held-out reference data for validation <ul> <li>op_dts: SDCLIREF v2 reference data during the projection period (2011-2030).</li> </ul> Each of the above-presented data sets consists of six columns. The five climate variables are indexed by their abbreviations. The sixth column time contains a string marking the respective timestamp. All CRCM5-LE model data contain a seventh column indicating the respective ensemble member. The evaluation results by Wasserstein Distance and Model Correction Inconsistency from Funk et al. (2024) are captured in the three additional files starting with 06_*.

opencc-by-4.0Nov 2023View details →

Determining non-significant bits on a C++ implementation of the LeNet-5 convolutional neural network to be used for storing error correcting codes to protect weights and biases. Robustness assessment of the network after integrating the proposed codes.

The architecture of the LeNet-5 convolutional neural network (CNN) was defined by LeCun in its paper "Gradient-based learning applied to document recognition" (<a href="https://ieeexplore.ieee.org/document/726791">https://ieeexplore.ieee.org/document/726791</a>) to classify images of hand written digits (MNIST dataset).This architecture has been customized to use Rectified Linear Unit (ReLU) as activation functions instead of Sigmoid.It consists of the following layers:<ul><li>conv1: Convolution 2D, 1 input channel (28x28), 3 output channels (28x28), kernel size 5, stride 1, padding 2.</li><li>relu1: Rectified Linear Unit (3@28x28).</li><li>max1: Subsampling buy max pooling (3@14x14).</li><li>conv2: Convolution 2D, 3 input channels (14x14), 6 output channels (14x14), kernel size 5, stride 1, padding 2.</li><li>relu2: Rectified Linear Unit (6@14x14).</li><li>max2: Subsampling buy max pooling (6@7x7).</li><li>fc1: Fully connected (294, 147)</li><li>fc2: Fully connected (147, 10)</li></ul>The fault hypotheses for this work include the occurrence of:<ul><li>S0/S1: multiple adjacent stuck-at-0 and stuck-at-1 faults to determine the least significant bits of weights and biases that could be used to store the proposed error correcting codes.</li><li>BF: single, double, and triple bit-flip faults to assess the robustness of the considered CNN</li></ul>In the memory cells containing all the parameters of the CNN:  <ul><li>w: weights (float32)</li><li>b: biases (float32)</li></ul>All the images (10000) from the MNIST dataset have been used as workload.The weights and biases of the LeNet-5 architecture have been protected using six different error correcting codes that have been deployed in the least significant bits of these elements.The parity check matrices (H = P I) that define these ECCs are:<ul><li>SEC(32, 26) (Hamming) under a classic policy (see methodology below):</li></ul>        11010010001000011101101000 100000        10101001000100011011010100 010000        01100100100010010110110010 001000        00011100010001001110001101 000100        00000011110000100001111011 000010        00000000001111100000000111 000001<ul><li>SEC(23, 18) (Hamming) under a conservative policy (see methodology below):</li></ul>        111100001111000000 10000        110011101000111000 01000        101011010100100110 00100        010110110010010101 00010        001101110001001011 00001<ul><li>SEC(13, 9) (Hamming) under an aggressive policy (see methodology below):</li></ul>        110111000 1000        101100110 0100        011010101 0010        111001011 0001<ul><li>DEC(32, 21) (low redundancy and reduced overhead DEC) under a classic policy (see methodology below):</li></ul>        111000011001010010000 10000000000        110110000011101000000 01000000000        101011000110000010001 00100000000        100101101000110001000 00010000000        011010101100100000100 00001000000        010101010100001001010 00000100000        001100110010010100100 00000010000        000011110001000110010 00000001000        000000001111001101001 00000000100        000000000000111100111 00000000010        000000000000000011111 00000000001<ul><li>DEC(28, 18) (low redundancy and reduced overhead DEC) under a conservative policy (see methodology below):</li></ul>        111111000000000000 1000000000        110100111100000000 0100000000        110000100011110000 0010000000        001110010011001100 0001000000        101100001010101010 0000100000        010001001101010110 0000010000        001011000101101001 0000001000        101000011000110101 0000000100        010001110000011011 0000000010        000010100110000111 0000000001<ul><li>DEC(17, 9) (low redundancy and reduced overhead DEC) under an aggressive policy (see methodology below):</li></ul>        111110000 10000000        111001100 01000000        110101010 00100000        101010110 00010000        101101001 00001000        100110101 00000100        100011011 00000010        110000111 00000001This dataset contains the raw data obtained from:<ul><li>running exhaustive fault injection campaigns for increasingly multiple stuck-at faults in the least significant bits of all weights and biases (simultaneously) and for all the images in the workload.</li><li>running statistical fault injection campaigns for single, double, and triple bit-flip faults, randomly targeting the considered locations and images in the workload.</li></ul><h3>Files information</h3><ul><li>no_ecc folder: Results obtained for the original (not protected) version of the CNN.<ul><li>golden_run.csv: Prediction obtained for all the images considered in the workload in the absence of faults (Golden Run). This is intended to act as oracle to determine the impact of injected faults.</li><li>sampling_SBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for single bit-flip faults.</li><li>sampling_DBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for double bit-flip faults.</li><li>sampling_TBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for triple bit-flip faults.</li><li>locating_sensitive_bits folder: Prediction obtained for all the images considered in the workload in presence of stuck-at-0/stuck-at-1 faults that simultaneously target the N least significant bits of all weights and biases. There is one file for each parameter of type of fault and range of targeted bits. Files for bits in the range [11, 0] are not included as they obtain eactly the same results as the Golden Run (faults do not alter the behaviour of the network).</li></ul></li><li>sec/classic, sec/conservative, and sec/aggressive folders: They contain the results obtained for the CNN protected by SEC(32, 26), SEC(23, 18), and SEC(13, 9), respectively.<ul><li>golden_run.csv: Prediction obtained for all the images considered in the workload in the absence of faults (Golden Run). This is intended to act as oracle to determine the impact of injected faults. It must be noted that this file could be different that the golden_run.csv file for the original version of the CNN, as deploying the ECC in the weights and biases may have affected the behaviour of the network.</li><li>sampling_SBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for single bit-flip faults. They should all be tolerated by the definition of the ECC.</li><li>sampling_DBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for double bit-flip faults. They could be more harmful than for the unprotected version of the CNN, as the ECC may erroneously flip correct bits.</li></ul></li><li>dec/classic, dec/conservative, and dec/aggressive folders: They contain the results obtained for the CNN protected by DEC(32, 21), DEC(28, 18), and DEC(17, 9), respectively.<ul><li>golden_run.csv: Prediction obtained for all the images considered in the workload in the absence of faults (Golden Run). This is intended to act as oracle to determine the impact of injected faults. It must be noted that this file could be different that the golden_run.csv file for the original version of the CNN, as deploying the ECC in the weights and biases may have affected the behaviour of the network.</li><li>sampling_DBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for double bit-flip faults. They should all be tolerated by the definition of the ECC.</li><li>sampling_TBF_10000.csv: Prediction obtained for running 10000 statistical fault injection experiments for triple bit-flip faults. They could be more harmful than for the unprotected version of the CNN, as the ECC may erroneously flip correct bits.</li></ul></li></ul><h3>Methodology information</h3>First, the CNN was used to classify all the images of the workload in the absence of faults to get a reference to determine the impact of faults. This is golden_run.csv file.To locate non-significant bits in weights and biases, fault injection experiments were executed targeting all elements of all parameters of the CNN using the following procedure:<ul><li>The initial mask targeted only the least significant bit</li><li>Until the mask targets all bits of the elements (32 bits as they are single-precision floating point values):<ul><li>Affect the bits (setting them to 0 or 1 in case of stuck-at-0 or stuck-at-1 faults) identified by the mask for all elements of all parameters.</li><li>Classify all the images of the workload in the presence of this fault. The obtained output was stored in a given .csv file.</li><li>Remove the fault from the CNN by restoring the affected bits to its previous value.</li><li>Add the next adjacent bit to the mask, so it targets an additional least significant bit.</li></ul></li></ul>The analysis of the obtained results may help in determining which bits can be used to store an ECC:<ul><li>which bits never affect the behaviour of the CNN, as the predicted classification is exactly the same than in the absence of faults.</li><li>which bits midly affect the behaviour of the CNN, as although the predicted classifications differ from those in the absence of faults, the accuracy of the network is barely affected.</li><li>which bits greatly affect the behaviour of the CNN, as the accuracy of the network is significantly affected.</li></ul>Accordingly, three different policies have been identified for deploying an ECC using these bits:<ul><li>Classic policy: The ECC protects as much bits as possible.</li><li>Conservative policy: The ECC protects all those bits that may affect the prediction of the network.</li><li>Aggressive policy: The ECC protects only those bits that significantly affect the accuracy of the network.</li></ul>After designing and deploying a single ECC and a double ECC for each of the identified policies, fault injection experiments were executed to verify their behaviour in the presence of faults.Single and double ECCs were tested against single and double bit-flip, respectively (all faults should be tolerated,) and double and triple bit-flips, respectively (a correct bit could be erroneously flipped.)Due to the heavy computational load of the decoders, statistical injection was used to run the required fault injection campaigns with a sample size (number of experiments) of 10000.Each experiment consisted in:<ul><li>Randomly selecting the image to process, and the parameter, element, and bits (mask) to be targeted by the fault.</li><li>Affecting the bits (inverting them) identified by the mask.</li><li>Classifying the selected image of the workload in the presence of this fault. The obtained output was stored in a given .csv file.</li><li>Removing the fault from the CNN by restoring the affected bits to its previous value.</li></ul><h3>List of variables (Name : Description (Possible values))</h3><ul><li>IMGID: Integer number identifying the considered image (1-9999).</li><li>TENSORID: Integer number identiying the parameter affected by the fault (0 - No fault, 1 - conv1.w, 2 - conv1.b, 3 - conv2.w, 4 - conv2.b, 5 - fc1.w, 6 - fc1.b, 7 - fc2.w, 8 - fc2.b).</li><li>ELEMID: Integer number identiying the element of the parameter affected by the fault (-1 - No fault, [0-2] - conv1.b, [0-74] - conv1.w, [0-5] - conv2.b, [0-149] - conv2.w, [0-146] - fc1.b, [0-43217] - fc1.w, [0-9] - fc2.b, [0-1469] - fc2.w).</li><li>MASK: 8-digit hexadecimal number identifying those bits affected by the fault ([00000000 - No fault, FFFFFFFF - all 32 bits faulty]).</li><li>FAULT: String identiying the type of fault (NF - No fault, BF - bit-flip, S0 - Stuck-at-0, S1 - Stuck-at-1).</li><li>SOFTMAX: 10 decimal numbers obtained after applying the softmax function to the provided output. They represent the probability of the image of belonging to the corresponding category for classification.</li><li>PRED: Integer number representing the category predicted for the processed image.</li><li>LABEL: integer number representing the actual category for the processed image.</li></ul>

opencc-by-nc-4.0Apr 2022View details →

Bias-corrected CORDEX daily precipitation dataset for the Carpathian Region

This dataset contains bias-corrected regional climate model (RCM) daily outputs for daily precipitation under the RCP8.5 scenario. The reference dataset is CARPATCLIM (Szalai et al., 2013) which covers the the Carpathian Region for the period 1961-2010. The dataset contains bias corrected daily outputs of the following high-resolution (0.11o) RCMs from the framework of EURO-CORDEX (Jacob et al., 2014) and Med-CORDEX (Ruti et al., 2016): - ALADIN - CCLM - HIRHAM - RACMO - RCA4 - RegCM - REMO - WRF   The dataset covers the following periods with grid spacing of 0.11o on a regular lon/lat grid (between latitudes 44°N and 50°N, and longitudes 17°E and 27°E): - 1976-2005 - 2021-2050 - 2070-2099   File format: NetCDF All data have been created following the work of Mezghani et al. (2017). Using the dataset please cite the following reference paper (also further details are given there): <a href="https://doi.org/10.28974/idojaras.2020.1.2">https://doi.org/10.28974/idojaras.2020.1.2</a>.   References: Jacob, D., Petersen, J., Eggert, B., Alias, A., Christensen, O.B., Bouwer, L.M., Braun, A., Colette, A., Déqué, M., Georgievski, G., Georgopoulou, E., Gobiet, A., Menut, L., Nikulin, G., Haensler, A., Hempelmann, N., Jones, C., Keuler, K., Kovats, S., Kröner, N., Kotlarski, S., Kriegsmann, A., Martin, E., van Meijgaard, E., Moseley, C., Pfeifer, S., Preuschmann, S., Radermacher, C., Radtke, K., Rechid, D., Rounsevel, M., Samuelsson, P., Somot, S., Soussana, J.-F., Teichmann, C., Valentini, R., Vautard, R., Weber, B. and Yiou, P. (2014) EURO-CORDEX New high resolution climate change projections for European impact research. Reg. Environ. Change, 14, 563–578. https://doi.org/10.1007/s10113-013-0499-2 Mezghani, A., Dobler, A., Haugen, J.E., Benestad, R.E., Parding, K.M., Piniewski, M., Kardel, I. and Kundzewicz, Z.W. (2017) CHASE-PL Climate Projection dataset over Poland – bias adjustment of EURO-CORDEX simulations. Earth Syst. Sci. Data, 9, 905–925. https://doi.org/10.5194/essd-9-905-2017 Ruti, P.M., Somot, S., Giorgi, F., Dubois, C., Flaounas, E., Obermann, A., Dell'Aquila, A., Pisacane, G., Harzallah, A., Lombardi, E., Ahrens, B., Akhtar, N., Alias, A., Arsouze, T., Aznar, R., Bastin, S., Bartholy, J., Béranger, K., Beuvier, J., Bouffies-Cloché, S., Brauch, J., Cabos, W., Calmanti, S., Calvet, J.-C., Carillo, A., Conte, D., Coppola, E., Djurdjevic, V., Drobinski, P., Elizalde-Arellano, A., Gaertner, M., Galán, P., Gallardo, C., Gualdi, S., Goncalves, M., Jorba, O., Jordi, G., L'Heveder, B., Lebeaupin-Brossier, C., Li, L., Liguori, G., Lionello, P., Maciás, D., Nabat, P., Onol, B., Raikovic, B., Ramage, K., Sevault, F., Sannino, G., Struglia, M.V., Sanna, A., Torma, C. and Vervatis, V. (2016) MED-CORDEX initiative for Mediterranean climate studies. Bulletin of the American Meteorological Society, 97, 1187–1208. https://doi.org/10.1175/BAMS-D-14-00176.1 Szalai, S., Auer, I., Hiebl, J., Milkovich, J., Radim, T., Stepanek, P., Zahradnicek, P., Bihari, Z., Lakatos, M., Szentimrey, T., Limanowka, D., Kilar, P., Cheval, S., Deak, Gy., Mihic, D., Antolovic, I., Mihajlovic, V., Nejedlik, P., Stastny, P., Mikulova, K., Nabyvanets, I., Skyryk, O., Krakovskaya, S.,Vogt, J., Antofie, T. and Spinoni, J. (2013) Climate of the Greater Carpathian Region. Final Technical Report. http://www.carpatclim-eu.org

opencc-by-4.0Jun 2022View details →

Downscaled and bias corrected 10 km water withdrawal of China, 1981-2010

This dataset contains (1) MonthlyDomIndcon_CN_0.1deg_1981-2010.nc -- the 0.1° gridded monthly domestic & industrial water consumption of China (2) MonthlyDomInduse_CN_0.1deg_1981-2010.nc -- the 0.1° gridded monthly domestic & industrial water withdrawal of China (3) MonthlyIrrigation_CN_0.1deg_1981-2010.nc --  the 0.1° gridded monthly irrigation water withdrawal of China These data are first spatially downscaled from Huang et al. (2018) (https://doi.org/10.5281/zenodo.1209296) based on population density, GDP and irrigation area, and then bias corrected against provincial-level statistics published by local water agencies. Refer to Huang et al. (2018) and Dong et al. (2022) for more details. Dong, N., Wei, J., Yang, M., Yan, D., Yang, C., Gao, H., et al. (2022). Model estimates of China's terrestrial water storage variation due to reservoir operation. Water Resources Research, 58, e2021WR031787. <a href="https://doi.org/10.1029/2021WR031787">https://doi.org/10.1029/2021WR031787</a>

opencc-by-sa-4.0Oct 2018View details →

Bias correction of simulated Brazilian wind power generation based on reanalysis data

Available data: - Brazilian wind power generation time series derived from MERRA-2 reanalysis data with wind speed and wind power bias correction. - Wind speed correction factors derived from INMET wind speeds (http://www.inmet.gov.br/portal/) as well as wind power correction factors dervied from ONS wind power generation time series are also provided. - Simulation of about 38 years of wind power generation with fixed capacity. Data used for validation: - Historical wind power generation data, which were used for validation of simulated time series, can be found at the ONS homepage (http://ons.org.br/Paginas/resultados-da-operacao/historico-da-operacao/geracao_energia.aspx).   Other Links: - Information on this will soon be found here: https://refuel.world/ - Code for generating time series, validation and analysis: https://github.com/KatharinaGruber/BrazilWind - Master thesis belonging to data: https://doi.org/10.5281/zenodo.1471221

opencc-by-4.0Sep 2019View details →

Bias-corrected simluated wind power generation time series for Brazil

Simulated and bias corrected wind power generation time series data sets for Brazil, its North-East and South, seven states and seven wind parks. The data sources, generation and validation of the datasets are described in the article "Assessing the Global Wind Atlas and local measurements for bias correction of wind power generation simulated from MERRA-2 in Brazil", preprint available on arXiv: arxiv.org/abs/1904.13083, final version DOI: <a href="https://doi.org/10.1016/j.energy.2019.116212">10.1016/j.energy.2019.116212</a> Code for generating the datasets is available at github.com/KatharinaGruber/BrazilWindpower_biascorr   The files "comp_*" contain comparisons of simulated and observed wind power generation time series with daily resolution for all regions. "comp_noc.RData" is for comparison of interpolation methods and contains time series generated with Nearest Neighbour interpolation (NN), Bilinear Interpolation (BLI) and Inverse Distance Weighting (IDW). "comp_wmsa.RData" is for comparison of wind speed mean approximation methods and contains time series generated with Nearest Neighbour interpolation (NN - no correction applied), mean approximation with measured data (IN) and mean approximation with the Global Wind Atlas (GWA). "comp_wsc.RData" is for comparison of spatiotemporal wind speed correction methods and contains time series generated with mean approximation with the Global Wind Atlas (wmsa) and combined mean approximation with the Global Wind Atlas and hourly and monthly mean approximation with measured data (wschm).   The files "statpowlist_*" contain hourly simulated wind power generation time series for three interpolation methods (NN, BLI, IDW), two mean approximation methods (wsmaIN - measured data (INMET), wsmaWA - Global Wind Atlas) as well as for spatiotemporal (hourly and monthly) wind speed bias correction (wschm) for each wind park available in The Wind Power dataset.

FIGURE 5 in A simulation-based examination of residual diversity estimates as a method of correcting for sampling bias

FIGURE 5. The performance of different implementations of the residual diversity estimate when a specific bias is forced to be the dominant influence. (5.1) Mean Spearman's rho values of four implementations of the RDE using the Smith and McGowan method. PFORM and PTAPH are set at 0.9 to minimise their influence, PLOC is variable. PMIST set at 0.1. (5.2) Mean Spearman's rho values of four implementations of the RDE using the Smith and McGowan method. LOC and PTAPH are set at 0.9 to minimise their influence, PFORM is variable. PMIST set at 0.1. The dashed red line indicates the critical value at p=0.05. Abbreviations as in Table 1.

FIGURE 7 in A simulation-based examination of residual diversity estimates as a method of correcting for sampling bias

FIGURE 7. Sample simulation comparing the results of the taxic, phylogenetic and residual diversity estimates to the true diversity. PFROM, PLOC and PTAPH set at 0.25. PMIST set at 0.1. Black box highlights instance where the Signor Lipps effect has been exaggerated by the PDE; the TDE and RDE both identify the rapid diversity decrease present in the true diversity. Abbreviations as in Table 1.

FIGURE 6 in A simulation-based examination of residual diversity estimates as a method of correcting for sampling bias

FIGURE 6. The performance of the phylogenetic diversity estimate when errors are introduced to the phylogeny. Mean Spearman's rho values of the PDE, TDE and the best performing implementation of the RDE. PLOC, PFORM and PTAPH set at 0.25. PMIST variable. The dashed red line indicates the critical value at p=0.05. Abbreviations as in Table 1.

FIGURE 4 in A simulation-based examination of residual diversity estimates as a method of correcting for sampling bias

FIGURE 4. The performance of different implementations of the residual diversity estimate examining faunas with varying degrees of homogeneity. PFORM, PLOC and PTAPH are set at 0.25. The rate of dispersal is increased relative to the rate of local extinction to increase the homogeneity of the faunas. The dashed red line indicates the critical value at p=0.05. Abbreviations as in Table 1.

FIGURE 1 in A simulation-based examination of residual diversity estimates as a method of correcting for sampling bias

FIGURE 1. An illustration of the taphonomic filter in the simulation, shown applied to a single taxon in a single time bin. The taxon is originally present in every locality in each region it occupies, but the taphonomic filter removes it from randomly selected localities