20253 research outputs found
Sort by
Global Pasture Watch - Annual grassland class and extent maps at 30-m spatial resolution (2000—2024) V2
No full textSub-dataset: Dominant grassland class, 2015-2017
Description
Global annual grassland class and extent for 2000—2024 produced by Parente et al. (2024) within the scope of the Global Pasture Watch initiative. The mapped grassland extent includes any land cover type, which contains at least 30% of dry or wet low vegetation, dominated by grasses and forbs (less than 3 meters) and a:
maximum of 50% tree canopy cover (greater than 5 meters),
maximum of 70% of other woody vegetation (scrubs and open shrubland), and
maximum of 50% active cropland cover in mosaic landscapes of cropland & other vegetation.
The grassland extent is classified into two classes:
Cultivated grassland: Areas where grasses and other forage plants have been intentionally planted and managed, as well as areas of native grassland-type vegetation where they clearly exhibit active and 'heavy' management for specific human-directed uses, such as directed grazing of livestock.
Natural/semi-natural grassland: Relatively undisturbed native grasslands/short-height vegetation, such as steppes and tundra, as well as areas that have experienced varying degrees of human activity in the past, which may contain a mix of native and introduced species due to historical land use and natural processes. In general, they exhibit natural-looking patterns of varied vegetation and clearly ordered hydrological relationships throughout the landscape.
Open shrubland (v2-beta): Land on which the vegetation is dominated by low-growing woody plants, characterized by a sparse distribution of shrubs and dominated by woody perennials. Typically covers 50—75% of the area, with significant open ground (with or without herbaceous understory) between them, where shrub canopies are less than 10 meters in diameter, and tree cover is below 10%, meaning they do not form a continuous or semi-continuous canopy.
The dataset is organized in 69 global mosaics (25 years for each time series) in COG (Cloud Optimized GeoTIFF) format, WGS84 Coordinate Systems (EPSG:4326) and pixel size equal to 0.00025 degrees, including:
Probabilities of cultivated grassland (values range from 0–100),
Probabilities of natural/semi-natural grassland (values range from 0–100), and
Probabilities of open shrubland (values range from 0–100), and
Dominant class (0-other land cover, 1-cultivated grassland and 2-natural/semi-natural grassland, 3-open shrubland).
All raster files are in unsigned 8-bit integer format and use 255 as no-data value (pixels ignored by prediction), following an specific naming convention:
Project name: Global Pasture Watch (gpw)
Class name: cultivated grassland (cultiv.grassland), natural/semi-natural grassland (nat.semi.grassland), open shrubland (open.shrubland) and dominant grassland (grassland)
Procedure combination: Random Forest (rf), median filter (med.filt) and balanced threshold (bthr).
Variable type: probability (p) and factor class (c)
Spatial resolution: 30m
Begin of time reference: date of first Landsat composite used by the modeling (20240101)
End of time reference: date of last Landsat composite used by the modeling (20241231)
Spatial extent: global (go)
Coordinate system: World Geodetic System 1984, used in GPS (epsg.4326)
Version: v2
Related resources
Maps of dominant grassland:
2000-2002 2003-2005 2006-2008 2009-2011 2012-2014 2015-2017 2018-2020 2021-2023 2024
Probability maps of cultivated grassland:
2000-2024 (All URLs)
Probability maps of natural/semi-natural grassland:
2000-2024 (All URLs)
Grassland reference samples based on VHR imagery (2000–2024):
GeoPackage files
Global machine learning models (Random Forest):
Parquet and joblib python files
Reference sampling design derived by FSCV:
GeoPackage and raster files
Harmonized reference samples based on existing LULC dataset:
GeoPackage and raster files
Source code for reproducibility:
GitHub release
Mapping feedback tool:
GeoWiki
Data catalogues:
OpenLandMap STAC Google Earth Engine
Support
For questions of bugs/inconsistencies related to the dataset raise a GitHub issue in https://github.com/wri/global-pasture-watc
Global subnational Gini coefficient (income inequality) and gross national income (GNI) per capita PPP datasets for 1990-2023 V3
No full textThis dataset provides a gridded subnational datasets for
Income inequality (Gini coefficient) at admin 1 level
Gross national income (GNI) per capita PPP at admin 1 level
The datasets are based on reported subnational admin data and spans three decades from 1990 to 2023
The datasets are presented in details in the following publication. Please cite this paper when using data.
Chrisendo D, Niva V, Hoffman R, Sayyar SM, Rocha J, Sandström V, Solt F, Kummu M. 2025. Rising income inequality across half of global population and socioecological implications. Nature Sustainability. https://doi.org/10.1038/s41893-025-01689-4
Code is available at following repositories:
Gini coefficient data creation: https://github.com/mattikummu/subnatGini
GNI per capita data creation: https://github.com/mattikummu/subnatGNI
analyses for the article: https://github.com/mattikummu/gini_gni_analyses
The following data is given (formats in brackets)
Gini coefficient:
Please note, two distinct datasets for Gini cofficient are given. One based on SWIID national dataset, and another for WID national dataset. These are separated in file names as follows: _disp_ for SWIID (disposable income); _WID_ for WID.
Income inequality (Gini coefficient) at admin 0 level (national) (GeoTIFF, gpkg, csv)
Income inequality (Gini coefficient) at admin 1 level (subnational) (GeoTIFF, gpkg, csv)
Slope for Gini coefficient at admin 1 level (GeoTIFF; slope is given also in gpk and csv files)
Uncertainty (standard deviation for each year and admin area; uncertainty for slope) for Gini (based on SWIID) at admin 1 level (gpkg)
Input data for the script that was used to generate the Gini coefficient (input_data_gini.zip)
Gross national income (GNI) pear capita PPP (in 2021 USD):
Gross national income (GNI) per capita PPP at admin 0 level (national) (GeoTIFF, gpkg, csv)
Gross national income (GNI) per capita PPP at admin 1 level (subnational) (GeoTIFF, gpkg, csv)
Slope for GNI per capita (log10) at admin 1 level (GeoTIFF; slope is given also in gpk and csv files)
Input data for the script that was used to generate the GNI per capita PPP (input_data_GNI.zip)
Files are named as follows
Format: raster data (GeoTIFF) starts with rast_*, polygon data (gpkg) with polyg_*, and tabulated with tabulated_*.
Admin levels: adm0 for admin 0 level, adm1 for admin 1 level
Product type:
_gini_disp_ for gini coefficient based on SWIID national dataset (disposable income)
_gini_WID_ for gini coefficient based on WID national dataset
_uncertainty_slope_gini_disp_ for slope uncertainty of SWIID based Gini
_uncertainty_SD_gini_disp_ for standard deviation of SWIID based Gini
_gni_perCapita_ for GNI per capita PPP
Metadata
Grids
Resolution: 5 arc-min (0.083333333 degrees)
Spatial extent: Lon: -180, 180; -90, 90 (xmin, xmax, ymin, ymax)
Coordinate ref system: EPSG:4326 - WGS 84
Format: Multiband geotiff; one band for each year over 1990-2021
Unit: no unit for Gini coefficient and PPP USD in 2017 international dollars for GNI per capita
Geospatial polygon (gpkg) files:
Spatial extent: -180, 180; -90, 83.67 (xmin, xmax, ymin, ymax)
Temporal extent: annual over 1990-2021
Coordinate ref system: EPSG:4326 - WGS 84
Format: gkpk
Unit: no unit for Gini coefficient and PPP USD in 2017 international dollars for GNI per capita
Version 3 changes (24.07.2025)
both datasets updated to cover 1990-2023 (previously 1990-2021)
extrapolation method updated
Gini data now produced also using WID national data as a base (previously only one based on SWIID was provided)
uncertainty analysis done for Gini (SWIID
Sample data for training EPIC-IIASA global gridded crop model emulators
No full textThis repository provides intermediary and output data required to reproduce the study Folberth et al. (submitted) "CROMES v1.0: A flexible CROp Model Emulator Suite for climate impact assessment". Associated code is provided in a separate repository at https://doi.org/10.5281/zenodo.14901127. Raw data can be obtained from the references.
The below folder structure shows how the project should be orgainzed to match the code repository. Herein, we provide separate zipped folders for the lowest level, e.g., processed, auxiliary.
project_name/ ### Main folder, by default labelled cromes in the associated code repository
+-- data/ ## All input and auxiliary data
¦ +-- raw/ # Immutable raw data. Not included here but referenced in the publication
¦ +-- processed/ # Cleaned and processed data. I.e., explanatory and target features for
¦ +-- auxiliary/ # Data required for evaluations and visualization
¦ +-- temp/ # Temporary storage of train/predict-ready feature files
+-- src/ ## All code required for training and evaluating emulators
¦ +-- features/ # Feature engineering scripts (included in separate repository)
¦ +-- models/ # Model training and evaluation scripts (included in separate repository)
¦ +-- visualization/ # Scripts for generating visualizations (included in separate repository)
+-- output/ ## Target folder for all outputs, here pre-populated
+-- models/ # Emulator model files
+-- feature_importance # Feature importances for pre-trained emulator models
+-- figures/ # Generated plots and figures (selected plots from the publication)
+-- tables/ # Generated tables (performance metrics, etc.)
+-- predictions/ # Generated crop yield predictions
+-- evals/ # Evaluation
Single- and double-cropping soybean production systems in Brazil: EPIC-IIASA GGCM simulations
No full textSoy production in Brazil continues to expand in response to growing global demand. This expansion has been enabled by improved soy varieties and greater agricultural flexibility, allowing for practices such as double-cropping. Being able to model recent past areal expansion and productivity increase of these cropping systems requires a good assessment of their productivity and environmental externalities (e.g., soil organic carbon, soil erosion, nutrient leaching) under alternative management practices.
To assess the productivity and environmental impacts of soybean production systems in Brazil, we employed a gridded modelling framework EPIC-IIASA GGCM, based on the process-based model EPIC (Environmental Policy Integrated Climate, https://epicapex.tamu.edu/about/epic). The analysis included key cropping systems and conservation practices commonly used in Brazil today: soy mono-cropping, no-till double cropping with corn, and no-till soy cultivation with pearl millet cover cropping. Additionally, we evaluated the role of irrigation and cultivar transitions as management options to support the sustainable transformation of soybean production.
Simulations were conducted on a 0.5° resolution grid covering the period from 1982 to 2016, contributing to Deliverable 6.1, “Non-food biomass production based on biophysical modelling”, developed under the CLEVER project (Creating Leverage to Enhance Biodiversity Outcomes of Global Biomass Trade), funded by the Horizon Europe programme (Topic: HORIZON-CL6-2021-BIODIV-01-15, Project Number: 101060765). Among other goals, the CLEVER project seeks to enhance biodiversity outcomes through improved modelling of global biomass trade.
These files are included:
1) CLEVER_BRA_SOY_YLD.zip including 20 NetCDF files containing gridded soybean yield data modelled for soybean production systems, based on the simulation design described in the ReadMe file
2) CLEVER_BRA_SOY_ENVI.zip including 60 NetCDF files containing gridded environmental externality data modelled for soybean production systems, based on the simulation design described in the ReadMe file.
3) ReadMe.docx: Simulation design and metadata description
CMIP7 ScenarioMIP infilling database for simple climate model workflow
No full textThe files here are infilling databases. They are used for the 'infilling' step of simple climate model assessment. This involves inferring emissions of one species (e.g. HFC32) based on emissions of another species (e.g. CO2) and relationships between these two species seen in other scenarios. For more details about infilling, see Lamboll et al., 2020.
The database is provided in three different formats. The versions of the scenarios used to compile this database is provided in the versions.json file.
The database was derived using the code in this repository: https://github.com/iiasa/emissions_harmonization_historical. The filenames are composed of identifiers related to the processing of each of the different input data sources. To identify the exact meaning of these identifiers, please see the processing code in https://github.com/iiasa/emissions_harmonization_historical
GLOBIOM output data for the forest sector part in CLEVER D7.2 deliverable
No full textSupplement for CLEVER D7.2 deliverable including output data for the forest sector part figures. CLEVER D7.2 forest sector part analyzes wood-based bioeconomy and novel forest sector supply chains. Results show that bioeconomy can be combined to sustainable forestry, but this needs novel forest sector supply chain solutions such as forest plantation, non-woody biomass use for bioenergy, and/or circular economy.
The record contains one readme file (readme_D72_ForestSupplyChains.docx) and one model output reporting csv file (CLEVER_D7.2_ForestSupplyChains_results.csv
Global Pasture Watch - Annual goat density layers at 1-km for 2000–2022 (including 95% prediction interval)
No full textGlobal annual layers of goat density at 1-km spatial resolution convering the period of 2000–2022. The layers were produced using harmonized and used as reference data (55,336 census polygons and 172,536 individual data entries), random forest predictive models and a large stack of multi-source harmonized gridded/raster spatial layers (128 individual raster spatial layers harmonized at 1 km spatial resolution).
Raster cell values represent heads km2 including:
Mean predicted values (_m_)
Upper prediction interval based on 97.5th percentiles (_p.975_)
Lower prediction interval based on 2.5th percentiles (_p.025_)
Based on 95% probability quantiles, prediction intervals are relatively wide; therefore, for a more effective use, we recommend converting them to standard deviation by dividing the range (p.975 - p.025) by four.
Raw/Uncalibrated headcounts are also provided and were computed by multiplying the density values by the actual area of potential land for livestock production.
In line with a request from our funders, livestock Layers will remain under embargo in Zenodo until the final acceptance of peer-reviewed publication. They can be accessed during the reviewing process by filling-in a form via Global Pasture Watch Early Access data program (https://survey.alchemer.com/S3/7859804/Pasture-Early-Adopters). All modeling framework presented in this work is publicly available at: https://github.com/wri/global-pasture-watch. We are currently preparing the data to be ingested in STAC and Google Earth Engine
Global Pasture Watch - Annual livestock headcount layers for cattle, goats, sheep, horses, and buffaloes at 1-km 2000–2022 (FAOSTAT-adjusted) (Part-2)
No full textGlobal annual layers of livestock distribution of cattle, goats, sheep, horses and buffaloes at 1-km spatial resolution convering the period of 2000–2022. The layers were produced using harmonized and used as reference data (55,336 census polygons and 939,257 individual data entries), random forest predictive models and a large stack of multi-source harmonized gridded/raster spatial layers (128 individual raster spatial layers harmonized at 1 km spatial resolution).
Raster cell values represent number of livestock animals (headcounts) calibrated according to FAOSTAT database. The adjustment was applied individually to each country, ensuring that the adjusted pixel values accurately reflected the national totals. For applications relying on headcount data from FAOSTAT the provided layers serve as a fully compatible alternative.
In line with a request from our funders, livestock maps will remain under embargo in Zenodo until the final acceptance of peer-reviewed publication. They can be accessed during the reviewing process by filling-in a form via Global Pasture Watch Early Access data program (https://survey.alchemer.com/S3/7859804/Pasture-Early-Adopters). All modeling framework presented in this work is publicly available at: https://github.com/wri/global-pasture-watch. We are currently preparing the data to be ingested in STAC and Google Earth Engine
HABITABLE Focus Group Data: Livelihoods, climate risks, adaptation, and migration impacts
No full textThis dataset is the product of HABITABLE () Work Package 5, "Migration Impacts," which aimed to improve the understanding of how different forms of migration and mobility affect social-ecological systems at places of origin.
It explores both direct and indirect impacts of migration on:
i) the coping and adaptation capacities of individuals, households, and communities facing climate risks; and
ii) the structural root causes of vulnerability, including social, material, cultural, and political dimensions.
Recommended use:
Climate vulnerability and adaptation research
Migration and development studies
Intersectional and community-based adaptation research
Study design and data collection:
Study type: Structured and visual participatory group discussions
Target population: Rural adults (18+), from migrant and non-migrant households in climate affected areas
Methodology: Participatory Rural Appraisal (PRA)-inspired tools, using visualizations and facilitated discussions
Data collection instruments: 12 thematic sessions (e.g., migration impacts, climate risks, adaptation, social change, inequality)
Sampling: Mixed groups based on gender, age, socioeconomic background; purposive and stratified participant selection
Data format: Coded textual segments and transcribed visual materials
Ethical considerations: Informed consent, anonymization of site and personal data
Geographic coverage and sample size:
Ghana: 2 sites, 24 sessions with each 6-8 participants
Mali: 4 sites, 48 sessions with each 6-8 participants
Kenya: 3 sites, 36 sessions with each 6-8 participants
Thailand: 3 sites, 36 sessions with each 6-8 participants
Data processing and structure:
Visual outputs and discussion notes were compiled into standardized reports and manually coded using a content-driven coding scheme. The resulting dataset includes text segments and image transcriptions, organized under thematic codes related to development trajectories, wellbeing, vulnerability, adaptation, migration impacts, gender, inequality, and environmental change.
Sensitivity and privacy:
No personally identifiable information was collected. Site names were pseudonymized, and sensitive content (e.g., conflict/cohesion issues) was handled with care to prevent harm.
Technical documents :
Data description: see file Description_Habitable_WP5_FGD_Data.pdf
Data collection tools: see file D51_Tools_Data_Collection_WP5_Migration_Impact.pd
