Summary This map shows friction values on a grid of 90x90 meters for Brazil. Friction values represent the average travel time by car, boat, train or foot (depending on the surface) to cross one grid-cell in horizontal or vertical space. Friction maps are fundamental for calculating accumulated costs maps that can serve as a substitute for infrastructure data in spatial modelling. The base-data used for creating this friction map comprises: A layer of road-infrastructure data obtained from the Brazilian Ministry of transportation (Ministério dos Transportes e Departamento Nacional de Infraestrutura de Transportes). The data is available at http://pnlt.imagem-govfed.opendata.arcgis.com/ . Planned roads where filtered out of the dataset. A layer of rail-roads in Brazil from the same source as (1). A layer of oficial hidroways in Brazil from the same source as (1). A layer of land-use classes in 2014 obtained from Mapbiomas. The data is available for download here: http://mapbiomas.org/ A layer of hydrological data that was composed from two different datasets from the Brazlian Water Agency (Agencia Nacional das Águas). The source data can be downloaded here: http://www.ana.gov.br/bibliotecavirtual/redeHidrografica.asp. The datasets from ANA are Drainage Network models based on SRTM data (2000) in the scale of 1:100.000 and in the scale of 1:250.000. We used the lower resolution data (1:250.000) to identify mayor rivers. However the spatial accuracy of this data is not sufficient for our purpose and affluent rivers to the main river are not included. We therefore developed a model to buffer the low resolution data-set with a 10km buffer on each side of the river. We than used this buffer areas to crop out the higher resolution data (1:100.000) thereby including affluent rivers of up to 10 km on each side of the main river. Furthermore the higher resolution data is spatially more accurate if compared to satellite imagery. A layer of sloped data based on global SRTM elevation data from 2000 available at: https://urs.earthdata.nasa.gov/ Description of the model: First all data was reprojected to WGS84, than the data was reclassified to hold travel times to cross one grid-cell for each class in seconds. The following assumptions on travel speed had been used for this purpose: Paved Roads: 60km/h Unpaved Roads: 40 km/h Waterbodies or navigable stretches in the flooding season: 10 km/h Hidroways: 20km/h Railways: 40km/h Forest: 3 km/h (comprises in the original data: forest, planted forest, coastal zone forest) Non-forest: 12 km/h (comprises: Agricultural areas, non-forest vegetation, pastures, others) Afterwards, all data-sources where rasterized at a resolution of 90x90meters and stacked in a rasterstack. Than the highest value of each grid-cell in the stack was extracted into a new summary raster. The summary layer is the raw friction map that needs to be corrected for the effect of slope. Slope was calculated from the SRTM1 data and rescaled to the same extent and resolution as the raw friction map. Slope was included as a constant factor with the following assumptions: slope between 0 and 5 degrees: slope constant 1 (travel speed is not impacted) slope between 5-10 degrees: slope constant 2 (travel speed is reduced by half) slope between 10-15 degrees: slope constant 3 ( travel speed is reduced to a third of the original speed) slope above 15 degrees: slope constant is 10 (travel speed is reduced to 10% of the original speed) Slope effects where not applied on water bodies. The corrected friction map was exported as a GeoTiff (WGS84) in a resolution of 90x90 meters. Additional Information: The friction map was created using the R-software environment and GDAL. The script to replicated the results or modify the model with other base data can be obtained on request. The friction map can be used to calculate accumulated cost maps with GIS software (costdistance function in ArcGIS and r.cost function in QGIS).