Index data for "Lomax et al. (2025), The relative productivity index: Mapping human impacts on rangeland vegetation productivity with quantile regression forests. Ecological Indicators 171 (113208)".

This record contains the final processed data for the four rangeland condition indices assessed in the following paper:

Lomax, Guy A., Thomas W.R. Powell, Timothy M. Lenton, and Andrew M. Cunliffe. ‘The Relative Productivity Index: Mapping Human Impacts on Rangeland Vegetation Productivity with Quantile Regression Forests’. Ecological Indicators 171 (February 2025): 113208. https://doi.org/10.1016/j.ecolind.2025.113208.

Description of the data records

rue_all.tif - a raster file with 19 layers containing annual rain use efficiency values (annual GPP divided by annual precipitation) for each pixel in the study area.

restrend_resids.tif - a raster file with 25 layers. Layers 1-6 contain the parameters of the pixel-wise linear regression relating annual GPP to annual precipitation and temperature; layers 7-25 contain the annual GPP residuals calculated from this relationship according to the RESTREND method:

yint, ppt_slope, t_slope - the intercept, GPP-precipitation slope and GPP-temperature slope of the pixel-wise linear regression model.

rsq - the overall R-squared of the pixel-wise linear regression model.

ppt_p_value and t_p_value - the pixel-wise p-values of the  containing the annual residuals from a pixelwise linear precipitation-GPP relationship as calculated using the RESTREND method.

resid_20XX - the annual GPP residuals from the GPP-precipitation-temperature linear relationship.

rpi.zip - an archive containing 19 annual raster files. Each raster represents a year in the dataset and contains four layers:

GPP - the observed total gross primary productivity.

quantile_pred - the predicted potential GPP given the biophysical conditions experienced by each pixel in that year (i.e., the 90th percentile prediction of the quantile regression forest model).

mean_pred - the predicted mean GPP given the biophysical conditions experienced by each pixel in that year.

rpi - the estimated relative productivity index for each pixel in that year, calculated as the observed GPP divided by the predicted potential GPP.

rpi_mean.tif - a raster file with a single layer giving the mean RPI per pixel across the study period.

lgs.zip - an archive containing three raster files. Each raster file represents the results of the LGS analysis with a different value of k (the number of clusters) and contains four layers:

cluster - the assigned cluster of each pixel.

GPP - the mean annual GPP of each pixel over the study period.

mean - the mean annual GPP of all pixels in the assigned cluster

potential - the estimated potential annual GPP of pixels in the assigned cluster, given by the 90th percentile of observed mean annual GPP values.

lgs_ratio - the estimated LGS ratio, calculated as the observed mean annual GPP divided by the estimated potential annual GPP.