1. Aggregate the potential natural vegetation dataset by Ramankutty & Foley (1999) from 15 to 12 land cover classes.
2. Use Historical Dataset on Human Environment (HYDE) global data for human population density and percentage cover by urban, crop and pasture lands for the year 1800 were used to classify the land terrestrial areas into the different anthrome classes.
3. Stratify the population density data into 6 density classes, namely, urban, > 2500 persons per sq. km; dense, > 100 persons per sq. km; residential, 10-100 persons per sq. km; populated, 1-10 persons per sq. km; remote, < 1 person sq. km; wild, 0 persons per sq. km.
4. Apply a sequence of classification thresholds to the natural vegetation data for land area covered by urban settlements, rice, irrigated and rainfed crops and pastures.