Streamflow reduction tables for commercial afforestation in South Africa, with catchment data

This dataset contained the national streamflow reduction (SFR) tables for commercial afforestation and the research catchment data used to verify them, harvested from WRC Report TT 173/02 (2002). The work was done in WRC project K5/1110, jointly funded by the Water Research Commission and the Department of Water Affairs and Forestry. The ACRU agrohydrological model was first verified against catchment experiments at Cathedral Peak (catchments II and IV), Westfalia (A to D), Jonkershoek (Lambrechtsbos A and B), Witklip (catchment V), Ntabamhlope (V7H004), Cedara (U2H018) and the Seven Oaks evaporation site (B29). For these sites the dataset gave catchment characteristics (weir, rain gauges, area, altitude, coordinates, mean annual precipitation, soils), sub-catchment areas, year-by-year land cover histories of planting and clearfelling, monthly climate means, goodness-of-fit statistics for simulated monthly total and low flows under detailed and default inputs, comparisons with regression-estimated flow reductions, and the confidence bands the authors proposed for applying the national tables. The model was then applied to all 843 quaternary catchments with a mean annual precipitation above 650 mm. Daily simulations were run for 1950 to 1993 under the modal Acocks veld type as baseline and under complete afforestation with eucalypts (representative age 4 years), pines (7 years) and wattle (4 years), each on shallow (0.6 m), medium (0.9 m) and deep (1.2 m) soils. For each catchment the tables gave area, mean annual precipitation, baseline annual total flow and low flow (driest three months), and the reductions in total flow and low flow for each genus and soil depth, in mm, as both medians and means. The data were harvested in 2026 from the born-digital report by the Agricultural Research Council for the Water Research Observatory. All 26 columns of the published summary rows of the SFR tables were reproduced exactly, and multi-line research catchment tables were checked cell by cell against page images. Areas were missing in the source for 17 catchments, a forestry climatic zone field was inferred from the colour coding of the tables, and conflicting values between tables and text were retained and flagged. The authors recommended the medium soil depth values as most representative and noted that low flow estimates carried considerably lower confidence than total flow estimates. Observed and simulated streamflow time series appeared only as figures and were not included. The dataset was provided as a long-format CSV of the SFR tables and two Excel workbooks with documentation and verification sheets.