Executive summary

Stage 3 of the Pūhoi to Warkworth (P2W) sediment assessment programme evaluates the performance of the GLEAMS erosion model used prior to the commencement of works. The work builds on the Stage 1 and Stage 2 analyses of measured sediment losses and sediment retention pond (SRP) performance. The objectives were to compare GLEAMS-predicted hillslope sediment yields with measured sediment loads, investigate reasons for any discrepancies, assess the suitability of GLEAMS for earthworks applications, and place the findings within the context of other New Zealand studies.

GLEAMS and USLE predictions were sourced from the original Assessment of Environmental Effects (AEE) and Construction Water Assessment (CWA) reports. Both models produced very similar specific sediment yield (SSY) estimates for Hill and Flat terrain types, despite relying on different rainfall erosivity inputs. Measured sediment loads were available for two SRP inlet sites. One site (SRP57350) produced SSYs comparable to the modelled values, while the other (SRP53000) produced substantially lower SSYs due to lower measured suspended sediment concentrations. All measured loads are likely to be underestimates because event peaks were not consistently sampled, meaning the monitoring dataset is not sufficiently robust to validate or invalidate the GLEAMS predictions.

The evaluation of GLEAMS highlights that it is an empirical, USLE-based erosion model developed for agricultural land and not designed to represent rilling, gullying, or concentrated flow processes typical of earthworks. Although the model has been applied in several New Zealand earthworks assessments, it was never intended to provide absolute sediment load predictions, and no calibration to earthworks-specific erosion data has been undertaken. However, a review of other New Zealand studies shows that the GLEAMS-derived hillslope erosion sediment yields for P2W fall within the broad range reported for disturbed land.

Overall, the apparent agreement between modelled and measured sediment yields at one site should not be interpreted as evidence of model accuracy, given the substantial uncertainty in the monitoring data and the structural limitations of GLEAMS. The model can provide indicative estimates of hillslope erosion but should not be relied upon for precise quantitative predictions without calibration and a clearer understanding of site-specific erosion processes. Further refinement of GLEAMS inputs are unlikely to meaningfully reduce model–measurement discrepancies, and any future modelling would require improved field data for model calibration and consideration of alternative, process-based approaches.

Earth Sciences New Zealand. Prepared for Auckland Council and NZTA, April 2026

Groundwater Loading Effects of Agricultural Management Systems GLEAMS