Schepen et al. (2025) Forecasting agricultural drought: the Australian Agricultural Drought Indicators

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2025
• Date: 2025-10-21
• Authors: Andrew Schepen, Andrew Bolt, Dorine Bruget, John Carter, Donald S. Gaydon, Mihir Gupta, Zvi Hochman, Neal Hughes, Chris Sharman, Peter Tan, Peter Taylor
• DOI: 10.5194/nhess-25-4053-2025

Research Groups

• Commonwealth Scientific and Industrial Research Organisation (CSIRO), Dutton Park, QLD, Australia
• Commonwealth Scientific and Industrial Research Organisation (CSIRO), St Lucia, QLD, Australia
• Queensland Government Department of Environment, Tourism, Science and Innovation, Dutton Park, QLD, Australia
• Australian Bureau of Agricultural and Resource Economics, Canberra, ACT, Australia
• Independent researcher: Sydney, Australia
• Commonwealth Scientific and Industrial Research Organisation (CSIRO), Sandy Bay, TAS, Australia

Short Summary

This study evaluates the skill of the Australian Agricultural Drought Indicators (AADI) system, which forecasts agricultural drought using biophysical and agro-economic models driven by the ACCESS-S2 climate model. It demonstrates that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit, providing earlier and more confident drought warnings than rainfall deficits alone.

Objective

• To evaluate the skill of drought indicator forecasts generated by the Australian Agricultural Drought Indicators (AADI) system, driven by the ACCESS-S2 dynamical global climate model, over a hindcast period from 1990–2018.

Study Configuration

• Spatial Scale: Approximately 5 km x 5 km grid across Australia for AADI outputs; ACCESS-S2 raw data on an approximate 80 km x 60 km grid, downscaled to 5 km.
• Temporal Scale: Hindcast period from 1990–2018 for AADI evaluation (ACCESS-S2 hindcasts 1981-2018). Forecast lead times up to 18 months, with primary evaluation up to 12 months. Forecasts issued monthly, evaluated quarterly (April, July, October, January).

Methodology and Data

• Models used:
  • AADI (Australian Agricultural Drought Indicators) system
  • ACCESS-S2 (dynamical global climate model)
  • APSIM (Agricultural Production Systems sIMulator) for crop yield simulation (wheat, sorghum)
  • AussieGRASS (pasture growth model)
  • farmpredict (statistical farm microsimulation model for farm profit)
  • Bayesian Joint Probability (BJP) modelling approach for climate forecast calibration
  • Method of Fragments (MOF) for empirical downscaling
  • Schaake Shuffle for spatial and temporal dependencies
• Data sources:
  • SILO gridded climate data (observational, 5 km resolution, 1960–2018)
  • ACCESS-S2 raw hindcasts (1981–2018)
  • Australian Agricultural and Grazing Industry Survey (AAGIS) data (1992–2022)
  • Soil type data (derived from National Generic Soil Group)

Main Results

• Antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year.
• Forecast confidence for farm profit increases with shorter lead times: median skill rises from 43% at 12 months to 67% at 6 months and 73% at 3 months.
• Median farm profit biases remain below 2% across all lead times, with high reliability.
• Forecasts for wheat, sorghum, and pasture are skillful and reliable in ensemble spread, though residual biases can occur (e.g., up to 20% for sorghum).
• The AADI system can identify drought-impacted areas with increased confidence up to 6 months earlier than traditional rainfall deficit indicators.
• Climate forecasts show moderate skill at 1-month lead time for most variables (temperature, evaporation, vapour pressure), with limited skill beyond 2 months, especially for rainfall and radiation.

Contributions

• Developed and evaluated a novel, comprehensive agricultural drought forecasting system (AADI) that integrates biophysical and agro-economic models with dynamical climate forecasts.
• Demonstrated that incorporating antecedent environmental and economic conditions significantly improves drought forecast skill beyond that of climate forecasts alone, particularly for farm profit.
• Provided a robust ensemble verification of agricultural drought indicators (crop yields, pasture growth, farm profit) across Australia, highlighting their accuracy, reliability, and lead time advantages over traditional meteorological indicators.
• Showcased the AADI system's ability to provide earlier warnings of drought impacts and its performance in both dry and wet conditions, supporting proactive risk management and decision-making.

Funding

• Australian Government, Department of Agriculture, Fisheries and Forestry.

Citation

@article{Schepen2025Forecasting,
  author = {Schepen, Andrew and Bolt, Andrew and Bruget, Dorine and Carter, John and Gaydon, Donald S. and Gupta, Mihir and Hochman, Zvi and Hughes, Neal and Sharman, Chris and Tan, Peter and Taylor, Peter},
  title = {Forecasting agricultural drought: the Australian Agricultural Drought Indicators},
  journal = {Natural hazards and earth system sciences},
  year = {2025},
  doi = {10.5194/nhess-25-4053-2025},
  url = {https://doi.org/10.5194/nhess-25-4053-2025}
}