High resolution drought forecasting
NIWA and the Ministry for Primary Industries (MPI) are working together to develop a new drought forecasting tool, with each organisation investing $100,000 in the project.
The tool uses innovative climate modelling, the latest in machine learning and other data-driven techniques. It will help farmers and growers better prepare for periods of dryness and drought.
The tool updates daily to provide forecasts at a much higher spatial resolution than previously available. This will enable the provision of district-level predictions of dryness and drought.
“Having a tool that draws on the best available science each day to provide advance warning of future dry spells will make a big difference to farmers' planning and decision making. This will not only contribute to the bottom line, but also to their own wellbeing and animal welfare.”
Nick Story, Director Rural Communities and Farming Support, MPI
Building on NIWA’s long history of drought forecasting
NIWA has provided Aotearoa New Zealand with seasonal rainfall predictions since 1999, when its Seasonal Climate Outlooks began.
In 2017 NIWA and MPI launched the NZ Drought Index (NZDI), an observational tool which measures the current state of dryness and drought.
Turning drought observations into forecasts: the NIWA35 deep learning model
The drought forecasting tool is part of the NIWA35 platform, which is using machine learning and deep learning to make high resolution sub-seasonal climate predictions up to 35 days ahead.
NIWA35 uses input data from a model issued daily by the US National Oceanic Atmospheric Administration (NOAA). NIWA uses the input data, alongside state-of-the-art deep learning, to create a downscaled (higher resolution) and more accurate forecast for Aotearoa New Zealand. This important step better captures the climatic variability that occurs across the country’s complex terrain.
NIWA35 has been producing rainfall predictions since March 2021 and predictions for soil moisture and potential evapotranspiration are in development. The main goal of this work will be to combine the outputs of these different predictions to create a forecast of meteorological drought risk that is consistent with the NZDI. This will enable those across the primary sector to be more informed about the potential for abnormal dryness and drought than ever before.
Regular briefings and real-time end user engagement
As part of the project, NIWA meteorologist Ben Noll provides regular briefings to MPI about drought risk across New Zealand, including presenting content from NIWA35 in development. This enables end users of the drought forecasting tool at MPI to give feedback in near real time as the project progresses, and this feedback to be incorporated into future development.
Next steps will include sharing a beta version of the drought forecast tool for testing by user groups in summer 2022/23.
The final product is expected to be made available publicly a year later (summer 2023/24).
An example image from the tool in development: January 2022 was a very dry month, as indicated by the dark red colouring and negative values on the Standardised Precipitation Index on the far right.
The NIWA35 outlook issued in late December 2021 indicated an elevated risk for January dryness. Note how the image on the far left (35-day forecast) is very similar to the image on the far right (observation).