Introduction
NIWA has produced a 29-year atmospheric regional reanalysis dataset – called the New Zealand Reanalysis (NZRA) – for all of Aotearoa New Zealand.
Atmospheric reanalysis datasets are comprehensive, long-term records of climate variables – such as precipitation, wind and cloud cover. They are produced by combining historical observational weather data (e.g. measurements from weather stations) with numerical weather prediction (NWP) models, resulting in consistent, gridded estimates of atmospheric variables over long periods (typically decades).
The NZRA was produced in partnership with NIWA’s international collaborators in numerical weather prediction, the Bureau of Meteorology in Australia and the UK Met Office, through the global Momentum Partnership.
Benefits of the NZRA
While global atmospheric datasets have been available for decades, the NZRA is the first very high-resolution regional reanalysis that covers all of New Zealand.
Because it covers a smaller geographic area than a global reanalysis, the NZRA can provide data at a higher spatial and temporal resolution. In addition, the NZRA better models New Zealand’s complex terrain and the intricate atmospheric processes that drive our local weather and climate.
Reanalysis datasets provide important insights into trends in weather and climate over time – and can help researchers understand things like expected frequency of severe weather events or how a local climate has been changing, and what might happen in the near future. This information can help us better understand and plan for future weather and climate related hazards.
Furthermore, reanalysis products like the NZRA are crucial to the success of any form of machine learning application related to weather and climate.
About the NZRA – technical information
- Covers 1990 to 2018.
- 1.5 km horizontal resolution over New Zealand – see map below for the extent of the NZRA domain.
- 30 min temporal resolution of output comprising variables on:
- Surface
- 21 pressure levels spanning 1000-100 hPa
- 70 model levels spanning surface-40 km.
- Initial and lateral boundary conditions derived from the Bureau of Meteorology’s 12 km resolution BARRA-R model.
Development of the NZRA generated 2 petabytes (PB) of data and used around 20 million core-hours of high-performance computing resource – equivalent to running around 250 high-end laptops continuously for a year.
Applications
The NZRA dataset is already being used across a range of science applications, including:
- as a training dataset for NIWA NEXT, an operational nowcasting system for wind and solar generation sites across New Zealand
- studying long term climatological trends in New Zealand’s Fire Weather Indices
- evaluating potential onshore and offshore wind power generation across New Zealand
- as training data for a high-resolution Neural Weather Model for New Zealand
- as input to a new snowpack model applied across the Southern Alps region to investigate snow depth variability across decades.
It has many more potential applications too, such as cross-disciplinary research on drought, hydrological and land-surface processes, public health, and risk/loss modelling as part of mitigation planning for climate change.
What next?
NIWA is planning to expand on the original NZRA dataset by:
- creating a dataset for 2019–Present Day, using the Bureau of Meteorology’s BARRA2 reanalysis dataset as input data
- working with our Momentum Partnership collaborators to develop and produce high-resolution regional reanalysis workflows that benefit from recent advances in convective-scale ensemble and data assimilation techniques and make use of the upcoming ERA6 global reanalysis product.
