ERA5-Land is a reanalysis dataset providing a consistent view of the evolution of land variables over several decades at an enhanced resolution compared to ERA5. ERA5-Land has been produced by replaying the land component of the ECMWF ERA5 climate reanalysis. Reanalysis combines model data with observations from across the world into a globally complete and consistent dataset using the laws of physics. Reanalysis produces data that goes several decades back in time, providing an accurate description of the climate of the past. ERA5-Land uses as input to control the simulated land fields ERA5 atmospheric variables, such as air temperature and air humidity. This is called the atmospheric forcing. Without the constraint of the atmospheric forcing, the model-based estimates can rapidly deviate from reality. Therefore, while observations are not directly used in the production of ERA5-Land, they have an indirect influence through the atmospheric forcing used to run the simulation. In addition, the input air temperature, air humidity and pressure used to run ERA5-Land are corrected to account for the altitude difference between the grid of the forcing and the higher resolution grid of ERA5-Land. This correction is called 'lapse rate correction'.
The ERA5-Land dataset, as any other simulation, provides estimates which have some degree of uncertainty. Numerical models can only provide a more or less accurate representation of the real physical processes governing different components of the Earth System. In general, the uncertainty of model estimates grows as we go back in time, because the number of observations available to create a good quality atmospheric forcing is lower. ERA5-land parameter fields can currently be used in combination with the uncertainty of the equivalent ERA5 fields. The temporal and spatial resolutions of ERA5-Land makes this dataset very useful for all kind of land surface applications such as flood or drought forecasting. The temporal and spatial resolution of this dataset, the period covered in time, as well as the fixed grid used for the data distribution at any period enables decisions makers, businesses and individuals to access and use more accurate information on land states.

1 km gridded estimates of daily and monthly rainfall for Great-Britain and Northern Ireland (together with approximately 3000 km2 of catchment in the Republic of Ireland) from 1890 to 2019. The rainfall estimates are derived from the Met Office national database of observed precipitation. To derive the estimates, monthly and daily (when complete month available) precipitation totals from the UK rain gauge network are used. The natural neighbour interpolation methodology, including a normalisation step based on average annual rainfall, was used to generate the daily and monthly estimates. The estimated rainfall on a given day refers to the rainfall amount precipitated in 24 hours between 9am on that day until 9am on the following day. The CEH-GEAR dataset has been developed according to the guidance provided in BS 7843-4:2012. Full details about this dataset can be found at https://doi.org/10.5285/dbf13dd5-90cd-457a-a986-f2f9dd97e93c

The amount of monthly rainfall in Northern Ireland varies from year to year. During the period in consideration, the lowest rainfall levels were recorded in April 2021 at just **** millimeters. Meanwhile, the most rainfall occurred in February 2020, when ***** millimeters fell. In that same month, there were **** rain days recorded, an unusually high number for February. A rain day is when there is a total of 1 mm or more of rain in a day. Seasonal rainfallSince 2010,********has been on average the wettest season in Northern Ireland. In 2023, however, ****** was the wettest season, with nearly *** mm of rainfall. That year, winter was the driest season, with *** mm of rainfall. Regional rainfallWhen compared to the rest of the UK, Northern Ireland receives less rain than both Scotland and Wales, but more than England. In 2024, the country experienced****** mm of rainfall. In comparison, Scotland and Wales received ***** and ******mm, respectively. This is due to the Scottish Highlands high levels of rain and Wales’ location in comparison to the Atlantic Ocean.

In December 2015, there were **** rain days recorded in Northern Ireland. This was the highest number recorded during the indicated period. A rain day is when more than 1 millimeter of rain falls within a day. The number of rain days in Northern Ireland amounted to **** in April 2025, a big decrease from the same month the previous year.

Code and Dataset for "On the Extrapolation of Generative Adversarial Networks for downscaling precipitation extremes in warmer climates"

This dataset accompanies the research paper titled "On the Extrapolation of Generative Adversarial Networks for downscaling precipitation extremes in warmer climates", currently under review for the AGU Journal GRL. The study introduces a novel Regional Climate Model (RCM) emulator focusing on high-resolution climate downscaling for the New Zealand region. For additional insights and access to the codebase utilized in this research, please refer to our Github Repository.

The code can also be found as a ".zip" file: *On-the-Extrapolation-of-Generative-Adversarial-Networks-for-downscaling-precipitation-extremes-main.

Aims

Our study focuses on two important gaps in the literature regarding the extrapolation of empirical downscaling algorithms. First, we examine how well relationships learned from a historical period extrapolate to future unobserved climates. We compare two widely used algorithms, a GAN and a deterministic CNN baseline, that use a similar architecture (i.e. convolutional layers) trained in a model-as-truth framework to downscale daily precipitation over New Zealand. We evaluate their accuracy in capturing climate change signals in mean and extreme precipitation. Second, we explore whether training on future vs. only historical periods combined with different-sized training datasets can improve extrapolation skill.

Geographic Focus

Our research focuses only on the New Zealand Region (165°E-184°W, 33°S-51°S).

Data Overview

Training and Evaluation Data

The training data used in this study (for our RCM emulator) spans the historical period and future period (SSP370) of simulation. It comprises daily accumulated precipitation as the primary target variable, alongside large-scale predictor variables.

• Resolution: The target variable is presented at a 12km resolution, reflecting the highest resolution face of RCM for the New Zealand region. Predictor variables are coarsened to a 1.5-degree resolution from original CCAM outputs using conservative interpolation.

• Period Coverage:

  • Training Data: 1960-2100 (Depending on Experiment, see Table 1 for list of experiment configurations)
  • Validation Data: 1985-2014 + 2070-2099 (to compute the climate change signal)

• Models:

  • Training on: ACCESS-CM2
  • Validated on: EC-Earth3, NorESM2-MM, CNRM-CM6-1, AWI-MR-1

File Structure

• Training Data:

  • Target/Ground Truth (Y): target_ACCESS-CM2_hist_ssp370_pr.nc
  • Predictor (X): predictor_ACCESS-CM2_hist_ssp370.nc

• Evaluation Data:
  All other GCMs can be accessed in one single file, predictor and target variables have the dimensions (time, lat, lon, GCM).

  • Target/Ground Truth (Y): Other_GCMs_hist_SSP370_target_fields_pr.nc
  • Predictor (X): Other_GCMs_hist_SSP370_predictor_fields.nc

Methodological Insights

• Regional Climate Model, Our Regional Climate Model training data is from the Conformal Cubic Atmospheric Model (CCAM) which is a global non-hydrostatic atmospheric model renowned for its variable-resolution cubic grid. . For more information about CCAM, please see the following paper.

• Predictor and Target Variables: Daily-averaged large-scale prognostic variables, including zonal wind, meridional wind, temperature, and specific humidity, are employed as predictors at the 500mb and 850mb pressure levels. These are normalized (see the GitHub repository for the mean and standard deviation fields). Precipitation is taken as is from CCAM and accumulated for each given day. Static predictors are also used in our model, which is stored in a GitHub repository.

• Training Framework: Our dataset benefits from the "perfect framework" training strategy, which uses CCAM-coarsened predictor variables. For more information about the perfect and imperfect training frameworks, see the following https://journals.ametsoc.org/view/journals/aies/3/2/AIES-D-23-0066.1.xml">review

Table 1: The six RCM emulator experiments performed in this study.

This dataset measures daily rainfall at 30 sites across Aotearoa New Zealand from 1960 to 2022.

Variables: site: location of monitoring station date: date rainfall: rainfall in mm rainfall_units: rainfall is measured in mm lat: Approximate latitude location of NIWA climate stations to represent a site. lon: Approximate longitude location of NIWA climate stations to represent a site. site_simple: site without macrons

DATA SOURCE: National Institute for Water and Atmospheric Research (NIWA) [Technical report available at https://www.mfe.govt.nz/publications/environmental-reporting/ministry-environment-atmosphere-and-climate-report-2020-updated]

Adapted by Ministry for the Environment and Statistics New Zealand to provide for environmental reporting transparency

Dataset used to develop the "Greenhouse gas concentrations" indicator [available at https://www.stats.govtnz/indicators/greenhouse-gas-concentrations]

This lowest aggregation dataset, was used to develop two ‘Our Atmosphere and Climate’ indicators. See Statistics New Zealand indicator links for specific methodologies and state/trend datasets (see ‘Shiny App’ downloads). 1) Rainfall (https://www.stats.govt.nz/indicators/rainfall) 2) Extreme rainfall (a. https://www.stats.govt.nz/indicators/extreme-rainfall

This dataset shows daily rainfall at 30 sites across New Zealand from 1960 to 2019.

More information on this dataset and how it relates to our environmental reporting indicators and topics can be found in the attached data quality pdf.

"Annual rainfall is the total accumulated rain over one year. Rain is vital for life, including plant growth, drinking water, river ecosystem health, and sanitation. Floods and droughts affect our environment, economy, and recreational opportunities.

This dataset shows annual average rainfall across New Zealand for 2005 as part of the data series for years 1972 to 2013. Annual rainfall is estimated from the daily rainfall estimates of the Virtual Climate Station Network (NIWA).

This dataset relates to the "Annual average rainfall" measure on the Environmental Indicators, Te taiao Aotearoa website.

Geometry: grid Unit: mm/yr"

This dataset is an augmentation of WRF forecasts over the Tairāwhiti region using a range of station observations.

Location and Time

The bounding latitudes and longitudes of the dataset are [177.90, 178.54, -39.03, -37.50], covering the Tairāwhiti region.

The dataset is available between March 1st 2024 and June 30th 2024.

Data sources

This dataset fuses data from multiple sources:

• Weather Research and Forecasting (WRF) model forecasts were obtained from the New Zealand Meteorological Service, and are at a 4km resolution. WRF is a weather forecast, rather than a reanalysis. We take the WRF forecast initialised most recently to the hour of interest, excluding the first six hours of any forecast to allow for a spin up period.
• National Climate Database (NCD) station observations were obtained through NIWA's Cliflo. There are 5 stations from the NCD producing hourly point-based precipitation observations.
• Station data from the Tethys database, including stations owned by Gisborne District Council and Fire and Emergency New Zealand. 60 sites are available in our region of interest at an hourly time period.
• Precipitation satellite data from NASA. This data is half hourly, so we resample to hourly data.
• Station data from automatic weather stations placed at schools in Tairāwhiti as part of the Mātaki Marangai project. There are 8 of these stations producing data every few minutes, so this data is resampled to be hourly. The raw data is available here.
• Rain gauges monitored by students during the Mātaki Marangai project. These data are quality checked and used to compare daily accumulations of adjusted WRF fields - more will be explained in the methodology section. The raw rain gauge data is available here.

Methodology

1. All data listed above are processed to become hourly data if it is not already.
2. WRF data are interpolated onto station observation points (that is, the data from the NCD, Tethys, and automatic weather stations).
3. The difference between the station observation and the WRF interpolated point is calculated, and a 2d delta field can be made at these points. This delta field can then be used to perform kriging interpolation (using the OrdinaryKriging Python tool) and adjust the gridded WRF dataset towards observations. This version of the adjusted dataset is called V1.
4. Not all locations have station observations, and so we use the satellite data to further improve the adjusted dataset in areas without stations. If there is a 0.2 degree square without site data, we take the satellite precipitation value for that square. This precipitation satellite data are calculated from cloud top temperatures, and is therefore of inferior quality to observations on the ground, but is useful in large regions without any ground observational data. We combine the satellite precipitation values and the station observations into a dataset and recalculate the 2d delta field over the region, and reperform kriging on the original WRF data (not the V1 dataset). We weight the station observations more than the satellite observations in the kriging process. This kriged dataset is called V2.
5. Finally, we compare the V2 dataset daily accumulations of precipitation to rain gauge observations. A daily 2d delta field is created and used to perform kriging interpolation once more. This final version of the dataset is called V3, and this is what is uploaded here.

Project Information

The Mātaki Marangai project was run by Bodeker Scientific, in collaboration with He Oranga Trust and the New Zealand Meteorological Service. The project was co-funded by New Zealand's Ministry of Business, Innovation and Employment (MBIE) Unlocking Curious Minds fund, and the MBIE Smart Ideas project DeepWeather.

Links to read more about the project:

• https://www.mataki-marangai.com/home
• https://www.bodekerscientific.com/projects/m%C4%81taki-marangai

Data for A History of Open Temperature and Rainfall with Uncertainty in New Zealand (HOTRUNZ): an open access 1 km resolution monthly 1910-2019 time-series of interpolated temperature and rainfall grids with associated uncertainty. The files use the open GeoTIFF geospatial data standard, that are compressed using the open 7z compression file format. There is data for both: nni: the variable estimate from natural neighbour interpolation unc: the uncertainty of the variable estimate for four weather variables: rain: total rainfall (mm) tavg: mean air temperature (°C) tmin: mean daily minimum air temperature (°C) tmax: mean daily maximum air temperature (°C) and these variables help form the file names alongside the year and month. So for example, nni-rain-1910-01.tif contains the natural neighbour interpolated values for total rainfall in January 1910, and unc-tavg-1979-05.tif contains the uncertainty values for mean air temperature in May 1979. For storage efficiency when archiving the data, all data value were multiplied by 10 so that they could be stored as 16 bit integer files. Therefore, all the data needs to be divided by 10 to restore the actual data values before the data is used for any analysis. For users outside of New Zealand who are struggling to download the large data files from here, these data files can also be downloaded from Zenodo at https://zenodo.org/record/5703749. There are also notes on how to source the input data used to create HOTRUNZ.

Precipitation in Ireland decreased to 1172.22 mm in 2024 from 1405.51 mm in 2023. This dataset includes a chart with historical data for Ireland Average Precipitation.

The annual number of rain days in the UK has fluctuated over the past three decades. In 2024, there were *** days in which * mm or more of rain fell. The year with the greatest number of rain days was 2000 when ***** days had at least * mm of rain. England is the driest country in the UK England is on average the driest country in the United Kingdom. In 2024, the country recorded an annual rainfall of **** mm. After England, Northern Ireland is the country that receives the least amount of rainfall across the UK. Wettest regions in Britain Despite Cardiff being the wettest city in the United Kingdom according to the Met Office, Scotland had received on average the largest volume of annual rainfall in the past 10 years. The northern and western regions of the UK – where rainfall is arriving from the Atlantic – tend to be the wettest in the country.

New Zealand Environmental Data Stack (NZEnvDS) comprises a set of 72 spatial layers for environmental modelling and site characterisation. NZEnvDS includes the layers that informed the Land Environments of New Zealand (LENZ), additional layers generated for LENZ that were never publicly released, and additional layers generated since, all covering mainland New Zealand and surrounding inshore islands. The original/top copy of the dataset is available at (https://doi.org/10.7931/m6rm-vz40), but the layers are reproduced here for ease of access. See (https://newzealandecology.org/nzje/3440/) for a publication describing NZEnvDS.

Generated as part of LENZ for the years 1950–1980 (Leathwick et al. 2002).

The precipitation layers weren’t originally included in LENZ but were generated as part of the project to create the LENZ vapour pressure deficit and annual water deficit layers.

Please cite the original paper as attribution when using these layers.

Rainfall sensors are currently deployed around the Burrishoole catchment to ensure maximum spatial coverage of rainfall variation. Data measured are mm of rain. Data coverage is for Furnace Newport Co. Mayo. Rainfall data collected since 1996. Rainfall data has been collected via tipping buckets which are used for the observations and measurements of rainfall. Data collected supports Met Eireann climate and meteorology analysis of rain over time periods. Data collected by the Marine Institute Newport Facility team in association with Met Eireann, Irelands meteorological service. Data has been collected regularly (e.g. daily) and is 100% complete for the period since the rainfall gauges deployed.

"Annual rainfall is the total accumulated rain over one year. Rain is vital for life, including plant growth, drinking water, river ecosystem health, and sanitation. Floods and droughts affect our environment, economy, and recreational opportunities.

This dataset shows annual average rainfall across New Zealand for 1998 as part of the data series for years 1972 to 2013. Annual rainfall is estimated from the daily rainfall estimates of the Virtual Climate Station Network (NIWA).

This dataset relates to the "Annual average rainfall" measure on the Environmental Indicators, Te taiao Aotearoa website.

Geometry: grid Unit: mm/yr"

Data for A History of Open Weather in New Zealand (HOWNZ): an open access 1 km resolution monthly 1910-2019 time-series of interpolated temperature and rainfall grids with associated uncertainty

There is data for both:

• nni: the variable estimate from natural neighbour interpolation
• unc: the uncertainty of the variable estimate

for four weather variables:

• rain: total rainfall
• tavg: mean air temperature
• tmin: mean daily minimum air temperature
• tmax: mean daily maximum air temperature

and these variables help form the file names alongside the year and month.

So for example, nni-rain-1910-01.tif contains the natural neighbour interpolated values for total rainfall in January 1910, and unc-tavg-1979-05.tif contains the uncertainty values for mean air temperature in May 1979.

For storage efficiency when archiving the data, all data value were multiplied by 10 so that they could be stored as 16 bit integer files. Therefore, all the data needs to be divided by 10 to restore the actual data values before the data is used for any analysis.