A dataset of sea breeze objects over Australia (1979-2024)

Overview This record contains a dataset of sea breeze objects derived from the Bureau of Meteorology Atmospheric Regional Reanalysis for Australia, version 2, with convective-scale configuration (BARRA-C2, Su et al., 2025), over a 1979-2024 period. BARRA2 is made available by the Bureau of Meteorology under a CC-BY 4.0 license via the Australian National Comptutational Infrastructure Sea breeze objects are defined on a time-latitude-longitude grid, at around 4.4 km horizontal spacing at hourly intervals. The objects are provided as a binary mask on a grid in .zarr format. Both hourly and daily masks are provided here and compressed into .zip archives. sea_breeze_mask_hourly.zip: Contains 1,978 .zarr directories of hourly sea breeze objects, with each directory containing no more than 10 days of objects. sea_breeze_mask_daily.zip: Contains three .zarr directories with daily sea breeze masks over the whole 1979-2024 period. Object identification method and software Sea breeze obects are defined using a moisture frontogenesis approach based on surface winds and specific humidity, with further filtering of objects using a set of morphological and physical characteristics. Filtering settings are provided in the .zarr metadata. Details on the methods can be found in Brown et al. (2025a). Code to produce the objects, and accomanying documentation, can be found in Brown et al. (2025b) or on GitHub. Daily data resampling The daily mask provided here has been resampled at each grid point from hourly data, by considering if a sea breeze is present on a given day. A day is either defined by UTC, Australian Eastern Standard Time (AEST, UTC + 10), or Australian Western Standard Time (AWST, UTC + 8), and each are provided as separate .zarr directories: UTC: sea_breeze_mask_utc.zarr AEST: sea_breeze_mask_aest.zarr AWST: sea_breeze_mask_awst.zarr For the AEST and AWST data, sea breeze objects occurring before 0900 or after 2100 are not included in the resampling. Opening the data After unzipping, the .zarr directories can be loaded into Python using xarray following the code below. This was tested with the following python packages: xarray version 2023.2.0 zarr version 2.13.3 import xarray as xr f = xr.open_dataset("sea_breeze_mask_utc.zarr", engine="zarr") Additional details: Spatial extent: 108.02E to 158.50E, 45.69S to 6.93S Temporal extent: 1979-01-01 to 2024-12-31 Time frequency: Daily and hourly概览 本数据集收录了基于澳大利亚气象局（Bureau of Meteorology）大气区域再分析第二版（BARRA2）的对流尺度配置（convective-scale configuration，BARRA-C2，Su et al., 2025）提取的海风目标数据集，时间跨度为1979年至2024年。BARRA2数据集由澳大利亚气象局以CC BY 4.0许可协议发布，通过澳大利亚国家计算基础设施（Australian National Computational Infrastructure）提供。 海风目标基于时间-纬度-经度网格定义，水平分辨率约为4.4千米，逐小时输出。数据集以.zarr格式的二进制掩码（binary mask）形式提供，同时包含逐小时与逐日掩码文件，并已打包为.zip压缩归档。 - sea_breeze_mask_hourly.zip：包含1978个.zarr目录，存储逐小时海风目标数据，每个目录内含不超过10天的目标数据。 - sea_breeze_mask_daily.zip：包含3个.zarr目录，涵盖1979-2024年全时段的逐日海风掩码数据。 海风目标识别方法与软件 本数据集采用基于地表风场与比湿的水汽锋生（moisture frontogenesis）算法定义海风目标，并通过一系列形态学与物理特征对目标进行二次筛选。筛选参数已存储于.zarr文件的元数据（metadata）中。相关方法细节可参见Brown等人（2025a）的研究成果；数据集生成代码与配套文档可从Brown等人（2025b）或GitHub平台获取。 逐日数据重采样说明 本数据集提供的逐日掩码是通过对逐小时数据进行逐网格重采样得到的，重采样逻辑为判断当日是否存在海风。逐日数据采用三种时区定义：协调世界时（UTC）、澳大利亚东部标准时间（AEST，UTC+10）与澳大利亚西部标准时间（AWST，UTC+8），并分别存储为独立的.zarr目录： - UTC：sea_breeze_mask_utc.zarr - AEST：sea_breeze_mask_aest.zarr - AWST：sea_breeze_mask_awst.zarr 针对AEST与AWST数据集，重采样时将排除09:00前与21:00后出现的海风目标。 数据加载方法 解压压缩包后，可通过Python的xarray库加载.zarr目录，示例代码如下。本代码已在以下Python包版本中测试通过： - xarray 2023.2.0 - zarr 2.13.3 python import xarray as xr f = xr.open_dataset("sea_breeze_mask_utc.zarr", engine="zarr") 补充说明 - 空间范围：108.02°E至158.50°E，45.69°S至6.93°S - 时间范围：1979年1月1日至2024年12月31日 - 时间频率：逐小时与逐日