A soil moisture dataset over the Brazilian semiarid region

This dataset contains soil moisture measured over the Brazilian Semiarid region from August 2015 to April 2019. An observational network of automatic ground stations was established by the National Center for Monitoring and Early Warning of Natural Disasters (CEMADEN) with the aim of supporting monitoring activities over the driest and most drought-affected region in the country. The sites are mostly characterised by family farming and rainfed agriculture. Soil volumetric water content is obtained by measuring the dielectric constant of the soil at 70 MHz frequency, which minimizes salinity and textural effects (model EC-5, Decagon Devices, Pullman, WA, USA). Probes use the default calibration, resulting in a precision of approximately ±0.03 m3 m-3. Measurements were made hourly at 0.1 m and 0.2 m depths; a subset of stations (10% of the original network, evenly distributed) had measurements at 0.1 m, 0.2 m, 0.3 m, and 0.4 m depths. Rainfall measurements are also included in this version of the dataset. Measurements were accumulated hourly. A total of 360 stations were selected after a quality control process. Data availability is approximately 2 years long on average, reaching up to 4 years in some stations. Gaps caused by instrument malfunction or data transmission problems are occasional on the data. Validation and consistency were assessed by comparing trends in soil vertical neighbour data and rainfall records collected by each station. Stations with at least two months of soil moisture data were kept on the database. Each file available here contains the time series and the metadata with information on the city, the Brazilian state, the geographic coordinates, and the soil texture (fraction of sand, silt and clay). Soil information was derived from 1 kg disturbed samples collected at each site. Data for individual stations are in the compressed file PCDs_.zip while a list summarising stations information is found in the file _Stations.xlsx. First applications of this dataset and further details on methodology and site can be found in Zeri et al. (2018, https://doi.org/10.3390/w10101421). Version 2 of the dataset includes rainfall measurements, 3 new stations, and the inclusion of records incorrectly removed during the quality control of version 1.本数据集收录了2015年8月至2019年4月期间巴西半干旱地区的土壤湿度观测数据。国家自然灾害监测与预警中心（National Center for Monitoring and Early Warning of Natural Disasters，简称CEMADEN）搭建了自动地面观测站网络，旨在为该国最干旱、受旱灾影响最严重的区域提供监测支撑。观测站点所在区域多以家庭农业和雨养农业为主要生产模式。 土壤体积含水量通过测定70MHz频率下的土壤介电常数获取，该方案可最大限度降低土壤盐度与质地对测量结果的影响，所用设备为美国华盛顿州普尔曼市Decagon Devices公司生产的EC-5型传感器。传感器采用默认校准参数，测量精度约为±0.03立方米每立方米。多数站点在0.1米和0.2米深度开展每小时一次的土壤湿度测量；另有占原网络10%的均匀分布站点，额外增设了0.3米、0.4米深度的观测。本版数据集同时包含降雨观测数据，降雨数据以小时累计值形式存储。 经质量控制流程筛选后，最终纳入本数据集的地面观测站共计360个。数据集的平均有效观测时长约为2年，部分站点的有效观测时长可达4年。数据中偶发因仪器故障或数据传输异常导致的观测缺失。研究人员通过对比站点的土壤垂直分层观测趋势与同期降雨记录，完成了数据的有效性与一致性校验。仅保留了至少拥有2个月有效土壤湿度观测数据的站点。 本平台提供的每份数据文件均包含土壤湿度时间序列数据与元数据信息，元数据涵盖站点所在城市、巴西所属州、地理坐标，以及土壤质地参数（砂粒、粉粒与黏粒占比）。土壤质地参数源自各站点采集的1kg扰动土壤样品分析结果。单站点的观测数据存储于压缩文件PCDs_.zip中，而站点信息汇总列表则存放在_Stations.xlsx文件内。本数据集的首批应用案例及观测方法、站点详情可参考Zeri等人2018年发表的研究（https://doi.org/10.3390/w10101421）。本数据集第二版新增了降雨观测数据、3个新站点，并补全了第一版质量控制流程中被错误剔除的观测记录。