Post-processed climate dataset for the Sierra Nevada de Santa Marta (SNSM) Region in Colombia

Monthly cleaned gap-filled climate dataset, derived from raw daily data from the Institute of Hydrology, Meteorology and Environmental Studies (IDEAM by its Spanish name), for a 30-year climate period, from January 1, 1989, to December 31, 2018. This dataset contains monthly precipitation (prec) and monthly maximum (tmax) and minimum (tmin) temperature as tabular time series (CSV) as well as spatial surfaces (GeoTIFF) interpolated at 1 km of spatial resolution. In addition, 1-km spatial surfaces of potential evapotranspiration (pet) are also provided. METHODOLOGY IDEAM stations within a buffer distance of 15 km of the study area were selected, resulting in 10 stations with daily precipitation data and 5 with daily maximum and minimum temperature data that were subject to quality control. Data were cleaned to remove values out of range, statistical outliers, repetitive measurements during five days, and measurements that changed excessively between two consecutive days following the procedures of Esquivel et al. (2018). Gaps in the daily weather data were filled using the nearest neighbor algorithm bounded by elevation (de Lima et al., in review). The gap-filled data was aggregated to a monthly time-step and consecutively monthly surfaces were generated for each year through spatial interpolations at 1 km of resolution (approx. 30 arc-seconds) by using the thin-plate spline smoothing method from ANUSPLIN version 4.4, which in turn, uses elevation as a covariate (Hutchinson and de Hoog, 1985). Five additional synthetic stations per variable were created for the interpolation based on the weather stations. Finally, monthly surfaces at 1 km of spatial resolution were generated using the modified equation of Hargreaves stated by Droogers & Allen (2002).

本数据集为经清洗与缺失值填充的气候数据集，其原始日数据取自水文、气象与环境研究所（西班牙文缩写为IDEAM），时间跨度为1989年1月1日至2018年12月31日，覆盖30年气候周期。
本数据集包含以表格时间序列（CSV格式）存储的月降水量（prec）、月最高气温（tmax）与月最低气温（tmin）数据，以及空间分辨率为1千米的插值空间表面（GeoTIFF格式）。此外，本数据集还提供空间分辨率为1千米的潜在蒸散量（pet）空间表面数据。

研究方法：选取研究区周边15千米缓冲范围内的IDEAM气象站点作为数据源，最终筛选出10个具备日降水量数据的站点与5个具备日最高、最低气温数据的站点，并对这些站点数据开展质量控制。依据Esquivel等人（2018）的研究流程，对原始数据开展清洗工作，剔除超出合理范围的数值、统计异常值、连续五日的重复观测值，以及连续两日之间发生异常突变的观测数据。基于以高程为约束条件的最近邻算法，对逐日气象数据中的缺失值进行填充（de Lima等，待刊）。将完成缺失值填充的日尺度数据聚合为月尺度时间序列，随后采用ANUSPLIN 4.4版本的薄板样条平滑法，以高程作为协变量（Hutchinson与de Hoog，1985），以约30角秒（约1千米）的空间分辨率开展空间插值，逐年生成各年份的月尺度气候空间表面。基于实测气象站点，为每个气象变量额外构建5个虚拟站点以辅助插值。最终，采用Droogers与Allen（2002）修正后的Hargreaves公式，生成空间分辨率为1千米的月尺度潜在蒸散量空间表面。