Hourly ground temperature data from a dynamically downscaled projection of past and future microclimates covering North America from 1980-1999 and 2080-2099

Ecological forecasting requires information about the climatic conditions experienced by organisms. Despite impressive methodological and computational advances, ecological forecasting still suffers from poor resolutions of environmental data. Published data comprise relatively few layers of surface climate and suffer from coarse temporal resolution. Hence, models using these data might underestimate heterogeneity of microclimates and miss biological consequences of climatic extremes. Moreover, we currently lack predictions about vegetation cover in future environments, a key factor for estimating the spatial heterogeneity of microclimates and hence the capacity for behavioral thermoregulation. Here, we describe microclimates and vegetation for the past and the future at spatial and temporal resolutions of 36 km (approximately 0.3°) and 1 h, respectively. We used the Weather Research and Forecasting model to downscale published, bias-corrected predictions of a global-circulation model from a resolution of 0.9° latitude and 1.25° longitude (approximately 100 km in latitude and 130 km in longitude). Output from this model was used as input for a microclimate model, which generated predictions for 19 variables for 1980-1999 and 2080-2099 at various heights, depths, sun angle and shade intensities. The data was evaluated using several criteria, each of which shed light on a different aspect of value to researchers. The metadata describe the modeling protocol, microclimate calculations, computer programs, and the evaluation process. The 19 predicted variables include albedo, snow layers, microclimate temperatures and pressures, among others. For a list of all variables please see the 'Model variables table' below.
The dataset is structured as follows:
(1) Main package: 19 monthly summaries, one for each microclimate variable (listed above) are available in this packagePackage structure schema/infographicR-script to extract and save NetCDF filesLocations table with latitude and longitude points covered in this data (csv)19 sub-packages (externally hosted, linked below) are available for this project, one for each microclimate variable.(2) Sub-packages: Within each sub-package are 44 tar files representing: 2 scenarios (past; future) across 22 geographical regions (see CoverageMap_Levy.png for distribution of regions)..tar file name template is [region]_[variable code]_[scenario]; i.e. B3_ISNOW_future.(3) .tar file: Each .tar file contains projection data in NetCDF format binary files for one region, one variable and for either past or future climates (1980-1999 and 2080- 2099).(4) NetCDF files
: Each NetCDF file is a time-series of data for a particular variable in one location (indexed by the longitudinal-latitudinal coordinates) for either past or future climates (1980-1999 and 2080-2099).Resolutions are of 36 km and 1 hour.
This sub-package contains past and future predictions of ground temperature in North America. In the microclimate model, ground temperatures (unit: K) were calculated at different shade densities (no shade, 25%, 50%, 75%, and full shade) by solving for the heat balance between surface fluxes and ground temperatures. Calculations were driven by outputs from the Weather Research and Forecasting model simulation. For more details, see Levy et al. (2016). There are 18 other sub-packages containing predictions for other variables, please see the main data package (doi:10.5072/FK2FX78N9G) for details and access.

生态预报需要获取生物体所经历的气候条件。尽管在方法学与计算技术上取得了长足进展，生态预报仍面临环境数据分辨率不足的难题。已公开的环境数据仅包含相对较少的地表气候图层，且时间分辨率较为粗糙。因此，基于此类数据构建的模型可能会低估微气候的异质性，同时忽略极端气候事件带来的生物学影响。此外，当前我们仍缺乏未来环境下植被覆盖的预测数据——而这正是估算微气候空间异质性、进而评估行为性体温调节能力的关键因素。

本数据集提供了过去与未来时段的微气候与植被数据，其空间分辨率为36 km（约0.3°），时间分辨率为1小时。我们采用天气研究与预报模型（Weather Research and Forecasting model，WRF），将全球环流模型（global-circulation model）公开的、经过偏差校正的预测结果从0.9°纬度×1.25°经度（纬度方向约100 km、经度方向约130 km）的分辨率进行降尺度处理。将该模型的输出作为微气候模型的输入，进而针对1980-1999年与2080-2099年两个时段，在不同高度、深度、太阳角度以及遮蔽强度下，生成19个变量的预测结果。

本数据集通过多项标准进行了评估，每项标准均可反映数据集在研究者关注的不同维度上的应用价值。元数据中详细描述了建模流程、微气候计算方法、计算机程序以及评估过程。这19个预测变量涵盖反照率、积雪层数、微气候温度与气压等。完整变量列表请参见下文的「模型变量表」。

本数据集的组织结构如下：
(1) 主数据包：本数据包包含19个月度汇总文件，分别对应上文提及的每一个微气候变量。包含数据包结构示意图/信息图、用于提取并保存网络通用数据格式（NetCDF）文件的R语言脚本、涵盖本数据集覆盖的经纬度点位的位置表（CSV格式）。本项目另有19个子数据包（外部托管，链接见下文），分别对应每一个微气候变量。
(2) 子数据包：每个子数据包内包含44个tar归档文件，分别对应2种情景（过去、未来）下的22个地理区域（区域分布详见CoverageMap_Levy.png）。tar文件的命名格式为`[区域]_[变量代码]_[情景]`，例如`B3_ISNOW_future`。
(3) tar归档文件：每个tar文件包含单个区域、单个变量、对应过去或未来气候时段（1980-1999年与2080-2099年）的NetCDF格式二进制投影数据。
(4) NetCDF文件：每个NetCDF文件为单个点位（以经纬度坐标索引）上特定变量的时间序列数据，对应过去或未来气候时段（1980-1999年与2080-2099年），分辨率为36 km与1小时。

该子数据包包含北美地区地表温度的过去与未来预测结果。在微气候模型中，研究人员通过求解地表通量与地表温度之间的热平衡方程，在不同遮蔽密度（无遮蔽、25%遮蔽、50%遮蔽、75%遮蔽以及全遮蔽）下计算得到地表温度（单位：开尔文）。计算过程以WRF模型的模拟输出为驱动。更多细节请参见Levy等人（2016）的研究。本数据集另有18个子数据包，涵盖其余变量的预测结果，详情与获取方式请参见主数据包（DOI:10.5072/FK2FX78N9G）。