Data from: Predicting ground temperatures across European landscapes

1. Ambient temperatures in natural environments can vary widely over short distances, especially on rugged ground where exposure to solar radiation depends on slope and aspect. Temperatures can also fluctuate rapidly, reaching values not revealed by climate data. Such fine-scale variation raises challenges for modelling species distributions under changing climates. 2. To avoid misunderstanding current species distributions and future changes, temperatures must be modelled at high resolutions. Most existing methods either require extensive parameterisation or are pre-parameterised for restricted localities and current conditions; here we describe a more versatile method intended for European landscapes under a wide range of scenarios. 3. The availability of high-resolution topographic data makes possible the use of projected solar irradiation to help predict local diurnal ground temperatures. Using time series from 83 points across Europe, we fitted statistical models for soil surface temperature based on geographical characteristics of the sites along with atmospheric variables obtained from a publicly-available database. The sites ranged from 40°N to 60°N and 2°W to 25°E, and from 43m to 1500m in elevation. 4. We compare models for monthly mean and daily afternoon temperatures, for open and tree-covered habitats. The effect of topography was greatest for the daily predictions, and generally more important in open than in tree-covered sites. Tests with data collected from other European locations in a different time period suggest that our models can predict monthly means with an error standard deviation of 2.7°C. We provide an R function that implements our models on the basis of readily-available data. 5. Our mean-temperatures model should be useful for understanding organisms' niches, dispersal possibilities and community dynamics, and for predicting species' refugia, range shifts and opportunities for adaptation under projected climate change.

1. 自然环境中的环境温度在短距离内可出现大幅波动，在崎岖地形中尤为显著——此类区域的太阳辐射（solar radiation）暴露量取决于坡度与坡向。温度亦可能快速起伏，出现气候数据未收录的极端数值。这类精细尺度的温度变异性，为气候变化背景下的物种分布建模带来了诸多挑战。
2. 若要避免对当前物种分布及未来变化产生误判，必须以高分辨率开展温度建模工作。现有多数建模方法要么需要开展大量参数化调试，要么仅针对特定区域与当前气候条件完成预参数化；本文介绍一种通用性更强的方法，适用于多种情景下的欧洲陆地景观。
3. 高分辨率地形数据的普及，使得借助预估太阳辐射（projected solar irradiation）预测局地昼夜地表温度成为可能。研究团队利用欧洲范围内83个测点的时间序列数据，基于各站点的地理特征以及从公开数据库获取的大气变量，拟合得到地表温度的统计模型。这些测点的纬度跨度为40°N至60°N、经度跨度为2°W至25°E，海拔范围介于43米至1500米之间。
4. 我们分别针对开阔生境与林木覆盖生境，比对了月均温度与每日午后温度的建模效果。地形效应在每日温度预测中表现最为显著，且总体而言在开阔生境中的影响强于林木覆盖生境。利用欧洲其他区域不同时段采集的独立数据开展测试，结果显示本模型可用于预测月均温度，其预测误差的标准差为2.7℃。我们还提供了一款基于易得数据即可运行模型的R函数（R function）。
5. 本研究所开发的月均温度模型，可用于解析生物的生态位、扩散潜力与群落动态，亦可用于预测气候变化情景下物种的气候避难所、分布范围迁移路径以及潜在适应机遇。