高分辨红外辐射探测器地表温度数据在江苏地区1980—2009年间适用性评估

地表温度在物理和生物过程中起着关键作用,也是评价地表热环境的重要指标。因此,了解地表温度时空变化对城市热岛监测及生态质量的评价具有重要意义。高分辨红外辐射探测器地表温度(HIRS LST)是目前时间尺度最长的全球逐小时地温数据集。为了解江苏省地表温度的时空分布情况并研究HIRS LST数据在江苏的适用性,论文选择江苏省1980—2009年49个站点的实测地表温度数据,利用相关系数、偏差、非偏性均方根误差等方法,对HIRS产品从多角度进行了验证。结果表明,HIRS地表温度与站点地表温度数据有较好的一致性。两者相关系数在整个区域都高于0.98,2种数据的距平相关系数在0.65~0.80之间。两者偏差和非偏性均方根误差表明,HIRS的数据在江苏北部和南部部分地区低估了地表温度,主要原因是其对高于32 ℃的地温事件发生次数存在较大程度的低估。然而,HIRS LST在很大程度上高估了在20~30 ℃之间的较高温日数。在年际变化方面,HIRS LST与观测数据在春季的相关性最高,冬季最低。趋势检验表明,2种数据在春、秋、冬3个季节均呈现出明显的增长趋势,增温趋势呈现出相似的空间变化。但是,该地区夏季的地表温度长期趋势被明显高估,HIRS数据并未反映出该地区北部大面积的降温趋势,而在其他季节的地表温度被低估。

Evaluation of the long-term high-resolution infrared radiation sounder land surface temperature during 1980-2009 in Jiangsu Province,China

Surface temperature plays a key role in physical and biological processes on Earth, and it is an important index for evaluating surface thermal environment. Understanding the temporal and spatial variations of surface temperature is of great significance for urban heat island monitoring and ecological quality evaluation. Current land surface temperature products derived from satellite remote sensing suffer from partial coverage or cloud-cover blockage problems, which have potential limitations on the study of climate and ecological environment. The recently developed NOAA satellites High-Resolution Infrared Radiation Sounder (HIRS) land surface temperature (LST) product is among the longest LST records. To examine the spatiotemporal distribution of LST in Jiangsu Province and data performance of HIRS LST at different temporal scales, we evaluated the HIRS LST and in situ measurement correlation coefficient (R), bias, and unbiased root mean square difference (ubRMSD) using the daily, annual, and seasonal mean values, together with the long-term linear trend during 30 years in the province. Great consistency between the two products is observed. Their correlation coefficients are higher than 0.98 for all stations, and those for daily anomalies range from 0.65 to 0.80 across the region. The bias and the ubRMSD indicate that the HIRS data have generally underestimated the LST across the northern and some southern areas, mainly because of its large underestimation of the occurrences of temperature higher than 32 ℃. Nevertheless, the HIRS LST has largely overestimated the occurrences of summer days with temperature ranging from 20-30 ℃. As for the intraannual variations, the HIRS LST shows highest correlation with in situ measurements in the spring, but the correlation is lowest in the winter. Trend test shows that both sets of data show significant increasing trends with similar patterns in the spring, autumn, and winter. However, the long-term trends are significantly overestimated across the region in the summer, and underestimated in other seasons in the HIRS LST data.