地球信息科学学报 ›› 2020, Vol. 22 ›› Issue (10): 2098-2107.doi: 10.12082/dqxxkx.2020.190423

• 遥感科学与应用技术 • 上一篇    下一篇

结合HASM和GWR方法的省级尺度近地表气温估算

周佳1,2(), 赵亚鹏2,3, 岳天祥2,3, 卢涛1,*()   

  1. 1.中国科学院成都生物研究所,中国科学院山地生态恢复与生物资源利用重点实验室,生态恢复与生物多样性保育四川省重点实验室,成都 610041
    2.中国科学院大学,北京 100049
    3.中国科学院地理科学与资源研究所 资源与环境信息系统国家重点实验室,北京 100101
  • 收稿日期:2019-08-05 修回日期:2019-12-09 出版日期:2020-10-25 发布日期:2020-12-25
  • 通讯作者: 卢涛 E-mail:zhoujia@cib.ac.cn;lutao@cib.ac.cn
  • 作者简介:周佳(1996— ),女,四川德阳人,硕士生,研究方向为地理信息系统。E-mail:zhoujia@cib.ac.cn
  • 基金资助:
    第二次青藏高原综合科学考察研究资助(SQ2019QZKK2002);国家重点研发计划项目(2016YFC0502101)

Near Surface Air Temperature Estimation by Combining HASM with GWR Model on a Provincial Scale

ZHOU Jia1,2(), ZHAO Yapeng2,3, YUE Tianxiang2,3, LU Tao1,*()   

  1. 1. Key Laboratory of Mountain Ecological Restoration and Bioresource Utilization & Ecological Restoration Biodiversity Conservation Key Laboratory of SichuanProvince, Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu 610041, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
    3. State Key Laboratory of Resourcesand Environment Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China
  • Received:2019-08-05 Revised:2019-12-09 Online:2020-10-25 Published:2020-12-25
  • Contact: LU Tao E-mail:zhoujia@cib.ac.cn;lutao@cib.ac.cn
  • Supported by:
    The Second Tibetan Plateau Scientific Expedition and Research Program, Grant(SQ2019QZKK2002);The National Key Research and Development Program of China(2016YFC0502101)

摘要:

卫星遥感反演得到的地表温度可用于近地表气温的估算,但现有方法的估算精度仍有进一步提升的空间。为了获取空间上连续且精度较高的近地表气温,本研究以四川省为例,首次将高精度曲面建模(HASM)用于遥感和气温实测数据的融合,并将综合了气温、地表温度、海拔、坡度、坡向的地理加权回归(GWR)拟合结果作为HASM模型的初始温度场,进而采用此种结合HASM和GWR的求解算法(HASM-GWR),融合MOD11C3地表温度产品与190个气象台站的气温实测数据,开展省级尺度近地表气温估算,并通过比较HASM-GWR、GWR以及普通线性回归(OLS)3种方法的估算精度,评估各模型对近地表气温的估算效果。结果表明,相比于传统估算模型,采用HASM-GWR数据融合方法能有效提高近地表气温的估算精度。采用该方法的近地表气温估算残差,72%介于-1~1 ℃,90%介于-2~2 ℃;且与GWR和OLS模型相比,估算结果的均方根误差(RMSE)分别降低了25.42%和39.83%。

关键词: 近地表气温, 卫星遥感数据, 高精度曲面模型, 地理加权回归模型, 普通线性回归模型, 数据融合, 精度, 四川省

Abstract:

As a part of natural climate variability,Near surface air temperature is an indispensable parameter that drives the energy and water exchanges among the hydrosphere, atmosphere and biosphere. Spatially and temporally resolved observations of near surface air temperatures are essential for understanding hydrothermal circulation at the land-atmosphere interface, and have had significant ecological impacts on many parts of the natural ecosystems. Given the ecological significance of near surface air temperature, the demand for accurate spatial data has risen greatly. Unfortunately, the gridded near surface air temperature data is generally limited by station coverage of meteorological observations, especially in extensive mountainous regions. Moreover, the uneven spatial distribution of meteorological stations may not effectively capture the true nature of the overall climate pattern. Given the strong correlation between land surface temperature and near surface air temperature, recent efforts have developed an alternative method for retrieving spatially continuous near surface air temperature from satellite-derived land surface temperature data sets. However, the degree of accuracy for current applications in near surface air temperature estimation still has a large room for improvement. Here we introduce a novel approach that combines High Accuracy Surface Modeling (HASM) with Geographically Weighted Regression (GWR) model for improving estimation of near surface air temperature in a data-fusion context. In this approach, application of fusion methods using Moderate Resolution Imaging Spectroradiometer (MODIS) products and ground-based observations was used. By fusing the MOD11C3 land surface temperature products and the air temperature data observed at 190 meteorological stations in Sichuan province, this study combines HASM with GWR model for improving estimation of near surface air temperature. To assess the feasibility of this modified model, we use 175 stations for model development and reserve15 for validation tests with three repetitions. The performance of combining HASM with GWR model (HASM-GWR) is also compared with multifactorial Geographically Weighted Regression (GWR) and Ordinary Least Squares (OLS) methods. The results indicated that the best estimation was found in HASM-GWR model. Specifically, the validation results from HASM-GWR model show that 72% of the estimated residual error is between -1 ℃ and 1 ℃, 90% is between -2 ℃ and 2 ℃, and the Root Mean Square Error (RMSE) reduces by 25.42% and 39.83% in comparison with other techniques. In addition, the near surface air temperature map obtained from HASM-GWR is better than that obtained by using other methods. Therefore, the proposed HASM-GWR model demonstrated an effective proficiency in near surface air temperature estimation, and it can be seen as an alternative to the popular data fusiontechniques.

Key words: near surface air temperature, remote sensing data, HASM, GWR, OLS, data fusion, accuracy, Sichuan province