Page 324 - 《高原气象》2026年第3期
P. 324
高 原 气 象 45 卷
932
原始空间特征, 并与 CLDAS 地表温度在陕北黄土 04. 005.
高原、 关中平原以及复杂地形区域的空间分布一致 Zhang H H, Loaiciga H A, Sauter T, 2024. A novel fusion-based
methodology for drought forecasting[J]. Remote Sensing, 16
性较高。
(5): 828. DOI: 10. 3390/rs16050828.
(2) 通过 SHAP 可解释性分析发现, 除 CL‐
Zhang P, Xu Z, Guan M, et al, 2022. Progress of Fengyun meteoro‐
DAS 地表温度(当日及前一日)外, NDVI 和纬度是 logical satellites since 2020[J]. Chinese Journal of Space Sci‐
影响融合模型输出的关键特征变量。CLDAS 地表 ence, 42(4): 724-732. DOI: 10. 11728/cjss2022. 04. yg14.
温度取值高低直接正相关地影响模型模拟的地表 Zhao T J, Wang S, Ouyang C J, et al, 2024. Artificial intelligence for
温度结果。高纬度、 低植被覆盖对地表温度模拟值 geoscience: Progress, challenges, and perspectives[J]. The Inno‐
vation, 5(5): 100691. DOI: 10. 1016/j. xinn. 2024. 100691.
的具有显著的正向贡献, 当纬度位于 35°N -36°N
Zhong Y M, Meng L K, Wei Z S, et al, 2021. Retrieval of all-weath‐
以北时, 纬度越高越有利于模型输出更高的地表温 er 1km land surface temperature from combined MODIS and AM‐
度值; NDVI 越小也越有利于模型输出更高的地表 SR2 data over the Tibetan Plateau[J]. Remote Sensing, 13(22):
温度值, 而当NDVI大于0. 2时, 贡献方向则相反。 4574. DOI: 10. 3390/rs13224574.
(3) 典型日期云下和全天候条件下的地表温 崔晓临, 程贇, 张露, 等, 2018. 基于 DEM 修正的 MODIS地表温度
产品空间插值[J]. 地球信息科学学报, 20(12): 1768-1776.
度融合结果都与相同条件下的 CLDAS 地表温度检
DOI: 10. 12082/dqxxkx. 2018. 180340. Cui X L, Cheng Y,
验效果相近。不同区域的表现与全省整体趋势基
Zhang L, et al, 2018. Spatial interpolation of MODIS land sur‐
本一致。地表温度融合结果在陕北地区的表现略 face temperature products based on DEM correction[J]. Journal
逊于关中和陕南地区, 后两者的表现较为接近且相 of Geo-information Science, 20(12): 1768-1776. DOI: 10.
对更好。区域差异可能受到云层分布格局、 地形复 12082/dqxxkx. 2018. 180340.
杂度以及模型对局部特征变量响应敏感性等多重 单帅, 沈润平, 师春香, 等, 2020. 中国北部积雪区冬季地表温度和
2m 气温再分析数据评估[J]. 高原气象, 39(1): 37-47. DOI:
因素的共同影响。
10. 7522/j. issn. 1000-0534. 2019. 00003. Shan S, Shen R P, Shi
参考文献(References): C X, et al, 2020. Evaluation of land surface temperature and 2m
air temperature from five reanalyses datasets across North China
Chen T Q, Guestrin C, 2016. XGBoost: A scalable tree boosting sys‐ in winter[J]. Plateau Meteorology, 39(1): 37-47. DOI: 10.
tem[C]. //Proceedings of the 22nd ACM SIGKDD International 7522/j. issn. 1000-0534. 2019. 00003.
Conference on Knowledge Discovery and Data Mining. New 丁利荣, 周纪, 张晓东, 等, 2023. 全天候地表温度遥感获取进展与
York, NY, USA: Association for Computing Machinery: 785- 挑战[J]. 遥感学报, 27(7): 1534-1553. DOI: 10. 11834/jrs.
794. https: //doi. org/10. 1145/2939672. 2939785. DOI: 10. 1145/ 20211323. Ding L R, Zhou J, Zhang X D, et al, 2023. Estima‐
2939672. 2939785. tion of all-weather land surface temperature with remote sensing:
Ghafarian Malamiri H R, Rousta I, Olafsson H, et al, 2018. Gap-fill‐ Progress and challenges[J]. National Remote Sensing Bulletin,
ing of MODIS time series land surface temperature (LST) prod‐ 27(7): 1534-1553. DOI: 10. 11834/jrs. 20211323.
ucts using Singular Spectrum Analysis (SSA)[J]. Atmosphere, 9 段四波, 茹晨, 李召良, 等, 2021. Landsat卫星热红外数据地表温度
(9): 334. DOI: 10. 3390/atmos9090334. 遥感反演研究进展[J]. 遥感学报, 25(8): 1591-1617. DOI:
Li Y X, Zhu S Y, Zhang G X, et al, 2024. Reconstruction of hourly FY- 10. 11834/jrs. 20211296. Duan S B, Ru C, Li Z L, et al, 2021.
4A AGRI land surface temperature under cloud-covered conditions Reviews of methods for land surface temperature retrieval from
using a hybrid method combining spatial and temporal information Landsat thermal infrared data[J]. National Remote Sensing Bulle‐
[J]. Remote Sensing, 16(10): 1777. DOI: 10. 3390/rs16101777. tin, 25(8): 1591-1617. DOI: 10. 11834/jrs. 20211296.
Lundberg S M, Lee S I, 2017. A unified approach to interpreting mod‐ 何坤龙, 赵伟, 刘晓辉, 等, 2021. 云雾覆盖下地表温度重建机器学
el predictions[C]. //Proceedings of the 31st International Confer‐ 习模型的训练集敏感性分析[J]. 遥感学报, 25(8): 1722-1734.
ence on Neural Information Processing Systems. Red Hook, NY, DOI: 10. 11834/jrs. 20211236. He K L, Zhao W, Liu X H, et al,
USA: Curran Associates Inc. : 4768-4777. DOI: 10. 5555/ 2021. Sensitivity analysis of the training set to the performance of
3295222. 3295230. the machine learning-based land surface temperature reconstruc‐
Mo Y P, Xu Y M, Chen H J, et al, 2021. A review of reconstructing tion for cloud covered pixels[J]. National Remote Sensing Bulle‐
remotely sensed land surface temperature under cloudy conditions tin, 25(8): 1722-1734. DOI: 10. 11834/jrs. 20211236.
[J]. Remote Sensing, 13(14): 2838. DOI: 10. 3390/rs13142838. 贾何佳, 李谢辉, 王磊, 等, 2022. 基于机器学习的西南地区遥感干
Zeng C, Long D, Shen H F, et al, 2018. A two-step framework for re‐ 旱监测与评估[J]. 高原气象, 41(6): 1572-1582. DOI: 10.
constructing remotely sensed land surface temperatures contami‐ 7522/j. issn. 1000-0534. 2022. 00006. Jia H J, Li X H, Wang L,
nated by cloud[J]. ISPRS Journal of Photogrammetry and Re‐ et al, 2022. Remote sensing drought monitoring and assessment
mote Sensing, 141: 30-45. DOI: 10. 1016/j. isprsjprs. 2018. in Southwestern China based on machine learning[J]. Plateau Me‐

