【摘要】：針對衛(wèi)星遙感技術(shù)監(jiān)測地表溫度(land surface temperature,LST)存在時空分辨率矛盾這一難題,以TsHARP溫度降尺度算法為基礎(chǔ),根據(jù)地表覆蓋類型的不同,分別選擇與LST相關(guān)性更好的光譜指數(shù)(歸一化植被指數(shù),NDVI;歸一化建造指數(shù),NDBI;改進的歸一化水體指數(shù),MNDWI;增強型裸土指數(shù),EBSI)提出了新的轉(zhuǎn)換模型,并從定性和定量兩個角度評價了TsHARP法和新模型的降尺度精度。結(jié)果表明:兩種模型在提高LST空間分辨率的同時又能較好地保持MODIS LST影像熱特征的空間分布格局,消除了原始1km影像中的馬賽克效應(yīng),兩種模型均能夠達到較好的降尺度效果;全局尺度分析表明,不管是在降尺度結(jié)果的空間變異性還是精度方面,本文提出的模型(RMSE:1.635℃)均要優(yōu)于TsHARP法(RMSE:2.736℃);TsHARP法在水體、裸地和建筑用地這些低植被覆蓋區(qū)表現(xiàn)出較差的降尺度結(jié)果,尤其對于裸地和建筑用地更為明顯(|MBE|3℃),新模型提高了低植被覆蓋區(qū)地物的降尺度精度;不同季節(jié)的降尺度結(jié)果表明,兩種模型都是夏、秋季的降尺度結(jié)果優(yōu)于春、冬季,新模型的降尺度結(jié)果四季均好于TsHARP法,其中春、冬季的降尺度精度提升效果要優(yōu)于夏、秋季。
[Abstract]:In view of the difficult problem of spatial and temporal resolution contradiction in monitoring surface temperature (land surface temperature,LST) by satellite remote sensing technology, based on the TsHARP temperature downscaling algorithm, according to the different types of land cover, The spectral index (normalized vegetation index, NDVI;) with better correlation with LST was selected respectively. The normalized construction index, NDBI; improved normalized water index and MNDWI; enhanced bare soil index, EBSI) proposed a new transformation model, and evaluated the downscaling accuracy of TsHARP method and new model from both qualitative and quantitative aspects. The results show that the two models can not only improve the spatial resolution of LST but also maintain the spatial distribution pattern of thermal characteristics of MODIS LST images, and eliminate the mosaic effect in the original 1km images. Both models can achieve a better downscaling effect. The global scale analysis shows that the proposed model (RMSE:1.635 鈩，

本文編號：2385194