应用中值融合模型的条件植被温度指数降尺度转换研究

Down-scaling Transformation of Vegetation Temperature Condition Index Using Median Fusion Model

为获得基于Landsat卫星遥感数据更为精确的定量化干旱监测结果,以陕西省关中平原为研究区域,基于Aqua MODIS数据反演的1 km空间分辨率的条件植被温度指数(VTCI)的定量化干旱监测结果(MODIS-VTCI)和Landsat OLI/TIRS数据反演的30 m空间分辨率的VTCI相对干湿监测结果(Landsat-VTCI),应用降尺度的中值融合模型(MFM)将基于MODIS数据反演的VTCI降尺度至30 m空间分辨率的VTCI定量化干旱监测,并对其结果进行验证。结果表明,应用降尺度的中值融合模型转换的VTCI定量化干旱监测结果(MFM-VTCI)与Landsat-VTCI的空间分布及纹理特征相似,两者间的相关系数和结构相似度均较大,均方根误差、差值影像及差值频数分布图所呈现的结果与定量化干旱监测结果和相对干湿监测结果间的系统误差相符,表明Landsat-VTCI与MFM-VTCI间的可比性较强。MFM-VTCI与累计降水间的相关性和MODIS-VTCI与累计降水间的相关性相近,均大于Landsat-VTCI与累计降水间的相关性,表明MFM-VTCI是定量化的干旱监测结果。

Vegetation temperature condition index（VTCI） is proved to be a quantitative drought monitoring approach by using the high temporal resolution remotely sensed data.However,with low temporal resolution data,the monitoring results are relatively wet and dry.A new model called the median fusion model（MFM） was developed for spatially down-scaling the coarse spatial quantitative VTCI（1 km） derived from the moderate-resolution imaging spectroradiometer（MODIS） data products and the relative VTCI（30 m） derived from the Landsat OLI/TIRS products in the Guanzhong Plain,China to a scale of the quantitative drought monitoring results（30 m） called MFM-VTCI,and their quantifications were proved.The results showed that the good agreements between the MFM-VTCIs and the Landsat-VTCIs were found in terms of correlation coefficient and structural similarity index（SSIM）values,and the two VTCIs had similar spatial distribution and texture features.The root mean square error（RMSE） and the differences between the MFM-VTCIs and the Landsat-VTCIs were consistent with the systematic error between the quantitative drought monitoring results and the relatively wet and dry monitoring results,indicating that it was comparable between the MFM-VTCIs and the Landsat-VTCIs.The correlation coefficients between the MFM-VTCIs and the cumulative precipitation were similar to those between the MODIS-VTCIs and the cumulative precipitation,which were larger than those between the Landsat-VTCIs and the cumulative precipitation,indicating that the down-scaled MFM-VTCIs were quantitative drought monitoring results.