以空间异质性较强的枯水期鄱阳湖为研究区,以搭载于同一卫星平台、具有同一观测时间和较高空间分辨率的ASTER数据为参照,分析研究了MODIS数据在土地覆盖分类中由空间尺度带来的不确定性。首先基于MODIS三角权重函数,建立了从ASTER到MODIS的尺度转换方法;然后对不同空间分辨率的数据进行土地覆盖分类,并基于误差矩阵和线性模型分析了MODIS土地覆盖分类结果的误差来源。结果表明,空间分辨率和光谱分辨率与成像方式这两类因素对MODIS与ASTER分类结果差异的贡献比例约为(6.6—11.2):2;MODIS像元尺度对研究区水体的分类不确定性影响较低,而对森林的不确定性影响可达63%。由此可见,在基于MODIS数据的土地覆盖分类研究中,空间尺度所产生的不确定性是比较显著的。这些研究结果对于土地覆盖分类及变化检测、尺度效应和景观生态学不确定性研究,有积极的参考意义。

Moderate-resolution Imaging Spectroradiometer (MODIS) data have been widely used in land cover classification with its advantages in multi-spectral and high-temporal features. The classification may suffer from uncertainty relevant to its moderate spatial resolution, yet the uncertainty remains unclear. To explore the issue, this paper used high spatial resolution data acquired from the Advanced Space borne Thermal Emission and Refl ection Radiometer (ASTER). Since ASTER and MODIS are onboard the same satellite platform, which allows simultaneous multi-resolution observation at coincident nadirs. The Poyang Lake area was taken as the study area for its diverse land covers in low-water period. The high-resolution ASTER data were upscaled to the same coarse resolution as MODIS, with a scaling function developed from a triangular spread function. Different classification methods were applied respectively for land cover classification using either MODIS or upscaled ASTER data. The uncertainties in classified results were subsequently analyzed based on error matrix and a linear model. The evaluation revealed that the uncertainties contributed from scale difference and that from differences in spectral resolution and imaging mode were (6.6—11.2):2. The scale-induced uncertainty also showed to be land-cover dependent. It was small for water body, but could be as large as 63% for forestland. The signif icance of scale effect is valuable for land cover classification, change detection analysis, scale effect evaluation, and uncertainty analysis in landscape ecology.