论文部分内容阅读
森林覆盖区积雪的提取精度很低,由于植被冠层的遮挡,冠层下的积雪很难被提取出来。基于Landsat 8OLI数据,针对玛纳斯河流域下游有大面积森林覆盖的特点,通过传统的积雪指数法,结合NDVI数据的积雪指数法和面向对象图像特征法分别提取积雪面积。结果表明:1传统的NDSI和S3积雪指数法无法较好地提取出森林覆盖下的积雪,提取精度分别为85.23%和87.54%。这两种方法适用于空间尺度较大、植被覆盖面积较大的区域,并不适合所选研究区;2结合NDVI数据后的NDSI、S3积雪指数模型能大大提高森林覆盖下的积雪面积,提取精度分别达到91.47%和90.60%。在影像空间分辨率较高,流域尺度较小,林区覆盖较多的情况下可采用此方法提取积雪;3随着海拔的升高,地形阴影影响逐渐增大,NDVI辅助积雪指数方法提取林区覆盖下积雪面积逐渐减小。因此采用光谱、纹理和空间信息结合的面向对象图像特征方法提取积雪,能够较好地识别出受地形影响下的雪像元,精度达到89.75%,可以满足实际应用的需求。
The accuracy of extraction of snow in the forest coverage area is very low. Because of the cover of the vegetation canopy, snow under the canopy can hardly be extracted. Based on the Landsat 8OLI data, aiming at the large area of forest coverage in the lower reaches of the Manas River, the snow area was extracted by the traditional snow index method, the snow index method combined with the NDVI data and the object-oriented image feature method. The results show that: 1 traditional NDSI and S3 snow index method can not extract the snow cover under the forest cover, the extraction accuracy is 85.23% and 87.54% respectively. These two methods are suitable for the area with large space scale and large area of vegetation coverage and are not suitable for the selected study area.2 NDVI and S3 snow index model combined with NDVI data can greatly improve the snow cover area under the forest cover , The extraction accuracy reached 91.47% and 90.60% respectively. This method can be used to extract snow cover when the image spatial resolution is high, the basin scale is small, and the forest area is covered more. 3 As the elevation increases, the impact of the terrain shadow gradually increases. NDVI assisted snow index method The snow cover area under the cover of the extracted forest area gradually decreased. Therefore, using the object-oriented image feature extraction method combined with spectral, texture and spatial information to extract snow, the snow pixels under the influence of the terrain can be well identified. The accuracy reaches 89.75%, which can meet the practical application needs.