基于数字相机拍摄影像的山区积雪消融动态观测研究—以天山积雪站为例

在山区,基于卫星影像的积雪监测很难获得高时间和高空间分辨率的数据。而山区复杂地形也降低了传统外推方法对积雪深度外推的精度。本研究在2011年于中国科学院天山积雪与雪崩研究站,利用数字相机对阳坡积雪面积进行监测。并在30mDEM的基础上校正影像得到融雪期研究区域积雪的时空变化。同时,利用山坡预设的雪尺监测阳坡积雪深度,使用分类回归树算法外推得到研究区内的积雪深度。在此基础上,评估不同温度变量驱动下SRM模型的融雪模块精度。研究表明,在山区使用数字相机拍摄研究积雪的时空变化是可行的。相对于卫星影像,在时间和空间分辨率显著提高。但仍然存在一些不足。分类回归树外推得到的研究区域内的积雪深度精度较高,以日最高气温和日小时积温驱动下SRM可以基本识别在日平均气温低于0℃下的融雪过程。

Observation and Analysis of the Process of Snowmelting at Tianshan Station Using the Images by Digital Camera

Snow cover is an essential part of the global cryosphere.With global climate change,the area of snow cover has shrunk seriously,especially in the Alpine area.Tianshan Mountain is the origin of the main rivers in central Asia and Xinjiang and seasonal snow cover is one of the primary water sources in the region.Therefore,the snow accumulation and melting process is of great significance to water management and the sustainable development of oases in Xinjiang arid area.At present,the hydrologic model used to study the snowmelt runoff on a scale of watershed includes conceptual models and distributed models.In mountain areas,snow cover monitoring based on satellite images is difficult to get the data of high temporal and spatial resolution.Besides,due to the complex terrain,the accuracy of traditional extrapolating methods used to extrapolate the snow depth reduces.With Tianshan Station for Snowcover and Avalanche Research as the research base,a digital camera was employed to monitor the area of snow cover on the sunny slope in the spring of 2011.After correcting the images by DEM of 30m resolution,the snow cover’s temporal and spatial variation process during snow-melting period within the researched area was obtained.However,images captured by calibrated Landsat5 and Snomap algorithm couldn’t identify the snow cover in the target region.At the same time,the snow depth was measured with the snowstake prepared on the sunny slope and obtained by the Classification And Regression Tree(CART) algorithm.Based on these results,the accuracy of snow-melting module of SRM model driven by different temperature variables was evaluated.The results show that it is possible to monitor snow cover by digital camera,which has significantly improved the temporal and spatial resolution of images compared to satellite.However,some technical problems should be solved.In addition,the snow depth extrapolation with CART also works well.The SRM model,driven by daily maximum air temperature and daily positive accumulative temperature can recognize the snow-melting process with daily mean temperature below 0℃.