论文部分内容阅读
为快速获得低等级河流的高精度河网,以松花江一级支流——阿什河为例,运用SWAT模型和河网密度相结合的方法合理计算集水面积阈值,通过最佳阈值的设定准确提取数字河网。研究结果表明:过大或过小的集水面积阈值都将导致所提取的河网与真实河网不符,当集水面积阈值>5 km2时,河网密度及其一阶导数趋于稳定、二阶导数趋于零,说明河网变化趋于稳定,提取河网接近真实河网;当集水面积阈值分别为5、10和20 km2时,三者的模拟结果在平原区差距较小、在山区差距较大,其中当阈值为10 km2时,提取的河网与真实河网吻合度最高。采用相对误差Re、相关系数R2和纳什效率系数Ens对模拟结果进行验证和评价,结果显示,当集水面积阈值为10 km2时,Re=6%、R2=0.996、Ens=0.987,模拟结果最好。
In order to obtain the high precision river network of low grade river quickly, taking the Ash River, a first tributary of the Songhua River as an example, a reasonable combination of SWAT model and river network density is used to calculate the catchment area threshold, Accurate extraction of digital river network. The results show that: the threshold of catchment area is too large or too small will lead to the river network is not consistent with the real river network, when the catchment area threshold> 5 km2, river network density and its first derivative tends to be stable, The second derivative tends to be zero, indicating that the river network changes tend to be stable and the river network is extracted close to the real river network. When the catchment area thresholds are 5, 10 and 20 km2, respectively, In the mountainous area, there is a big gap between them. When the threshold is 10 km2, the extracted river network has the highest degree of coincidence with the real river network. The simulation results were validated and evaluated using relative error Re, correlation coefficient R2 and Nash efficiency coefficient Ens. The results showed that when the catchment area threshold was 10 km2, Re = 6%, R2 = 0.996, Ens = 0.987, it is good.