2018北京市建成区0.8米GF2黑臭水体分布产品

以覆盖了2018年北京市建成区范围的0.8米GF-2 PMS原始影像集为基础，完成了图像的预处理，包括影像的融合、几何精校正、大气校正、水体提取等步骤。其中融合是首先将4米的MSS4个波段间的几何偏差先做了复原，然后再与PAN融合，形成0.8米4个波段的DN值影像。不仅能够精确地校正了波段间的偏差，还能保持很高的波谱保真度。几何精校正是以全球公开的Sentinel-2（空间分辨率10米）正射影像产品作为参考影像，通过影像匹配的方法自动获取控制点用于几何精校正，并利用了区域网平差保证了整个研究区内不同时相影像数据的几何一致性。绝对几何精度在10米以内，接边精度在2个像元以内。大气校正采用了相对辐射归一化方法，以与GF-2时相接近的Sentinel-2的L2A影像作为参考影像，输出GF-2反射率产品。经验证，反射率误差在4个波段的平均相对误差在30%以内。在反射率产品基础上，提取河流水体中心线，并带入黑臭水体模型，判别出黑臭水体河段并输出。经验证，黑臭水体反演的精度超过70%。

Based on the 0.8-meter GF-2 PMS raw image dataset covering the built-up area of Beijing in 2018, image preprocessing was completed, including steps such as image fusion, geometric precise correction, atmospheric correction, and water body extraction. Specifically, for image fusion: first, geometric deviations among the four 4-meter MSS bands were corrected, then fused with the PAN band to generate a 0.8-meter DN value image with four spectral bands. This method not only accurately corrects inter-band deviations but also maintains high spectral fidelity. For geometric precise correction, the globally released Sentinel-2 (10-meter spatial resolution) orthorectified image products were used as reference imagery. Control points were automatically obtained via image matching for geometric correction, and block adjustment was applied to ensure geometric consistency of multi-temporal image data across the entire study area. The absolute geometric accuracy is within 10 meters, and the edge-matching accuracy is within 2 pixels. Atmospheric correction was performed using the relative radiometric normalization method, with Sentinel-2 L2A images acquired at a time close to that of the GF-2 images as references, to generate GF-2 reflectance products. Validation results show that the average relative error of reflectance across the four bands is within 30%. Based on the reflectance products, the centerlines of river water bodies were extracted, then input into the black and odorous water body model to identify and output sections of black and odorous water bodies. Validation results indicate that the inversion accuracy of black and odorous water bodies exceeds 70%.

搜集汇总

背景与挑战

背景概述

该数据集是基于2018年北京市建成区0.8米GF-2 PMS影像生成的黑臭水体分布产品，经过多步骤预处理和验证，反演精度超过70%，适用于水体污染研究。