Remote sensing identification of urban black-odor water bodies in Shenyang city based on GF-2 image

Yue Yao, Qian Shen, Li Zhu, Hongjie Gao, Hongye Cao, Hui Han, Jianguo Sun, Junsheng LiNational 2019 · DOI

Abstract

Urban "black-odor" water bodies represent a significant environmental challenge in China. This study develops a remote sensing identification model based on the Black and Odorous water Index (BOI) using high-resolution Gaofen-2 (GF-2) satellite imagery. By analyzing the spectral characteristics of 96 water samples in Shenyang, researchers found that black-odor water exhibits a flatter reflectance slope in the green-to-red spectral range compared to normal water. The BOI model achieved high identification accuracy (up to 100% on validation sets) and proved effective when using Rayleigh-corrected reflectance ($R_{rc}$), providing a practical tool for city-wide water quality monitoring.

Introduction

Rapid urbanization in China has led to the emergence of "black-odor" water bodies—urban rivers and ponds characterized by dark colors and foul smells due to organic pollution and low dissolved oxygen. Under the national "Water Ten Plan," eliminating these pollutants is a priority. However, traditional monitoring via manual sampling is labor-intensive and lacks spatial continuity. Satellite remote sensing, particularly using high-resolution sensors like GF-2, offers a promising alternative for monitoring narrow urban channels that are often invisible to lower-resolution satellites like Landsat.

Methodology and Study Area

The study focused on the urban area of Shenyang, China. Field surveys conducted between 2015 and 2016 collected spectral data and water quality parameters (transparency, dissolved oxygen, and ammonia nitrogen) from 46 "normal" water sites and 50 "black-odor" sites.

Distribution of water sampling points in Shenyang City, including both normal and black-odor water bodies across various urban river systems.

The Black and Odorous water Index (BOI)

Spectral analysis revealed that while normal water typically shows a distinct reflectance peak in the green band, black-odor water displays a "flat" signature between the green and red wavelengths. To quantify this differentiator, the researchers proposed the BOI:

$$BOI = \frac{R_{rs}(Green) - R_{rs}(Red)}{R_{rs}(Blue) + R_{rs}(Green) + R_{rs}(Red)}$$

Where $R_{rs}$ is the remote sensing reflectance. For practical satellite applications where full atmospheric correction is challenging, the study demonstrated that Rayleigh-corrected reflectance ($R_{rc}$) can be used as a robust proxy.

Results and Discussion

Model Performance

The BOI model demonstrated exceptional accuracy in distinguishing polluted waters from healthy ones:

  • Using $R_{rs}$: A threshold of $BOI < 0.065$ identified black-odor water with 100% accuracy on the validation set.
  • Using $R_{rc}$: A threshold of $BOI < 0.05$ achieved 96.8% accuracy, making it highly suitable for operational satellite monitoring without intensive atmospheric correction.

Temporal Monitoring in Shenyang

Application of the model to GF-2 imagery from 2015 and 2016 tracked the effectiveness of local remediation efforts. The results showed gradual improvement in the Mantang and Xinkai rivers, while identifying that the Huishanming channel remained severely polluted, highlighting the index's utility for policy enforcement and environmental management.

Sensitivity to Atmospheric Conditions

Simulation results indicated that the model's effectiveness is sensitive to atmospheric clarity. The BOI is most reliable under clear conditions where the Aerosol Optical Thickness (AOT at 550nm) is $\le$ 0.5. In hazier conditions, atmospheric "noise" begins to mask the subtle spectral differences of the water.

Conclusion

The BOI provides a robust, high-resolution method for identifying urban black-odor water bodies. While calibrated for the water characteristics of Shenyang, the index offers a scalable framework for urban water quality management. Future work will focus on validating the index across different geographic regions and refining atmospheric correction models to enhance its utility in diverse urban environments.

Summary updated

Abstract

ä¸­å›½åŸŽå¸‚é»‘è‡­æ°´ä½“æƒ å†µä¸¥é‡ï¼ŒåŸºäºŽé¥æ„Ÿç›‘æµ‹é»‘è‡­æ°´ä½“åˆšåˆšèµ·æ­¥ï¼Œå¾ˆå¤šé—®é¢˜å¾ è§£å†³ã€‚ä»¥æ²ˆé˜³å¸‚åŸŽå¸‚å»ºæˆåŒºå† ä¸»è¦æ²³æµä¸ºç ”ç©¶åŒºï¼ŒäºŽ2015年—2016年开展地面调查,获取了浑河和蒲河46ä¸ªä¸€èˆ¬æ°´ä½“çš„æ ·ç‚¹ï¼Œå’Œè¾‰å±±æ˜Žæ¸ ã€æ»¡å ‚æ²³ã€ç»†æ²³ä»¥åŠå¾®å±±æ¹–è·¯é™„è¿‘ã€ä¸é¦™æ¹–åŒ—éƒ¨50ä¸ªé»‘è‡­æ°´ä½“çš„æ ·ç‚¹æ•°æ®ï¼ŒåŒ æ‹¬æ°´é¢å ‰è°±å’Œä¸»è¦æ°´è´¨å‚æ•°ã€‚åˆ†æžäº†é»‘è‡­æ°´ä½“ä¸Žä¸€èˆ¬æ°´ä½“çš„å ‰è°±ç‰¹å¾ï¼Œå‘çŽ°åŸŽå¸‚é»‘è‡­æ°´ä½“åå°„çŽ‡å ‰è°±åœ¨ç»¿å ‰â€”çº¢å ‰æ³¢æ®µå˜åŒ–æ¯”ä¸€èˆ¬æ°´ä½“å¹³ç¼“ï¼ŒåŸºäºŽè¿™ä¸€ç‰¹ç‚¹æå‡ºäº†ä¸€ç§åŸºäºŽåå°„çŽ‡å ‰è°±æŒ‡æ•°BOI(Black and Odorous water Index)çš„é»‘è‡­æ°´ä½“è¯†åˆ«æ¨¡åž‹ï¼Œå¹¶å°†å ¶ä¸Žçº¢ç»¿æ³¢æ®µæ¯”å€¼æŒ‡æ•°è¿›è¡Œå¯¹æ¯”ï¼Œå ·æœ‰æ›´å¥½çš„è¯†åˆ«ç²¾åº¦ã€‚ç»“æžœè¡¨æ˜Žï¼š(1)基于遥感反射率(Rrs)计算的BOI小于0.065时,可判为黑臭水体。(2)由于GF2æ°´ä½“å›¾åƒç²¾ç¡®å¤§æ°”æ ¡æ­£å­˜åœ¨å›°éš¾ï¼Œå¯ä»¥åˆ©ç”¨ç‘žåˆ©æ•£å°„æ ¡æ­£åå°„çŽ‡(Rrc)替代Rrs,BOI小于阈值0.05æ—¶ï¼Œå¯åˆ¤åˆ«ä¸ºé»‘è‡­æ°´ä½“ï¼›åŒæ—¶æ¨¡æ‹Ÿè¯æ˜Žï¼Œå½“æ°”æº¶èƒ¶å ‰å­¦åŽšåº¦é€æ¸å¢žå¤§æ—¶ï¼Œé»‘è‡­æ°´ä½“ä¸Žä¸€èˆ¬æ°´ä½“çš„å ‰è°±å·®å¼‚å°†é€æ¸å‡å°ï¼Œå› æ­¤è¿™ç§æ–¹æ³•ä¸»è¦é€‚ç”¨äºŽæ¯”è¾ƒæ¸ æ™°çš„å›¾åƒã€æ°”æº¶èƒ¶å ‰å­¦åŽšåº¦æ¯”è¾ƒå°(如AOT(550)≤0.5时)。(3)基于Rrc的BOI模型可以较好的应用于GF2å›¾åƒä¸Šï¼Œå ·æœ‰è¾ƒå¥½çš„è¯†åˆ«ç²¾åº¦ã€‚å¯¹2015年—2016å¹´3景GF-2å½±åƒæå–çš„ç»“æžœæ˜¾ç¤ºï¼Œæ»¡å ‚æ²³å’Œæ–°å¼€æ²³é»‘è‡­çŽ°è±¡å¾—åˆ°é€æ­¥æ”¹å–„ï¼Œè¾‰å±±æ˜Žæ¸ é»‘è‡­çŽ°è±¡ä¾ç„¶ä¸¥å³»ã€‚æœ¬æ–‡å‘å±•çš„é»‘è‡­æ°´ä½“é¥æ„Ÿè¯†åˆ«ç®—æ³•ä¸»è¦æ˜¯åŸºäºŽæ²ˆé˜³é»‘è‡­æ°´ä½“çš„å ‰è°±ç‰¹å¾ï¼Œä» åœ¨æ²ˆé˜³å¸‚è¿›è¡Œäº†éªŒè¯ï¼Œå°†æ¥è¿˜éœ€åœ¨å ¶ä»–åŸŽå¸‚è¿›ä¸€æ­¥éªŒè¯ï¼Œå¹¶ä¸”éœ€è¦æ›´å¤šåœ°è€ƒè™‘å¤šç§å› ç´ å¯¹æ°´ä½“åå°„çŽ‡çš„å½±å“ã€‚
BibTeX
@inproceedings{yao2019remote,
  author = {Yue YAO and Qian Shen and Li Zhu and Hongjie GAO and Hongye Cao and Hui HAN and Jianguo Sun and Junsheng Li},
  title = {{Remote sensing identification of urban black-odor water bodies in Shenyang city based on GF-2 image}},
  booktitle = {National},
  year = {2019},
  doi = {10.11834/jrs.20197482},
}
Figure