Next Article in Journal
Assessing Riverbank Change Caused by Sand Mining and Waste Disposal Using Web-Based Volunteered Geographic Information
Next Article in Special Issue
Bathymetric Modelling of High Mountain Tropical Lakes of Southern Ecuador
Previous Article in Journal
Micro-Catchments, Macro Effects: Natural Water Retention Measures in the Kylldal Catchment, Germany
Previous Article in Special Issue
Assessment and Mapping of Riverine Flood Susceptibility (RFS) in India through Coupled Multicriteria Decision Making Models and Geospatial Techniques
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 16
Issue 5
10.3390/w16050732
water-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Water Quality Estimation Using Gaofen-2 Images Based on UAV Multispectral Data Modeling in Qinba Rugged Terrain Area

by
Dianchao Han
1,
Yongxiang Cao
1,2,*,
Fan Yang
1,2,
Xin Zhang
3 and
Min Yang
3,*
1
Power China Group, Northwest Engineering Corporation Limited, Xi’an 710065, China
2
Shaanxi Union Research Center of University and Enterprise for River and Lake Ecosystems Protection and Restoration, Xi’an 710065, China
3
School of Resources Engineering, Xi’an University of Architecture and Technology, Xi’an 710055, China
*
Authors to whom correspondence should be addressed.
Water 2024, 16(5), 732; https://doi.org/10.3390/w16050732
Submission received: 26 December 2023 / Revised: 22 February 2024 / Accepted: 26 February 2024 / Published: 29 February 2024
(This article belongs to the Special Issue Application of GIS Models and Remote Sensing in Water Quality Evaluation, Land and Coastal Zone Management)
Browse Figures
Versions Notes

Abstract

This study presents an innovative method for large-scale surface water quality assessment in rugged terrain areas, specifically tailored for regions like the Qinba Mountains. The approach combines the use of high-resolution (10 cm) multispectral data acquired by unmanned aerial vehicles (UAVs) with synchronized ground sampling and 1 m resolution multispectral imagery from China’s Gaofen-2 satellite. By integrating these technologies, the study aims to capitalize on the convenience and synchronized observation capabilities of UAV remote sensing, while leveraging the broad coverage of satellite remote sensing to overcome the limitations of each individual technique. Initially, a multispectral estimation model is developed for key water quality parameters, including chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP), utilizing data from UAVs and coordinated ground samples. Subsequently, a comparison is made between the spectral band ratios (R/G and NIR/G) obtained from the UAV data and those from the Gaofen-2 satellite data, revealing a substantial similarity. Ultimately, this integrated methodology is successfully employed in monitoring water quality across a vast area, particularly along the midstream of the Hanjiang River in the Qinba Mountain region. The results underscore the feasibility, advantages, improved efficiency, and enhanced accuracy of this approach, making it particularly suitable for large-scale water quality monitoring in mountainous terrain. Furthermore, this method reduces the burden associated with traditional ground-based spectral acquisitions, paving the way for a more practical and cost-effective solution in monitoring vast water bodies.

1. Introduction

Although China has a vast land and dense rivers and lakes, it is a country with severe water scarcity [1]. The seven major river watersheds in China, represented by the Yangtze River, have varying degrees of alarming pollution [2]. According to a report released by the Ministry of Ecology and Environment of China in January 2023 on the status of surface water environmental quality in China from January to December 2022, the proportion of water quality reaching grades I to III reached 87.9% of 3641 samples, an increase of 3.0% compared to 2021, and the proportion of water quality reaching grade V was 0.7%, both of which meet the requirements. Among the 210 monitored key lakes (reservoirs), the proportion of water quality reaching grades I to III was 73.8%, an increase of 0.9 percentage points compared to the previous year; the proportion of water quality reaching grade V was 4.8%, a decrease of 0.4 percentage points. The main pollutant indices were chemical oxygen demand, total phosphorus, and permanganate index [3]. Overall, the water quality in China is good, and it has improved compared to the previous year. However, water areas near populated areas have shown a trend of eutrophication [4]. The eutrophication of water bodies is caused by an increase in the concentration of phosphorus, nitrogen, and other elements, leading to a proliferation of algae and other plankton.
Water quality testing aims to determine water quality parameters such as chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP) to estimate trends in the quality of water bodies [5]. Currently, three main methods are used, including water quality testing, remote sensing detection, and mathematical model prediction [6].
Water sample collection for water quality testing provides single-point water quality parameters and the limited number of sampling points restricts the representative results of the entire water body [7]. At the same time, this traditional method has high requirements for the technical level of researchers, the precision of analytical instruments, and the quality of sampling, and chemical testing based on field sampling is expensive, operationally complex, and slow, which can easily lead to delays in decision-making and cannot effectively and accurately help investigate overall water quality [8]. Mathematical models can predict future trends from current and past single-point water quality parameters [9]. Remote sensing detection has the advantages of low cost, high speed, and large-area observation, and is widely used in dynamic water quality testing [10]. However, remote sensing observation time is limited by satellite orbits, and ground synchronous observation may be affected by weather and work arrangements during satellite overpass periods [11].
In recent years, unmanned aerial vehicles (UAVs) have gradually become a hot topic in various fields: with the expansion of their applications, they have been applied to the inversion of crop yields and chlorophyll content, but most of them are used to invert water quality parameters in small water bodies [12]. The specific method is to obtain spectral reflectance data from small water areas through the multispectral camera carried by the drone, and to use empirical and semi-empirical methods to construct concentration inversion models for water quality parameters such as COD, TP, and turbidity, and study their spatial distribution [13]. Among these methods, statistical and analytical methods are used to select the optimal band for the correlation between the remote sensing data and the water quality parameters, and to realize the inversion of the water quality parameters by experience [14]. The advantages of the semi-empirical method include simple structure and convenient use. It does not require a large number of measured water samples, and has a certain practical physical significance. The sensitive band is determined by the actual spectral characteristics corresponding to different water quality parameters, and then inverted through the sensitive band or their combination [15]. In recent years, some water quality detection studies have introduced theoretical methods such as artificial neural networks (ANNs) [16] and support vector machines (SVMs) [17] into the model construction process, with good results and recognition from many scholars. The detection method of water quality by drone remote sensing can find some features of pollution migration and pollution sources that are difficult to show by conventional methods [18]. In addition, it has other advantages such as low cost, flexible working hours, convenient simultaneous ground observation, and high speed [19]. However, UAVs are more suitable for water quality observation in small water bodies, and their efficiency is low for rivers or river basins with larger scales [20].
This study aims to merge the benefits of drone remote sensing for synchronized ground observations with the extensive coverage of satellite remote sensing, addressing the drones’ limited scale and the satellites’ challenges in synchronized ground observations. A regional surface water quality detection method based on multispectral observation of rotorcraft UAVs, synchronous ground sampling, and satellite remote sensing multispectral data was developed. The method first uses multispectral observation of rotorcraft UAVs and synchronous ground water sample collection and testing to establish multispectral models for water quality parameters such as COD, TN, and TP. Then, the multispectral data from the UAVs were compared with the 1 m resolution multispectral data from China’s Gaofen-2 for similarity comparison. Finally, the model was applied to the multispectral data from Gaofen-2 for large-scale water quality monitoring. This method was successfully applied to the midstream of the Han River in the Qinling Mountains, where synchronous multispectral data collection using UAVs avoided the high workload of ground spectral sampling in conventional water quality remote sensing monitoring. It is particularly suitable for modeling and detection work relating to water quality remote sensing in mountainous regions.

2. Materials and Methods

2.1. Study Area

The study area is located in the Fenghuang Ridge of the Qinba Mountain region, which is part of the Hanjiang River watershed, the largest tributary of the Yangtze River. The region has a northern subtropical humid monsoon climate, with an average annual precipitation of 1085.13 mm. Precipitation is concentrated between June and September, accounting for 52% to 76% of the annual total, while the dry season (December, January, and February) accounts for only 1.1% to 4.5% of the annual total. The average temperature is 15.0 °C, with January as the coldest month at an average temperature of 3.4 °C and an extreme minimum temperature of −7.4 °C. July is the hottest month with an average temperature of 25.5 °C and an extreme maximum temperature of 41.7 °C. The Chenjiagou River has a length of 3.38 km, a watershed area of 2.43 km2, and a height difference of 644.6 m. The dry season flow is 240.8 m3/d. The Xiaomixi stone coal mine is located in the Banjiuguan formation (Figure 1) [21,22]. The Banjiuguan formation mainly consists of black carbonaceous slate and carbonaceous siliceous rocks with coarse-grained rocks interbedded in the middle to upper parts, showing high hardness and fine-grained crystalline structure with plate-like texture. Its composition is mainly composed of quartz, sericite, muscovite, carbonaceous matter, etc., and it is a set of deep marine deep water basin-type coal-rich (siliceous) carbonaceous sedimentary rocks with multi-cyclic volcanic rocks [23,24].

2.2. Multispectral Data Acquisition and Preprocessing for UAVs

The acquisition of multispectral images of surface water in the study area was completed by the DJI Phantom 4 multispectral system [25]. The Phantom 4 multispectral system is equipped with an integrated multispectral imaging system that integrates one visible light camera and five multispectral cameras (blue, green, red, red edge, and near-infrared) responsible for visible light imaging and multispectral imaging. It also uses the TimeSync time synchronization system to achieve millisecond-level error in the camera imaging time by synchronizing the clock system of the flight controller, camera, and RTK to the microsecond level. The position of the center point of each camera lens and the center point of the antenna are combined with equipment attitude information for real-time compensation to obtain more accurate position information for the image. The technical parameters are shown in Table 1.
This study used PhotoScan (Agisoft Metashape Professional Version 1.8.0) to perform 3D reconstruction. High-quality 3D models were generated through processing and analyzing multispectral data, and then true-orthogonal multispectral images (TDOMs) were generated through orthorectification [26]. Due to the presence of distortions and color aberrations in the camera itself, it was necessary to calibrate the camera using a sun sensor model. During the calibration process, accurate camera parameters needed to be input to eliminate the influence of distortions and color aberrations on data processing. By identifying common feature points in aerial photos for photo alignment, the camera’s various parameters were adjusted using optimization to improve data quality. By collecting dense clouds, point cloud data were transformed into grid data. The grid data were then used to perform geometric correction and resampling on the original tile images, followed by orthorectification to stitch and fuse the tile data to generate a complete orthorectified image. The geometric resolution of the UAV multispectral orthorectified image was 10 cm, with map projection using CGCS2000 coordinate system and Gauss–Kruger projection. The multispectral image used five channels: blue, green, red, red edge, and near-infrared.

2.3. Water Sample Collection

This study selected 14 observation points from upstream to downstream in different sections of the Hanjiang River, Haoping River, Xiaomi Stream, and Dami Stream, and collected 14 water samples with 500 mL in polyvinyl chloride bottles (Figure 2). The water quality sample collection was carried out simultaneously with UAV photography data collection. Before data collection, the sampling plan was designed, considering the influence of river flow velocity. During the actual sample collection, water samples were collected 0.5 m below the water surface, with 500 mL collected from each point and uniformly distributed within the study area. A water sampling method was used to collect water samples within the drone observation area. Water sample collection at river centers was conducted utilizing bridges, suspension bridges, or a 3–5-m-long pole. In this process, a water intake cylinder was attached to either a rope or the end of the pole, with a float marker positioned approximately 0.5 m above it to indicate the sampling depth. For the bridge and suspension bridge method, the rope was securely fastened to the structure and then gradually lowered into the river, allowing adjustments to the sampling depth based on the float marker’s position. For the long pole method, the pole was extended into the river from a stable position along the banks, with similar depth adjustments made using the float marker. In both approaches, the cylinder was maintained in a stable position until it was filled with river water, followed by a gradual retrieval process to prevent spillage or contamination of the sample. The collected water samples were then prepared for subsequent water quality analysis.
This study used the methods of total phosphorus (TP), total nitrogen (TN), and chemical oxygen demand (COD) in water samples according to the Chinese national standard “Water Quality Determination of Total Phosphorus (GB11893-89)” [27], the Chinese environmental industry standard “Water Quality Determination of the Chemical Oxygen Demand (HJ828-2017)”, and “Water Quality Determination of Total Nitrogen (HJ636-2012)” [28,29]. The determination of chemical oxygen demand was carried out by titrating the unreduced potassium dichromate in the water sample with ammonium ferrous sulfate, and calculating the mass concentration of oxygen consumed from the amount of potassium dichromate consumed. Total nitrogen was determined by alkaline potassium persulfate solution at a temperature of 120~124 °C, which transformed nitrogen compounds in the sample into nitrate, and then measured the absorbance at 220 nm and 275 nm using ultraviolet spectrophotometry. Total phosphorus was determined by digesting the sample with potassium persulfate (or nitric acid-perchloric acid) under neutral conditions, and then oxidizing all the phosphorus to orthophosphate. In an acidic medium, orthophosphate reacts with ammonium molybdate in the presence of antimony salt to form a heteropoly acid of phosphomolybdate, which is immediately reduced by ascorbic acid to form a blue complex. The experiment was carried out according to the relevant standards to obtain the content of total nitrogen, total phosphorus, and chemical oxygen demand.

2.4. Data Acquisition and Processing of Gaofen-2 Data

The Gaofen-2 satellite, China’s first civil optical satellite with a spatial resolution within 1 m, was launched into orbit at the Taiyuan Satellite Launch Center aboard a Long March 4B carrier rocket on 19 August 2014 and officially entered into service on 21 August 2014. Gaofen-2 carries two cameras: a 0.8 m panchromatic (PAN) camera and a 3.2 m multispectral (MSI) camera (Table 2). Its imaging width is 45 km, and it can repeat observations of any region on the Earth’s surface within 5 days [30].
This study used two scenes of Gaofen-2 remote sensing data downloaded from the website of GeoCloud, operated by the China Geological Survey, collected on 9 January 2023, with multispectral data undergoing radiation calibration, atmospheric correction, and orthorectification [31]. Then, the Gram–Schmidt pan sharpening method was used to fuse the orthorectified panchromatic band and multispectral bands to generate 1 m resolution multispectral fused images [32]. Preprocessing was carried out using ENVI5.3 (Exelis Inc., Alexandria, VA, USA) software [33]. After preprocessing, the fused high-resolution multispectral image was extracted using the Normalized Difference Water Index (NDWI) method to extract the water distribution in the study area [34]. The NDWI results were then used to mask the fused high-resolution multispectral image, retaining only the pixels corresponding to water bodies.

2.5. Establishment of UAV Multispectral Water Quality Estimation Model

The blue band (450 nm), green band (560 nm), red band (650 nm), and near-infrared band (840 nm) selected from the calibrated UAV multispectral images according to the Gaofen-2 multispectral bands were denoted as B, G, R, and NIR, respectively. Based on previous research, the band-specific correlation results with COD, TP, and TN water quality parameters were obtained. The R/G ratio parameter was selected as a sensitive band parameter for COD and TP, while the NIR/G ratio parameter was selected as a sensitive band parameter for TN [35,36]. In this study, we developed concentration retrieval models for COD, TP, and TN water quality parameters using linear, power, and quadratic polynomial regression methods. The inversion model was established using Microsoft Excel 2013 (Microsoft Corporation, Redmond, WA, USA).

2.6. Comparison of Multispectral Images from UAVs and Gaofen-2

To demonstrate the reliability of the water quality estimation model established in this study, we also calculated the R/G and NIR/G ratio images using both Gaofen-2 data and drone multispectral data [36]. Then, we conducted a relative bias analysis between these two ratio images. Relative bias is a measure of the deviation of a single measurement result from the average value in two repeated measurements. The relative bias test for two repeated measurements is calculated using Formula (1).
R D = A − B A + B × 100 %
RD refers to relative deviation, A represents the R/G and NIR/G ratio images calculated using the Gaofen-2 data in this study, and B refers to the R/G and NIR/G ratio images calculated using the UAV multispectral data.

3. Results

3.1. Relative Deviation between UAV and Gaofen-2 Multispectral Images

This study calculated the relative deviation of the R/G and NIR/G band ratio images from unmanned aerial vehicles (UAVs) and GF-2 remote sensing images. The range of R/G values was 0–22.38%, with high values concentrated at the confluence of the Hanjiang tributaries. Low values were concentrated in the mainstream of the Hanjiang (Figure 3a). The proportion of pixels with a relative deviation within 10% was 92.47% (Figure 3c), indicating that the R/G band ratios from UAV multispectral data and GF-2 remote sensing data are very similar. The range of NIR/G values was 0–56.56%, with high values distributed in the mainstream of the Hanjiang, and low values distributed along the banks of the river and near the Han River (Figure 3b). The proportion of pixels with a relative deviation within 30% was 88.94% (Figure 3d), indicating that the NIR/G band ratios from the UAV multispectral data and GF-2 remote sensing data are also relatively close. Therefore, the results of this study’s relative deviation analysis show good consistency and stability in band ratio values between the satellite and UAV images. This indicates that the water quality model established using UAV band ratios can be applied to satellite remote sensing images.

3.2. Establishment of Water Quality Estimation Models

This study used the model functions of previous researchers who used ground spectral data for water quality modeling and calculated R/G and NIR/G ratio images on the UAV multispectral data of each sampling point, and then took the average value within a 5 m × 5 m area around the sampling point as the independent variable [35,36]. The corresponding chemical oxygen demand (COD), total phosphorus (TP), and total nitrogen (TN) were used as the dependent variables. The regression models were established using Microsoft Excel 2013 software and a linear regression model was selected as the estimation model for COD, an exponential function regression model was chosen as the estimation model for TP, and a polynomial regression model was selected as the estimation model for TN. The r2 values of the three modes were all over 0.5 (Table 3).
Satellite remote sensing data play an important role in water quality monitoring. The water quality status of surface water can be evaluated and monitored via satellite remote sensing images, providing important information for water environment management and protection. By analyzing the spectral characteristics and reflectance of satellite remote sensing images, pollutants in water bodies can be quickly identified and quantitatively evaluated, providing data support for monitoring and managing water pollution. Based on the water quality status in various regions, the water quality condition is evaluated by referring to the Chinese national standard “Environmental Quality Standards for Surface Water (GB3838-2002)” (Table 4), and a water quality distribution map is generated. The schematic diagram obtained is shown below [37].
The estimation results of chemical oxygen demand (COD) showed that the water bodies in the mainstream of the Hanjiang River and its tributaries such as the Ruhe River, Donghe River, and Dadaohe River have a large area with COD concentration of 20–30 mg/L, which is classified as Class IV water according to the Chinese national standard “Environmental Quality Standards for Surface Water (GB3838-2002)” (Figure 4) [37]. The water bodies in the upstream of the Haoping River and Ruhe River, Donghe River, and Dadaohe River have a COD concentration of 30–40 mg/L, which is classified as Class V water according to the Chinese national standard “Environmental Quality Standards for Surface Water (GB3838-2002)” (Figure 4) [37]. There are smaller areas of water bodies in the vicinity of Donghe Town and Liushui Town, as well as in the upstream of the Ruhe River and Donghe River, with chemical oxygen demand (COD) values ranging from 0 to 15 mg/L, which is classified as Class II water (Figure 4). There are scattered waters in the upstream of the Xiaomi Stream with COD concentration over 40 mg/L, which is classified as Class V water (Figure 4). The COD concentration in the water bodies in the study area is basically in line with the local standard Scheme for Dividing the Functional Zones of Surface Water Bodies in the Hanjiang River System (Shaanxi section) (DB61-262-1997) established by Shaanxi Province [38].
According to the total phosphorus (TP) estimation results, the total phosphorus contents of the mainstream of the Hanjiang River and the tributaries of the Ruhe River, Donghe River, and Dadaohe River were 0.1–0.2 mg/L, which met the three types of water codes for the Chinese national standard “Environmental Quality Standard for Surface Water” (GB3838-2002) (Figure 5) [37]. The total phosphorus contents in the upper reaches of the Haoping River, Ruhe River, Donghe River, and Dadaohe River, as well as the Xiaomi Stream near Yunfeng Village, were 0.2–0.3 mg/L, which met the fourth-class water quality standard (GB3838-2002) (Figure 5) [37]. The total phosphorus concentration in the study area was basically in line with the “Hanjiang River Surface Water Functional Zoning Scheme” (DB61-262-1997) formulated by Shaanxi Province [38].
According to the estimation of total nitrogen (TN), the total nitrogen contents of the main water bodies of the Hanjiang River and its tributaries, such as Yunfeng Village and the Ruhe River, Donghe River, and Dadaohe River, were 0.5~1.0 mg/L, which met China’s “Environmental Quality Standard for Surface Water” (GB3838-2002) (Figure 6) [37]. In addition, the total nitrogen contents in the upper reaches of the Ruhe River, Donghe River, and Dadaohe River were 1.0~1.5 mg/L, which conforms to the Chinese national standard (GB3838-2002) [37] (Figure 6), which is classified as Class IV water. The total nitrogen concentration in the study area was basically in line with the “Hanging River Surface Water Functional Zoning Scheme” (DB61-262-1997) formulated by Shaanxi Province [38].

4. Discussion

In this study, we have employed a hybrid approach, combining drone multispectral technology with high-resolution satellite remote sensing for water quality monitoring in the Qinling Mountainous region of the Hanjiang Valley. This region, known for its rugged terrain and variable ecological conditions, presents unique challenges for traditional monitoring methods [39]. Our approach aims to overcome these challenges by leveraging the complementary strengths of drones and satellites. To evaluate the effectiveness of our approach, we conducted a comparative analysis of water quality data obtained from traditional ground-based methods and our integrated drone–satellite system. The results were satisfactory. While ground-based methods were limited to small, accessible areas and were subject to safety concerns in rugged terrain [40], our drone–satellite system was able to capture multispectral data over a much larger area, with significantly improved efficiency and safety [41]. Specifically, our drone system, equipped with high-resolution cameras and spectral sensors, was able to capture detailed images of water bodies with a resolution of up to 0.5 m. This allowed us to identify and map water quality parameters such as COD, TP, and TN with unprecedented accuracy [42]. When compared to ground-based methods, the drone system achieved a significant increase in data collection efficiency [43]. Moreover, by integrating these drone-collected data with high-resolution satellite remote sensing data, we were able to develop a predictive model for water quality parameters [40]. This model, trained on a dataset comprising both drone and satellite data, was able to estimate water quality parameters over a wide area [44]. In addition to improved efficiency, our drone–satellite hybrid method also offers several practical advantages. Firstly, it enables rapid deployment and real-time monitoring, allowing for timely responses to water quality changes [45]. Secondly, it reduces the need for extensive ground-based sampling, which can be costly and time-consuming in mountainous regions. Finally, the system’s scalability and adaptability make it suitable for a wide range of water quality monitoring applications in diverse geographical settings.
Currently, numerous studies utilize unmanned aerial vehicle (UAV) multispectral data and satellite remote sensing data for earth surface observation. Some studies used UAV multispectral data to conduct scale-up corrections and construct regional-scale soil salt content inversion models based on satellite bands [46]. Other studies combined UAVs and satellites to obtain regional-scale structural parameters of Populus euphratica forests and monitor above-ground biomass [47]. Additionally, some researchers used UAVs and GF-1 to obtain the above-ground biomass in Shengjin Lake and to fuse UAV and satellite images to obtain more comprehensive and accurate information, improving biomass inversion accuracy [48]. These studies demonstrate that combining UAV multispectral data and satellite multispectral data can leverage the respective advantages of both technologies to facilitate surface observation work. In surface water quality observation, researchers have used Landsat 30 m resolution remote sensing data, Sentinel-2 10 m resolution remote sensing data, and UAV 8 cm resolution data to map and compare chlorophyll a content in small reservoirs. The results indicate good consistency among the three sensors at different spatial resolutions (10 m, 30 m, and 8 cm) [49]. Building on previous research, this study analyzed the relative deviation of the R/G and NIR/G band ratios from UAV and satellite data. The majority of the R/G band ratio relative deviations were less than 10%, while the NIR/G band ratio relative deviations were mostly under 40%. Therefore, despite significant differences in pixel reflectance values between the UAV and satellite data bands, the band ratio results were relatively stable. This is consistent with previous research findings and indicates that water quality models based on UAV multispectral data and synchronized water samples can be reliably applied to high-resolution satellite remote sensing data.
Based on the comparison between our study’s findings and the water quality monitoring results reported by the Shaanxi Provincial Aquatic Research and Technology Promotion Station for the Ankang section of the Hanjiang River mainstream from 2020 to 2022, several noteworthy observations emerge. Firstly, regarding total phosphorus concentration, our study’s range of 0.001 mg/L to 0.335 mg/L overlaps significantly with the station’s reported range of 0.018 mg/L to 0.370 mg/L [50]. This suggests a general consistency in the trends and patterns of phosphorus concentration in the river over the three-year period. However, a closer examination reveals some intriguing differences. Specifically, our study’s lower limit for total phosphorus is notably lower than that reported by the station, indicating potentially lower levels of phosphorus in certain parts of the river or at specific times. This could be attributed to various factors such as seasonal variations, changes in river flow rate, or even differences in sampling locations and methodologies. Similarly, for total nitrogen concentration, our study’s range of 0.641 mg/L to 1.147 mg/L shows a considerable overlap with the station’s range of 0.470 mg/L to 1.030 mg/L [50]. Once again, this indicates a broad agreement in the overall trends and patterns of nitrogen concentration in the river. Nonetheless, our study’s upper limit for total nitrogen is slightly higher than that reported by the station, suggesting potentially higher levels of nitrogen in some areas or during certain periods. These differences, while subtle, could have important implications for understanding the river’s water quality and its potential impact on the surrounding ecosystem. For instance, even slight variations in phosphorus and nitrogen concentrations can significantly affect algal growth and the overall health of the aquatic environment. While our study’s results are generally consistent with the actual observations reported by the Shaanxi Provincial Aquatic Research and Technology Promotion Station, the noted differences highlight the importance of continuous and comprehensive monitoring efforts to accurately assess and manage water quality in the Hanjiang River. Future studies should aim to expand the scope and resolution of such monitoring efforts to gain a more nuanced understanding of the river’s water quality dynamics and their potential environmental impacts.
While the integrated use of drones and satellite remote sensing technology for water quality monitoring offers numerous advantages such as efficiency, wide coverage, and relatively high accuracy, the method also exhibits some weaknesses. Firstly, the flight altitude, stability, and the quality of the multispectral sensors carried by drones can affect the quality and resolution of the data collected. Complex or variable meteorological conditions, such as strong winds, low visibility, or high temperatures, can limit the operation of drones and subsequently impact data acquisition. Secondly, satellite remote sensing, despite its wide coverage, is often affected by factors such as revisit time, cloud cover, and atmospheric disturbances. This can lead to a lack of timely information, especially in scenarios where frequent or immediate monitoring is required. Moreover, both drones and satellite remote sensing rely on algorithms and models to estimate water quality parameters. These models may require adjustment or calibration in different water bodies and environmental conditions, and certain water quality parameters may be difficult to measure accurately solely through remote sensing techniques.

5. Conclusions

In this study, high-resolution satellite remote sensing data and unmanned aerial vehicle (UAV) multispectral remote sensing data were applied to conduct rapid monitoring of surface water quality parameters such as COD, TN, and TP in the mainstream area of the Hanjiang River in the Qinba mountainous region. During the work, we found that high-resolution satellite remote sensing data have the advantage of a large observation range, but it is difficult to ensure synchronized ground sampling and spectral measurements, which poses difficulties in establishing water quality parameter models. UAV multispectral data have the advantage of flexible observation and can ensure synchronized UAV remote sensing data acquisition and water sample collection.
This study employed traditional linear regression, power exponential regression, and polynomial regression methods for modeling, with the aim of validating the feasibility of water quality models based on drone multispectral data and high-resolution satellite data estimation methods. In future work, more advanced modeling schemes can be employed to obtain superior water quality estimation models. Therefore, the regional surface water quality detection method based on multi-rotor drone multispectral observation, synchronized ground sampling, and satellite remote sensing multispectral data are an effective and scalable approach that can be widely applied to water quality remote sensing modeling and detection in vast mountainous areas.

Author Contributions

Conceptualization, Y.C.; methodology, M.Y.; software, D.H.; validation, X.Z. and F.Y.; formal analysis, X.Z.; investigation, M.Y.; writing—original draft preparation, M.Y.; writing—review and editing, X.Z.; supervision, D.H.; project administration, Y.C.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Natural Science Basic Research Program of Shaanxi (Program No. 2022JQ-295 and 2021JM-350) and the National Natural Science Foundation of China (Program No. 42272342), and the APC was funded by Northwest Engineering Corporation Limited Major Science and Technology Projects, grant number XBY-ZDKJ-2020-08.

Data Availability Statement

The data presented in this study are available on request from the corresponding authors.

Acknowledgments

We are thankful to GeoCloud of the China Geological Survey for their provision of the Gaofen-2 high-resolution remote sensing data. The authors would like to thank the reviewers for their very helpful and constructive reviews of this manuscript.

Conflicts of Interest

Author Dianchao Han was employed by the company Power China Group, Northwest Engineering Corporation Limited. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Wang, J.; Li, Y.; Huang, J.; Yan, T.; Sun, T. Growing water scarcity, food security and government responses in China. Glob. Food Secur. 2017, 14, 9–17. [Google Scholar] [CrossRef]
  2. Huang, J.; Zhang, Y.; Bing, H.; Peng, J.; Dong, F.; Gao, J.; Arhonditsis, G. Characterizing the river water quality in China: Recent progress and on-going challenges. Water Res. 2021, 201, 117309. [Google Scholar] [CrossRef] [PubMed]
  3. The Ministry of Ecological Environment Announced the Environmental Quality of Surface Water in the Fourth Quarter of 2022 and from January to December in China. Available online: https://www.mee.gov.cn/ywdt/xwfb/202301/t20230129_1014067.shtml (accessed on 29 January 2023).
  4. Frumin, G.T.; Gildeeva, I.M. Eutrophication of water bodies—A global environmental problem. Russ. J. Gen. Chem. 2014, 84, 2483–2488. [Google Scholar] [CrossRef]
  5. Chen, T. Design and Implementation of Water Pollution Monitoring Visualization Based on IDL. Master’s Degree, University of Electronic Science and Technology of China, Chengdu, China, 2010. [Google Scholar]
  6. Wang, X.; Shi, Q.; Abdusalih, N.; Pan, X. Review on Monitoring of Lake Water Quality with the Remote Sensing Technology in Arid Areas of the West China. J. Agric. Sci. 2005, 26, 70–74. [Google Scholar]
  7. Bouraoui, F.; Grizzetti, B. Modelling mitigation options to reduce diffuse nitrogen water pollution from agriculture. Sci. Total Environ. 2014, 468, 1267–1277. [Google Scholar] [CrossRef] [PubMed]
  8. Branch, R.A. Remote Sensing of Water Surface Variability Near River Mouths. Doctoral Dissertation, University of Washington, Seattle, WA, USA, 2020. [Google Scholar]
  9. Chu, H.-J.; He, Y.-C.; Chusnah, W.N.; Jaelani, L.M.; Chang, C.-H. Multi-Reservoir Water Quality Mapping from Remote Sensing Using Spatial Regression. Sustainability 2021, 13, 6416. [Google Scholar] [CrossRef]
  10. Zhou, C.; Ye, H.; Sun, D.; Yue, J.; Yang, G.; Hu, J. An automated, high-performance approach for detecting and characterizing broccoli based on UAV remote-sensing and transformers: A case study from Haining, China. Int. J. Appl. Earth Obs. Geoinf. 2022, 114, 103055. [Google Scholar] [CrossRef]
  11. Zhou, D.; Xiao, J.; Bonafoni, S.; Berger, C.; Deilami, K.; Zhou, Y.; Frolking, S.; Yao, R.; Qiao, Z.; Sobrino, J.A. Satellite Remote Sensing of Surface Urban Heat Islands: Progress, Challenges, and Perspectives. Remote Sens. 2019, 11, 48. [Google Scholar] [CrossRef]
  12. Liu, Y.; Xia, K.; Feng, H.; Fang, Y. Inversion of water quality elements in small and micro-size water region using multispectral image by UAV. J. Environ. Sci. 2019, 39, 1241–1249. [Google Scholar]
  13. Shi, R.; Zhang, H.; Yue, R.; Zhang, X.; Wang, M.; Mei, S. A wavelet theory based remote sensing inversion of chlorophyll a concentrations for inland lakes in arid areas using TM image data. Acta Ecol. Sin. 2017, 37, 1043–1053. [Google Scholar]
  14. Adjovu, G.E.; Stephen, H.; James, D.; Ahmad, S. Overview of the Application of Remote Sensing in Effective Monitoring of Water Quality Parameters. Remote Sens. 2023, 15, 1938. [Google Scholar] [CrossRef]
  15. Li, Y.; He, B. A Semi-Empirical Retrieval Method of Above-Ground Live Forest Fuel Loads by Combining SAR and Optical Data. Remote Sens. 2023, 15, 5. [Google Scholar] [CrossRef]
  16. Zhan, Y.; Wang, X.; Jiang, S. Study on the artificial neural network model for water quality remote sensing inversion in the Wei River. Remote Sens. Technol. Appl. 2009, 24, 63–67. [Google Scholar]
  17. Wang, X.; Zhou, Z.; Yan, J. Application of GA-SVM for multispectral remote sensing inversion of water quality parameters in the Wei River. J. Remote Sens. 2009, 13, 740–744. [Google Scholar]
  18. Yuan, Q.; Shen, H.; Li, T.; Li, Z.; Li, S.; Jiang, Y.; Xu, H.; Tan, W.; Yang, Q.; Wang, J.; et al. Deep learning in environmental remote sensing: Achievements and challenges. Remote Sens. Environ. 2020, 241, 111716. [Google Scholar] [CrossRef]
  19. Li, A.; Fan, M.; Qin, G.; Xu, Y.; Wang, H. Comparative Analysis of Machine Learning Algorithms in Automatic Identification and Extraction of Water Boundaries. Appl. Sci. 2021, 11, 10062. [Google Scholar] [CrossRef]
  20. Zhao, S.; Hou, Y.; Zhang, A.; Wang, H.; Ma, J. Remote sensing inversion of water quality of Fuyang river based on UAV multi-spectrral images. Geomat. Spat. Inf. Technol. 2022, 3, 81–85. [Google Scholar]
  21. Jia, Z.; Lv, T. Geological Characteristics and Stone-bearing Coal Analysis of Dove Formation in Shuanglong Area, Ankang City. Coal 2014, 23, 52–53. [Google Scholar]
  22. Gong, H.; Xu, Y.; Chen, H.; Ke, H. Temporal and Spatial Variation and Influencing Factors of Heavy Metals in the Water of the Chenjiagou River in a Certain Place. Northwestern Geol. 2023, 4, 169–183. [Google Scholar]
  23. Yang, X.; Ma, H. Thought on the characteristics and comprehensive utilization of stone coal. Shaanxi Coal 2013, 1, 11–13. [Google Scholar]
  24. Jia, Z. Geological Characteristics of the Stone Coal Belt in Haoping, Ankang City. Value Eng. 2014, 33, 310–311. [Google Scholar]
  25. Choosumrong, S.; Hataitara, R.; Sujipuli, K.; Chitpan, M. Bananas diseases and insect infestations monitoring using multi-spectral camera RTK UAV images. Spat. Inf. Res. 2023, 31, 371–380. [Google Scholar] [CrossRef]
  26. Yue, J.; Feng, H.; Jin, X.; Yuan, H.; Li, Z.; Zhou, C.; Yang, G.; Tian, Q. A Comparison of Crop Parameters Estimation Using Images from UAV-Mounted Snapshot Hyperspectral Sensor and High-Definition Digital Camera. Remote Sens. 2018, 10, 1138. [Google Scholar] [CrossRef]
  27. GB 11893-89; Water Quality-Determination of Total Phosphorus-Ammonium Molybdate Spectrophotometric Method. China Environmental Science Press Co., Ltd.: Beijing, China, 1990.
  28. HJ 828-2017; Water Quality-Determination of the Chemical Oxygen Demand-Dichromate Method. China Environmental Science Press Co., Ltd.: Beijing, China, 2017.
  29. HJ 636-2012; Water Quality-Determination of Total Nitrogen-Alkaline Potassium Persulfate Digestion UV Spectrophotometric Method. China Environmental Science Press Co., Ltd.: Beijing, China, 2012.
  30. Liang, S.; Wei, H.; Gan, F.; Chen, L.; Xiao, C. Preliminary Application Evaluation of GF-2 Satellite Data in Remote Sensing Geological Survey. Spacecr. Recovery Remote Sens. 2015, 4, 63–72. [Google Scholar]
  31. Prieto-Amparan, J.A.; Villarreal-Guerrero, F.; Martinez-Salvador, M.; Manjarrez-Domínguez, C.; Santellano-Estrada, E.; Pinedo-Alvarez, A. Atmospheric and Radiometric Correction Algorithms for the Multitemporal Assessment of Grasslands Productivity. Remote Sens. 2018, 10, 219. [Google Scholar] [CrossRef]
  32. Huang, Z.; Chen, Q.; Chen, Q.; Liu, X. Variational Pansharpening for Hyperspectral Imagery Constrained by Spectral Shape and Gram–Schmidt Transformation. Sensors 2018, 18, 4330. [Google Scholar] [CrossRef]
  33. Hamad, R.; Balzter, H.; Kolo, K. Predicting Land Use/Land Cover Changes Using a CA-Markov Model under Two Different Scenarios. Sustainability 2018, 10, 3421. [Google Scholar] [CrossRef]
  34. Yang, X.; Zhao, S.; Qin, X.; Zhao, N.; Liang, L. Mapping of Urban Surface Water Bodies from Sentinel-2 MSI Imagery at 10 m Resolution via NDWI-Based Image Sharpening. Remote Sens. 2017, 9, 596. [Google Scholar] [CrossRef]
  35. Zhao, Q.; Han, L.; Wang, S.; Zhang, D.; Fan, J.; Sun, G.; Yang, J.; Li, Y.; Wang, X. Study on the inversion of COD in water bodies in Guangzhou based on remote sensing. Sci. Technol. Innov. 2021, 9, 38–39. [Google Scholar]
  36. Liu, J.; Kuang, R.; Li, J.; Hu, M. Inversion Model of TN, TP Concentration Based on Measured Spectral Reflectance Data in Poyang Lake. Southwest China J. Agric. Sci. 2020, 9, 2088–2094. [Google Scholar]
  37. GB 3838-2002; Environmental Quality Standards for Surface Water. China Environmental Science Press Co., Ltd.: Beijing, China, 2002.
  38. DB61-262-1997; Dividing the Functional Zones of Surface Water Bodies in the Hanjiang River System (Shaanxi Section). China Environmental Science Press Co., Ltd.: Beijing, China, 1998.
  39. Xu, Y.; Chen, H.; Ke, H.; Gong, H.; Cheng, X.; Zhang, M.; Wang, X.; Zhao, Z. Analysis of White Pollution of River Aluminum in Stone Coalmining Area in Haoping River Basin and Its Causes. Northwestern Geol. 2023, 4, 128–140. [Google Scholar]
  40. Zhang, F.; Chan, N.W.; Liu, C.; Wang, X.; Shi, J.; Kung, H.-T.; Li, X.; Guo, T.; Wang, W.; Cao, N. Water Quality Index (WQI) as a Potential Proxy for Remote Sensing Evaluation of Water Quality in Arid Areas. Water 2021, 13, 3250. [Google Scholar] [CrossRef]
  41. Gholizadeh, M.H.; Melesse, A.M.; Reddi, L. A Comprehensive Review on Water Quality Parameters Estimation Using Remote Sensing Techniques. Sensors 2016, 16, 1298. [Google Scholar] [CrossRef] [PubMed]
  42. Chen, B.; Mu, X.; Chen, P.; Wang, B.; Choi, J.; Park, H.; Xu, S.; Wu, Y.; Yang, H. Machine learning-based inversion of water quality parameters in typical reach of the urban river by UAV multispectral data. Ecol. Indic. 2021, 133, 108434. [Google Scholar] [CrossRef]
  43. Koparan, C.; Koc, A.B.; Privette, C.V.; Sawyer, C.B. In Situ Water Quality Measurements Using an Unmanned Aerial Vehicle (UAV) System. Water 2018, 10, 264. [Google Scholar] [CrossRef]
  44. Essam, S.; Zhang, Y.; Suliman, A. Mapping concentrations of surface water quality parameters using a novel remote sensing and artificial intelligence framework. Int. J. Remote Sens. 2017, 38, 1023–1042. [Google Scholar]
  45. Sandau, R.; Bries, K.; D’Errico, M. Small satellites for global coverage: Potential and limits. ISPRS J. Photogramm. Remote Sens. 2010, 65, 492–504. [Google Scholar] [CrossRef]
  46. Xi, X.; Zhao, G.; Gao, P.; Cui, K.; Li, T. Inversion of soil salinity in coastal winter wheat growing area based on sentinel satellite and unmanned aerial vehicle multi-spectrum: A case study in kenli district of the Yellow River Deltae. Sci. Agric. Sin. 2020, 53, 5005. [Google Scholar]
  47. Yang, X.; Zan, M.; Munige, M. Estimation of above ground biomass of Populus euphratica forest using UAV and satellite remotesensing. Trans. Chin. Soc. Agric. Eng. 2021, 37, 77. [Google Scholar]
  48. Gao, Y.; Liang, Z.; Wang, B.; Wu, Y.; Liu, S. UAV and satellite re-mote sensing images based above ground biomass inversion in the meadows of Shengjin Lake. J. Lake Sci. 2019, 31, 517. [Google Scholar]
  49. Cillero, C.; Domínguez, G.; Delgado, M. An UAV and satellite multispectral data approach to monitor water quality in small reservoirs. Remote Sens. 2020, 12, 1514. [Google Scholar] [CrossRef]
  50. Qu, S.; Meng, Q.; Bo, L. Monitoring analysis and evaluation of physical and chemical indicators of Yinghu water quality. China Fish. 2023, 6, 68–70. [Google Scholar]
Water 16 00732 g001
Figure 1. The map of the study area.
Figure 1. The map of the study area.
Water 16 00732 g001
Water 16 00732 g002
Figure 2. The distribution map of UAV data and water sample data collection points.
Figure 2. The distribution map of UAV data and water sample data collection points.
Water 16 00732 g002
Water 16 00732 g003
Figure 3. Comparison of UAV and Gaofen-2 multispectral data. ((a) Spatial distribution of relative deviation in R/G band ratio between UAV and Gaofen-2 images; (b) Spatial distribution of relative deviation in NIR/G band ratio between UAV and Gaofen-2 images; (c) Histogram distribution of relative deviation in R/G band ratio between UAV and Gaofen-2 images; (d) Histogram distribution of relative deviation in NIR/G band ratio between UAV and Gaofen-2 images).
Figure 3. Comparison of UAV and Gaofen-2 multispectral data. ((a) Spatial distribution of relative deviation in R/G band ratio between UAV and Gaofen-2 images; (b) Spatial distribution of relative deviation in NIR/G band ratio between UAV and Gaofen-2 images; (c) Histogram distribution of relative deviation in R/G band ratio between UAV and Gaofen-2 images; (d) Histogram distribution of relative deviation in NIR/G band ratio between UAV and Gaofen-2 images).
Water 16 00732 g003
Water 16 00732 g004
Figure 4. The estimation results of COD concentration based on GaoFen-2 data.
Figure 4. The estimation results of COD concentration based on GaoFen-2 data.
Water 16 00732 g004
Water 16 00732 g005
Figure 5. The estimation results of TP concentration based on Gaofen-2 data.
Figure 5. The estimation results of TP concentration based on Gaofen-2 data.
Water 16 00732 g005
Water 16 00732 g006
Figure 6. The estimation results of TN concentration based on Gaofen-2 data.
Figure 6. The estimation results of TN concentration based on Gaofen-2 data.
Water 16 00732 g006
Table 1. The instrumental parameters of the DJI Phantom 4 multispectral system.
Table 1. The instrumental parameters of the DJI Phantom 4 multispectral system.
Multispectral Bands (nm)Spatial Resolution (cm)Focal Length (mm)Field of VIEW (°)Aperture
450/560/650/730/840105.7462.7f/2.2
Table 2. Parameters of Gaofen-2 data.
Table 2. Parameters of Gaofen-2 data.
SensorsBandsSpectral Range (nm)Spatial Resolution of Subsatellite Point (m)Swath Width (km)Sway (°)Pass Frequency (Day−1)
PAN1450~9000.820±261
MSI1450~5203.220±261
2520~590
3630~690
4770~890
Table 3. The water quality estimation models.
Table 3. The water quality estimation models.
ParametersModelsR2
Chemical oxygen demand (COD) y = 37.465 R G − 10.662 0.804
Total phosphorus (TP) y = 0.2414 R G 0.8648 0.808
Total nitrogen (TN) y = 1.9126 N I R G 2 − 3.3941 N I R G + 2.1469 0.9177
Table 4. The water quality classification of Environmental Quality Standards for Surface Water (GB3838-2002). (Unit: mg/L).
Table 4. The water quality classification of Environmental Quality Standards for Surface Water (GB3838-2002). (Unit: mg/L).
Class IClass IIClass IIIClass IVClass V
COD≤15≤15≤20≤30≤40
TP≤0.02≤0.1≤0.2≤0.3≤0.4
TN≤0.2≤0.5≤1.0≤1.5≤2.0
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Han, D.; Cao, Y.; Yang, F.; Zhang, X.; Yang, M. Water Quality Estimation Using Gaofen-2 Images Based on UAV Multispectral Data Modeling in Qinba Rugged Terrain Area. Water 2024, 16, 732. https://doi.org/10.3390/w16050732

AMA Style

Han D, Cao Y, Yang F, Zhang X, Yang M. Water Quality Estimation Using Gaofen-2 Images Based on UAV Multispectral Data Modeling in Qinba Rugged Terrain Area. Water. 2024; 16(5):732. https://doi.org/10.3390/w16050732

Chicago/Turabian Style

Han, Dianchao, Yongxiang Cao, Fan Yang, Xin Zhang, and Min Yang. 2024. "Water Quality Estimation Using Gaofen-2 Images Based on UAV Multispectral Data Modeling in Qinba Rugged Terrain Area" Water 16, no. 5: 732. https://doi.org/10.3390/w16050732

APA Style

Han, D., Cao, Y., Yang, F., Zhang, X., & Yang, M. (2024). Water Quality Estimation Using Gaofen-2 Images Based on UAV Multispectral Data Modeling in Qinba Rugged Terrain Area. Water, 16(5), 732. https://doi.org/10.3390/w16050732

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop