Inversion of Chlorophyll-a Concentration in Shanghai Offshore Waters Based on BRDF Correction

doi:10.11988/ckyyb.20230373

Abstract

Abstract:

To investigate the ecological status of Shanghai’s offshore waters, we applied the water color remote sensing technology to estimate chlorophyll-a concentration. Considering the impact of BRDF (Bidirectional Reflectance Distribution Function) on inversion accuracy, we utilized LandSat-8 remote sensing data and measured water quality data in association with Lee’s model and QAA (Quasi-analytical algorithm) for BRDF correction. Results indicate that suspended sediment concentration is a major factor affecting BRDF. The angle of sunlight has minimal impact on BRDF when the sun is not vertically incident. After BRDF correction, the mean value of R² increased by 22.2% to 0.9, whereas mean RMSE decreased to 0.74. Before the correction, chlorophyll-a concentrations in Shanghai’s offshore waters were overestimated by an average of 2 mg/m³. BRDF correction notably enhances chlorophyll-a inversion accuracy, presenting innovative insights for offshore water color inversion.

Key words: chlorophyll-a, BRDF correction, offshore waters, Landsat-8

CLC Number:

X87环境遥感

HAN Zhen, CHEN Shuai-kang, LI Pi-xue, CHEN Hao-cheng. Inversion of Chlorophyll-a Concentration in Shanghai Offshore Waters Based on BRDF Correction[J]. Journal of Yangtze River Scientific Research Institute, 2024, 41(9): 169-177.

Figures/Tables 15

Fig.1 Technical flow chart of the method

Fig.2 Schematic diagram of water surface spectral reflection

Table 1 Values of G 0 w(Ω), G 1 w(Ω), G 0 p(Ω), G 1 p(Ω) at different sunlight angles

θ_s/(°)	θ_v/(°)	$φ$ /(°)	$G 0 w$ (Ω)	$G 1 w$ (Ω)	$G 0 p$ (Ω)	$G 1 p$ (Ω)
0	0	0	0.060 4	0.040 6	0.040 2	0.131 0
0	30	90	0.059 6	0.051 6	0.040 8	0.142 0
15	30	90	0.059 0	0.056 2	0.041 1	0.146 1
30	30	90	0.058 4	0.060 1	0.041 8	0.149 2
0	40	135	0.058 1	0.058 1	0.041 4	0.145 8
15	40	135	0.061 4	0.052 4	0.042 5	0.140 8
30	40	135	0.062 4	0.052 4	0.043 4	0.140 6

Table 1 Values of G 0 w(Ω), G 1 w(Ω), G 0 p(Ω), G 1 p(Ω) at different sunlight angles

θ_s/(°)	θ_v/(°)	$φ$ /(°)	$G 0 w$ (Ω)	$G 1 w$ (Ω)	$G 0 p$ (Ω)	$G 1 p$ (Ω)
0	0	0	0.060 4	0.040 6	0.040 2	0.131 0
0	30	90	0.059 6	0.051 6	0.040 8	0.142 0
15	30	90	0.059 0	0.056 2	0.041 1	0.146 1
30	30	90	0.058 4	0.060 1	0.041 8	0.149 2
0	40	135	0.058 1	0.058 1	0.041 4	0.145 8
15	40	135	0.061 4	0.052 4	0.042 5	0.140 8
30	40	135	0.062 4	0.052 4	0.043 4	0.140 6

Table 2 Steps of QAA_v6 algorithm

步骤	表达式
1	r_rs(λ)=R_rs(λ)/(0.52+1.7R_rs(λ))
2	u(λ)= $- g 0 + g 0 2 + 4 g 1 r r s (λ) 2 g 1$ ,g₀=0.089,g₁=0.124 5
3	a(λ₀)=a(670)=a_w(670)+0.39 $R r s (670) R r s (443) + R r s (490) 1.14$ , a_w(670)=0.439
4	b_bw(λ)=0.000 95(λ/550)^-4.3
5	b_bp(λ₀)=b_bp(670)= $u (λ 0) a (λ 0) 1 - u (λ 0)$ -b_bw(670)
6	η=2 $1 - 1.2 e x p - 0.9 r r s (443) r r s (555)$
7	b_bp(λ)=b_bp(λ₀) $λ 0 λ η$
8	a(λ)= $1 - u (λ) u (λ)$ b_bw(λ)+b_bp(λ)

Table 2 Steps of QAA_v6 algorithm

步骤	表达式
1	r_rs(λ)=R_rs(λ)/(0.52+1.7R_rs(λ))
2	u(λ)= $- g 0 + g 0 2 + 4 g 1 r r s (λ) 2 g 1$ ,g₀=0.089,g₁=0.124 5
3	a(λ₀)=a(670)=a_w(670)+0.39 $R r s (670) R r s (443) + R r s (490) 1.14$ , a_w(670)=0.439
4	b_bw(λ)=0.000 95(λ/550)^-4.3
5	b_bp(λ₀)=b_bp(670)= $u (λ 0) a (λ 0) 1 - u (λ 0)$ -b_bw(670)
6	η=2 $1 - 1.2 e x p - 0.9 r r s (443) r r s (555)$
7	b_bp(λ)=b_bp(λ₀) $λ 0 λ η$
8	a(λ)= $1 - u (λ) u (λ)$ b_bw(λ)+b_bp(λ)

Fig.3 Flowchart of BRDF correction

Fig.4 Offshore of Shanghai and distribution of sampling points

Fig.5 Distribution of bb/a under different water quality concentrations

Fig.6 Distribution of bb/a under different sunlight incident angles

Fig.7 Distribution of bb/a for water bodies with high suspended sediment concentration before and after BRDF correction

Fig.8 Single-band correlation analysis

Fig.9 Correlation analysis of band combinations

Fig.10 Comparison of chlorophyll-a concentration between measurements and inversion results

Table 3 Evaluation results for the accuracy of the chlorophyll-a inversion model

方法	R²	MRE/%	RMSE/(mg·m^-3)
校正前_经验法	0.759 8	36.48	1.16
校正后_经验法	0.926 3	25.40	0.63
校正前_NDCI	0.787 4	43.69	1.06
校正后_NDCI	0.904 8	26.21	0.71
校正前_WCI	0.669 0	47.72	1.32
校正后_WCI	0.868 1	19.84	0.87

Fig.11 Spatial distribution of chlorophyll-a concentration before and after BRDF correction

Fig.12 Difference in chlorophyll-a concentration before and after BRDF correction

[1]	ZHENG Pu, XU Xin-fa, XU Xiao-hua, CHEN Xin-yu, ZHANG Ren-gao. Retrieval of Chlorophyll-a Concentration in Poyang Lake by Using Multi-annual and Simultaneous Satellite-Ground Remote Sensing [J]. Journal of Yangtze River Scientific Research Institute, 2024, 41(7): 48-56.
[2]	QIAN Chun-long, ZENG Yi-chuan, YUAN Wei-hao, WU Yi. Optimal Construction of Chl-a Prediction Model for Poyang Lake Based on Time Series [J]. Journal of Yangtze River Scientific Research Institute, 2023, 40(10): 14-21.
[3]	CHENG Chun-mei,LI Yuan,DING Yi, TU Qian-guang,QIN Peng. Remote Sensing Estimation of Chlorophyll-a and Total Suspended Matter Concentration in Qiantang River Based on GF-1/WFV Data [J]. JOURNAL OF YANGTZE RIVER SCIENTIFIC RESEARCH INSTI, 2019, 36(1): 21-28.
[4]	HUANG Zhuo, CAO Xiao-huan, LIU Hong-fang, QIN Shuang. Synergistic Effect of Combined Mercury (Hg) and DiButyl Phthalate (DBP) on Scenedesmus quadricauda [J]. JOURNAL OF YANGTZE RIVER SCIENTIFIC RESEARCH INSTI, 2012, 29(10): 28-32.