MODIS PWV neural network differential correction model integrating multiple nonlinear factors

doi:10.11947/j.AGCS.2024.20230233

Abstract

Abstract:

MODIS water vapor products have become important atmospheric water vapor productsdue to their advantages of high spatial resolution, however, due to uncertain factors such as precipitation, cloud mask, and surface reflection spectrum, the inversion accuracy is limited. To effectively improve the quality of MODIS water vapor products, this article analyzes nonlinear factors such as cloud, land cover type, pixel attitude, time, and position, and a MODIS PWV neural network differential correction model integrating multiple nonlinear factors is constructed for the first time. Firstly, the correlation between MODIS PWV and GNSS PWVat the same site is analyzed, PWV_diff (the difference between the two) is taken as the target value of the neural network model, and the input information for the model is 19 nonlinear factors such as cloud mask confidence, land cover type, periodic day of year, periodic hour of day, sensor zenith angle, solar zenith angle, sensor azimuth angle, and solar azimuth angle. Compared with traditional linear correction model, the root mean square error (RMSE) of corrected MODIS PWV is reduced from 3.271 3 mm to 2.360 2 mm, and the accuracy is improved by 27.85%. The performance of the model isalso evaluated by using high spatiotemporal ERA5 PWV data as a reference value. The experimental results show that the corrected RMSE of the corrected MODIS PWV using the proposed model in this paper is 2.037 4 mm, which improves the accuracy by 57.99% compared to the uncorrected MODIS PWV (RMSE=4.850 3 mm); furthermore, neural network differential correction models are constructed for MODIS PWV products under four different confidence levels provided by cloud mask products, the results show that the RMSEs of MODIS PWV products with cloud confidence levels >99%, >95%, >66%, and <66% are increased by 60.03%, 61.21%, 55.72%, and 54.57% compared to uncorrected MODIS PWV. This indicates that the model constructed in this article has high universality in improving the accuracy of MODIS PWV products under various cloud covers, and is expected to provide high-precision water vapor information for climate change and rainfall forecasting research.

Key words: MODIS PWV, GNSS PWV, PWV differential correction, BP neural network, ERA5 PWV

CLC Number:

P228

Mengyao WANG, Shubi ZHANG, Wenyuan ZHANG, Yang LIU. MODIS PWV neural network differential correction model integrating multiple nonlinear factors[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(11): 2138-2148.

Figures/Tables 17

Fig.1

Fig.2

Fig.3

Tab.1

Tab.2

Fig.4

Fig.5

Fig.6

Fig.7

Fig.8

Tab.3

Fig.9

Tab.4

Tab.5

Tab.6

Fig.10

Tab.7

References 27

[1]	SOLOMON S, ROSENLOF K H, PORTMANN R W, et al. Contributions of stratospheric water vapor to decadal changes in the rate of global warming[J]. Science, 2010, 327(5970): 1219-1223.
[2]	谷晓平, 王新明, 吴战平, 等. 基于MODIS近红外数据的贵州高原大气水汽反演研究[J]. 高原气象, 2009, 28(2): 446-451.
	GU Xiaoping, WANG Xinming, WU Zhanping, et al. Retrieval of atmospheric water vapor using MODIS near infrared channel data in Guizhou Plateau[J]. Plateau Meteorology, 2009, 28(2): 446-451.
[3]	ZHAO Qingzhi, YAO Wanqiang, YAO Yibin, et al. An improved GNSS tropospheric tomography method with the GPT2w model[J]. GPS Solutions, 2020, 24(2): 60.
[4]	刘备, 王勇, 娄泽生, 等. CMONOC观测约束下的中国大陆地区MODIS PWV校正[J]. 测绘学报, 2019, 48(10): 1207-1215. DOI:. doi: 10.11947/j.AGCS.2019.20180386
	LIU Bei, WANG Yong, LOU Zesheng, et al. The MODIS PWV correction based on CMONOC in Chinese mainland[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(10): 1207-1215. DOI:. doi: 10.11947/j.AGCS.2019.20180386
[5]	DE OLIVEIRA G, BRUNSELL N A, MORAES E C, et al. Evaluation of MODIS-based estimates of water-use efficiency in Amazonia[J]. International Journal of Remote Sensing, 2017, 38(19): 5291-5309.
[6]	GURBUZ G, JIN Shuanggen. Long-time variations of precipitable water vapour estimated from GPS, MODIS and radiosonde observations in Turkey[J]. International Journal of Climatology, 2017, 37(15): 5170-5180.
[7]	LI Xueying, LONG Di. An improvement in accuracy and spatiotemporal continuity of the MODIS precipitable water vapor product based on a data fusion approach[J]. Remote Sensing of Environment, 2020, 248: 111966.
[8]	张豹. 地基GNSS水汽反演技术及其在复杂天气条件下的应用研究[D]. 武汉: 武汉大学, 2016.
	ZHANG Bao. Ground-based GNSS water vapor inversion technology and its application in complex weather conditions[D]. Wuhan: Wuhan University, 2016.
[9]	赵庆志. 地基GNSS水汽反演关键技术研究及其应用[D]. 武汉: 武汉大学, 2017.
	ZHAO Qingzhi. Research on key technologies of ground-based GNSS water vapor retrieval and its application[D]. Wuhan: Wuhan University, 2017.
[10]	LI Zhenhong, MULLER J P, CROSS P. Comparison of precipitable water vapor derived from radiosonde, GPS, and moderate-resolution imaging spectroradiometer measurements[J]. Journal of Geophysical Research: Atmospheres, 2003, 108(D20): e2003jd003372.
[11]	张文渊, 郑南山, 张书毕, 等. 附加高水平分辨率PWV约束的GNSS水汽层析算法[J]. 武汉大学学报(信息科学版), 2021, 46(11): 1627-1635.
	ZHANG Wenyuan, ZHENG Nanshan, ZHANG Shubi, et al. GNSS water vapor tomography algorithm constrained with high horizontal resolution PWV data[J]. Geomatics and Information Science of Wuhan University, 2021, 46(11): 1627-1635.
[12]	YAO Yibin, SHAN Lulu, ZHAO Qingzhi. Establishing a method of short-term rainfall forecasting based on GNSS-derived PWV and its application[J]. Scientific Reports, 2017, 7(1): 12465.
[13]	马赛, 岳迎春, 上官明, 等. 基于GNSS的MODIS大气可降水量校正模型[J]. 南京信息工程大学学报(自然科学版), 2021, 13(2): 154-160.
	MA Sai, YUE Yingchun, SHANGGUAN Ming, et al. A correction model for MODIS precipitable water vapor based on GNSS data[J]. Journal of Nanjing University of Information Science & Technology (Natural Science Edition), 2021, 13(2): 154-160.
[14]	GUI Ke, CHE Huizheng, CHEN Quanliang, et al. Evaluation of radiosonde, MODIS-NIR-Clear, and AERONET precipitable water vapor using IGS ground-based GPS measurements over China[J]. Atmospheric Research, 2017, 197: 461-473.
[15]	王佳, 李琼, 魏加华. 青海MODIS可降水量数据校正研究[J]. 青海大学学报, 2021, 39(2): 77-84.
	WANG Jia, LI Qiong, WEI Jiahua. Study on the data correction of MODIS precipitable water vapor of Qinghai province[J]. Journal of Qinghai University, 2021, 39(2): 77-84.
[16]	张环宇, 唐伯惠. 融合物理机理与随机森林算法的FY-4A AGRI数据晴空大气可降水量遥感反演[J]. 遥感学报, 2021, 25(8): 1836-1847.
	ZHANG Huanyu, TANG Bohui. Remote sensing retrieval of total precipitable water under clear-sky atmosphere from FY-4A AGRI data by combining physical mechanism and random forest algorithm[J]. National Remote Sensing Bulletin, 2021, 25(8): 1836-1847.
[17]	CHANG Liang, XIAO Ruya, PRASAD A A, et al. Cloud mask-related differential linear adjustment model for MODIS infrared water vapor product[J]. Remote Sensing of Environment, 2019, 221: 650-664.
[18]	KHATRI P, HAYASAKA T. Comparison of cloud properties between SGLI aboard GCOM-C satellite and MODIS aboard terrasatellite[J]. Remote Sensing, 2023, 15(4): 1075.
[19]	TIAN Guoliang, HUANG Qiaolin, HE Hongyan, et al. Analysis of the influence of undulate terrain on the geometric accuracy of remote sensing images[C]//Proceedings of the 8th International Conference on Instrumentation & Measurement, Computer, Communication and Control. Harbin: IEEE, 2018.
[20]	龙腾飞, 焦伟利, 张兆明, 等. 卫星影像像元观测天顶角和方位角的计算方法: CN201410352963.8[P]. 2014-10-29.
	LONG Tengfei, JIAO Weili, ZHANG Zhaoming, et al. Calculation method of zenith angle and azimuth angle observed by satellite image element: CN201410352963.8[P]. 2014-10-29.
[21]	MA Xiongwei, YAO Yibin, ZHANG Bao, et al. An improved MODIS NIR PWV retrieval algorithm based on an artificial neural network considering the land-cover types[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 3170078.
[22]	WANG Zhaoyang, ZHOU Xinghua, LIU Yanxiong, et al. Precipitable water vapor characterization in the coastal regions of China based on ground-based GPS[J]. Advances in Space Research, 2017, 60(11): 2368-2378.
[23]	YUAN Peng, BLEWITT G, KREEMER C, et al. An enhanced integrated water vapour dataset from more than 10 000 global ground-based GPS stations in 2020[J]. Earth System Science Data, 2023, 15(2): 723-743.
[24]	王帅民. 基于GNSS和再分析资料的ZTD/PWV精度评定与模型构建方法研究[D]. 济南: 山东大学, 2021.
	WANG Shuaimin. Research on accuracy evaluation and model establishment of ZTD/PWV based on GNSS and reanalysis[D]. Jinan: Shandong University, 2021.
[25]	ZHU Dantong, ZHANG Kefei, YANG Liu, et al. Evaluation and calibration of MODIS near-infrared precipitable water vapor over China using GNSS observations and ERA-5 reanalysis dataset[J]. Remote Sensing, 2021, 13(14): 2761.
[26]	何清, 王震坤. 正态性检验方法在教学研究中的应用[J]. 高等理科教育, 2014(4): 18-21, 77.
	HE Qing, WANG Zhenkun. Discussion on the application of normality test method in teaching research[J]. Higher Education of Sciences, 2014(4): 18-21, 77.
[27]	沈花玉, 王兆霞, 高成耀, 等. BP神经网络隐含层单元数的确定[J]. 天津理工大学学报, 2008, 24(5): 13-15.
	SHEN Huayu, WANG Zhaoxia, GAO Chengyao, et al. Determining the number of BP neural network hidden layer units[J]. Journal of Tianjin University of Technology, 2008, 24(5): 13-15.

对比数据	均值/mm	平均绝对误差/mm	均方根误差/mm	相关系数
探空站PWV	0.212 1	0.892 8	1.274 5	0.988 2
GNSS PWV	0.065 3	1.471 6	2.113 9	0.947 3

模型	均值	平均绝对误差	均方根误差
未校正	-0.013 1	3.610 6	4.474 3
传统线性校正模型	-0.006 6	2.633 5	3.271 3
神经网络差分校正模型	-0.002 4	1.745 5	2.360 2

模型	均值	平均绝对误差	均方根误差
未校正	-0.681 9	3.748 2	4.850 3
神经网络差分校正模型	-0.007 7	1.566 6	2.037 4

方案	均值	平均绝对误差	均方根误差
方案Ⅰ	-0.005 6	1.866 2	2.517 2
方案Ⅱ	-0.001 8	1.855 1	2.478 6
方案Ⅲ	-0.007 0	1.858 1	2.494 5
方案Ⅳ	-0.003 0	2.297 9	2.956 4

云类型	模型信息	均值	平均绝对误差	均方根误差
确信晴空（99%）	未校正	-0.60 22	2.824 5	3.690 2
确信晴空（99%）	神经网络差分校正模型	0.053 7	1.067 1	1.475 0
可能晴空（95%）	未校正	-0.971 0	2.847 7	3.874 0
可能晴空（95%）	神经网络差分校正模型	0.091 9	1.197 3	1.685 5
不确定（66%）	未校正	-1.562 3	2.808 9	3.947 4
不确定（66%）	神经网络差分校正模型	0.046 1	1.239 5	1.747 9
云（cloud）	未校正	2.065 0	3.624 2	4.719 3
云（cloud）	神经网络差分校正模型	-0.002 9	1.651 1	2.144 2