Potential of solar-induced chlorophyll fluorescence (SIF) to access long-term dynamics of soil salinity using OCO-2 satellite data and machine learning method

• Published in: Geoderma 2024-04, Vol.444, p.116855, Article 116855
• Main Authors: Du, Ruiqi, Xiang, Youzhen, Chen, Junying, Lu, Xianghui, Wu, Yuxiao, He, Yujie, Xiang, Ru, Zhang, Zhitao, Chen, Yinwen
• Format: Article
• Language: English
• Subjects: GOSIF; OCO-2; Soil salt; Solar-induced fluorescence; Spaceborne observations
• ISSN: 0016-7061; 1872-6259
• DOI: 10.1016/j.geoderma.2024.116855

•Sensetivity of SIF to soil salt varies globally with land uses.•SIF has limitations on herbaceous cover and sparse vegetation.•Globally, 54.98 Mha salt-affected soils were observed during 2000～ 2020.•The United States and China are the two countries most affected by soil salinity. The accumulation of soil salt becomes a worldwide widespread phenomenon, being a major threat to global production. As an environmental stress, soil salinity can reduce the vegetation photosynthetic activity. Solar-induced chlorophyll fluorescence (SIF) is an electromagnetic signal actively released by vegetation during photosynthesis. SIF not only can capture lower vegetation photosynthetic activity due to environmental stress promptly, but also is less affected by atmosphere and soil background. However, the ability of SIF observation to detect soil salinity remains unclear. Here, we use standardizedsolar-induced chlorophyll iluorescence index (SIFI) from SIF observation time series (2000 ∼ 2020) of global OCO-2 based SIF product (GOSIF) to develop soil salinity model. The results show that: SIF observation can identify salt affected soil (EC ≥ 2 ∼ 4 dS m−1) and salinity class a global scale. The SIFI calculated from SIF observation at May ∼ August (hereafter SIFI5-8) is the optimal sensitivity indices for rainfed cropland, herbaceous cover, irrigated cropland, shrubland, grassland. SIFI10-11 is the optimal sensitivity indices for forest and sparse vegetation; (2) By comparison, the ovrerall classification accuracy of predicted salinity class is above 70 %. The order of classification accuracy is rainfed cropland > irrigated cropland > bare area > forest > grassland > shrubland > herbaceous cover > sparse vegetation; (3) During at least three-quarters of the period from 2000 to 2020, the area of salt affected soil (EC ≥ 2 ∼ 4 dS m−1) was 4.9 Mkm2; (4) The annual change rate of global soil salt content is generally between −0.05 and 0.05 dS m−1 yr−1. Soil salinity in South Africa and West Asia increased greatly with an annual change rate of 0.02 ∼ 0.03 dS m−1 yr−1. These results demonstrate the ability of SIF observation to estimate soil salinity, providing a new perspective for explaining and evaluating soil salinity variation.