Inversion of soil pH during the dry and wet seasons in the Yinbei region of Ningxia, China, based on multi-source remote sensing data
Introduction
Like drought, high salinity is one of the major issues limiting global agricultural production, particularly in arid and semi-arid regions. It is estimated that almost 950 million hectares of soil worldwide has low fertility due to salinization (Aldabaa et al., 2015). Soil alkalization is a process of soil salinization during which the replaceable sodium in soil colloids undergoes chemical substitution with other cations, resulting in the absorption of a large amount of Na+ ions; Na+ then binds CO32− to form Na2CO3, leading to increased soil pH and the formation of alkaline soil. Alkaline soil is characterized by a high pH and poor physical properties (e.g., poor structure and low permeability), resulting in soil compaction and retarded plant growth. Soil alkalization therefore has severe negative impacts on crop yield and agricultural productivity, threatening ecosystem health and economic sustainability (Li et al., 1998; Indrasumunar et al., 2012). According to statistical analyses, soil alkalinity is strongly affected by the monsoon climate (Li et al., 2016). Since seasonal precipitation in arid regions is unevenly distributed, the characteristics of alkaline soils show dynamic changes in terms of space and time. Thus, dynamic monitoring is essential in fully understanding the current status of soil alkalization and developing effective strategies aimed at soil restoration and land reclamation (Bannari et al., 2018). Zhang et al. (2018a) revealed that the annual mean precipitation and annual mean air temperature are strong predictors of regional soil pH levels (an indicator of soil acidity or alkalinity). Understanding the changes in soil pH that occur during different seasons is therefore important in understanding the dynamics of soil alkalization and preventing soil degradation.
Various techniques have been used to study soil pH, including traditional soil sampling and laboratory analyses, as well as field measurements using newly-developed sensors. The recent development of remote sensing technology (e.g., satellite platforms, global positioning systems, unmanned aerial vehicles, and ground sensors) combined with geographic information systems have provided new opportunities for research into soil science (Chai et al., 2015). Furthermore, with the application of spectral analysis, remote sensing has become a major approach during large-scale monitoring and evaluation of soil alkalization (Roelofsen et al., 2015; Wang et al., 2018a; Davis et al., 2019; Ren et al., 2020; dos Teixeira et al., 2020), and a number of studies have used remote sensing images to estimate soil pH (Liu et al., 2013; Li et al., 2018; Fu et al., 2020). For example, Miao et al. (2018) used the gray correlation model to estimate soil pH values across Hengshan County in Shanxi Province, China, based on hyperspectral data, while Zhang et al. (2018a) combined ground-measured hyperspectral and satellite image spectra to establish a soil pH estimation model with improved accuracy. Although the spectral indices used in previous studies can provide accurate estimates of soil alkalization, their effectiveness varies with the environmental conditions, land cover, soil type, and salt content (Wang et al., 2017; Farahmand et al., 2020). During spectral analysis, establishment of a reliable quantitative model is a prerequisite for accurate predictions of the type, composition, and nature of unknown samples. However, due to the complicated causes of soil alkalization and the sophisticated composition of soil pH, contradictory results have been obtained from different regions (Wang et al., 2017; Wang et al., 2019). Appropriate selection of feature variables is therefore crucial for quantitative analysis of spectral data. The most commonly used methods of variable selection include Pearson's correlation (Wang et al., 2019), Principal component analysis (PCA) (Xu et al., 2018), stepwise regression (SR) (Wang, 2019), gray relational analysis (GRA) (Wang, 2019), and variable importance in projection (Wang, 2019).
With inversion models, estimations of soil properties using soil spectra tend to be based on statistical analyses, such as multivariate stepwise regression (MLR) and partial least squares regression (PLSR) (Luan et al., 2013). In comparison, machine learning inversion models have strong nonlinear fitting capabilities and excellent data mining qualities that increase the use of spectral reflection data (Li et al., 2020). For example, Wang et al. (2018b) found that the machine learning algorithm showed strong data mining capability when a large number of variables were involved, and their correlations with soil salt content were relatively low. Li et al. (2017) found that the back propagation neural network (BPNN), support vector machine (SVM), and random forest (RF) also showed high inversion accuracy, far superior to statistical analysis models. However, the above-mentioned machine learning models tend to regard the relationship between independent and dependent variables as globally unchanged, ignoring those changes caused by spatial variability (Wu et al., 2016). To solve this issue, Fotheringham et al. (1996) proposed a local model, geographically weighted regression (GWR), which improved the inversion accuracy and effectiveness by introducing an unsteady relationship between the independent and dependent variables. To date, a number of models have been established to predict levels of soil salinization (Yuan et al., 2019; Li et al., 2019); yet, few reports have compared their prediction accuracy on alkalized soil when pH is used as the major indicator.
The Yinbei region in Ningxia, China, is an arid area with wide-spread occurrence of saline-alkaline soil. The total area of salinized soil in the Yinbei region is ~11.4 × 104 hm2 (The alkaline represented takyr solonetzs), approximately 2.15 times the area of non-saline farmland. Because of its flat terrain and deep soil layer, the Yinbei region is considered an important reserve of arable land in Ningxia. Accurate and efficient analyses of local soil salt contents and pH levels are therefore critical in assessing soil health and optimizing land management in this region. A number of studies have examined the mechanisms, and carried out dynamic monitoring and prediction of soil salinization on a local scale (Zhang et al., 2018b; Zhang et al., 2019a); however, most of these studies focus on the total soil salt content over a specific period of time (Sidike et al., 2014; Zhang et al., 2019b). Monitoring the spatiotemporal patterns of soil pH during different seasons is essential in understanding the dynamics of soil alkalization and facilitating soil improvements in the Yinbei region.
The objectives of this study were: (1) to explore the differences in soil pH during the dry and wet seasons, and determine the relationship between ground-measured hyperspectral data and Landsat 8 OLI image spectral data in the Yinbei region; and (2) to construct an optimal soil pH inversion model using BPNN, SVM, RR, and GWR based on different input variables selected by PCA, RS, and GRA. The results could provide a useful framework for the establishment of a spectral quantitative model of Solonetzs in this region.
Section snippets
Study area
Yinbei region (38°26′60″ - 39°14′09″ N, 105°57′40″ - 106°52′52″ E) is located in northern Ningxia Province, China (Fig. 1), covering an area of ~2060 km2 and representing part of the temperate arid region of northwest China. It lies between the diluvial fan and plain at the eastern foot of Helan Mountain. The study area experiences a warm temperate monsoon climate, with an annual mean temperature of 9 °C, low precipitation (annual mean: 150–203 mm), and strong evaporation (annual mean > 1825
Distribution of soil alkalinity variables
The results of soil physicochemical analysis provided general information on soil alkalinity in the study area (Table 3). Descriptive statistics revealed that the mean soil moisture content was lower in the dry season than the wet season (11.58% vs. 17.51%). In contrast, the mean soil pH in the dry season was 9.28, higher than in the wet season (9.11). Meanwhile, both the soil CO32− and HCO3− concentrations were markedly higher in the dry season than the wet season. The coefficients of
Discussion
According to the laboratory analyses (Table 3), the soil pH was slightly higher in the dry season than the wet season, possibly because more soluble substances are leached as a result of rainfall during the wet season. During the rainy period, soluble salt ions such as Na+, CO32−, and HCO3− are leached from the soil, while during the dry period, soluble ions migrate to the soil surface due to evaporation. Alternating rainy and dry periods can therefore lead to prominent changes in soil
Conclusions
This study used ground-measured hyperspectral data and Landsat 8 OLI remote sensing images to invert the soil pH in the Yinbei region of Ningxia. Different input variables and modeling methods (global and local regression) were combined to obtain the optimal spectral inversion model of soil pH during the dry and wet seasons. The main conclusions were as follows:
- (1)
The soil pH in the study area showed weak spatial variability, which was slightly greater in the dry season than the wet season. Under
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the National Natural Science Foundation of China (42067003).
References (64)
- et al.
Combination of proximal and remote sensing methods for rapid soil salinity quantification
Geoderma.
(2015) - et al.
Assessing soil salinity using soil salinity and vegetation indices derived from IKONOS high-spatial resolution imageries: applications in a date palm dominated region
Geoderma.
(2014) - et al.
Detecting salinity hazards within a semiarid context by means of combining soil and remote-sensing data
Geoderma.
(2006) - et al.
Remote sensing inversion of saline and alkaline land based on reflectance spectroscopy and D-TELM algorithm in Wuyuan areas
Infrared Phys. Technol.
(2020) - et al.
Laboratory prescreening of Bradyrhizobium japonicum for low pH, Al and Mn tolerance can be used to predict their survival in acid soils
Soil Biol. Biochem.
(2012) - et al.
Assessment of hydrosaline land degradation by using a simple approach of remote sensing indicators
Agric. Water Manage
(2005) - et al.
Modelling the electrical conductivity of soil in the Yangtze delta in three dimensions
Geoderma.
(2016) - et al.
Spatial multi-scale variability of soil nutrients in relation to environmental factors in a typical agricultural region, Eastern China
Sci. Total Environ.
(2013) - et al.
Estimating soil salinity from remote sensing and terrain data in southern Xinjiang Province, China
Geoderma
(2019) - et al.
An evaluation of remote sensing derived soil pH and average spring groundwater table for ecological assessments
Int. J. Appl. Earth Obs.
(2015)
Estimating soil salinity in Pingluo County of China using QuickBird data and soil reflectance spectra
Int. J. Appl. Earth Obs.
Machine learning regression algorithms for biophysical parameter retrieval: opportunities for Sentinel-2 and -3
Remote Sens. Environ.
Estimation of soil salt content (SSC) in the Ebinur Lake wetland National Nature Reserve (ELWNNR), Northwest China, based on a bootstrap-BP neural network model and optimal spectral indices
Sci. Total Environ.
Capability of Sentinel-2 MSI data for monitoring and mapping of soil salinity in dry and wet seasons in the Ebinur Lake region, Xinjiang, China
Geoderma.
Estimating temporal changes in soil pH in the black soil region of Northeast China using remote sensing
Comput. Electron. Agric.
Experimental study on shear strength characteristics of sulfate saline soil in Ningxia region under long-term freeze-thaw cycles
Cold Reg. Sci. Technol.
Prediction of salinity ion content in different soil layers based on hyperspectral data
Trans. Chin. Soc. Agric. Eng.
A geostatistical Vis-NIR spectroscopy index to assess the incipient soil salinization in the Neretva River valley, Croatia
Geoderma.
Using Remote Sensing Techniques for Appraisal of Irrigated Soil Salinity. // New Zealand: MODSIM 2007 International Congress on Modelling and Simulation
Patterns of variation in soil salinity: example of a salt marsh in SE of Spain
Arid Soil Res. Rehabil.
Sentinel-MSI VNIR and SWIR bands sensitivity analysis for soil salinity discrimination in an arid landscape
Remote Sens.
Toward a semiautomatic machine learning retrieval of biophysical parameters
IEEE J-STARS
Extraction and modeling of regional soil salinization based on data from GF-1 satellite
Acta Pedol. Sin.
The effect of the geomorphologic type as surrogate to the time factor on digital soil mapping
Open J. Soil Sci.
Comparing Sentinel-2 MSI and Landsat 8 OLI in soil salinity detection: a case study of agricultural lands in coastal North Carolina
Int. J. Remote Sens.
The Basic Method of Grey System (in Chinese)
Tropical soil pH and sorption complex prediction via portable x-ray fluorescence spectrometry
Geoderma.
Estimating soil salinity in the dried Lake bed of Urmia Lake using optical sentinel-2B images and multivariate linear regression models
J. India Soc. Remote
The geography of parameter space: an investigation of spatial non-stationarity
Int. J. Geogr. Inf. Sci.
Combining UAV-based hyperspectral imagery and machine learning algorithms for soil moisture content monitoring
PeerJ
Inversion of net primary productivity in the arid region of Northwest China based on various regressions
Resour. Sci.
Calibration and validation of soil salinity estimation model based on measured hyperspectral and Aster image
Trans. Chin. Soc. Agric. Eng.
Cited by (17)
Spatiotemporal dynamics and anthropologically dominated drivers of chlorophyll-a, TN and TP concentrations in the Pearl River Estuary based on retrieval algorithm and random forest regression
2022, Environmental ResearchCitation Excerpt :Traditional Chl-a monitoring measures including ground sampling, transportation and laboratory chemical analysis require massive labor work, cost a big expenditure and take a long research cycle, which can only reflect the water quality around the monitoring points with a limited data size; therefore, monitoring measures with short operation cycle, wide ground coverage and convenient data acquisition are needed (Tsatsaros et al., 2021). Remote sensing monitoring and estimating photosensitivity parameters based on the established retrieval algorithm have been applied to various regions and water bodies (Fu, 2021; Han et al., 2021; Jia et al., 2021). The number of studies on the retrieval model of water quality parameters, such as Chl-a, clarity, and total suspended solids, based on watercolor remote sensing theory is growing rapidly (Isada et al., 2022; Jia et al., 2021; Wang et al., 2021b).
Comparison of the use of Landsat 8, Sentinel-2, and Gaofen-2 images for mapping soil pH in Dehui, northeastern China
2022, Ecological InformaticsCitation Excerpt :The recent development of remote sensing technology, including satellites, airborne sensors, and unmanned aerial vehicles, combined with geographic information systems has provided a new approach for soil science (Chai et al., 2015). Due to the recent improvements in the spatial and spectral resolutions of the data obtained using these methods, many studies on soil pH prediction have been conducted using optical satellite imagery (Pahlavan-Rad and Akbarimoghaddam, 2018; Zhang et al., 2018; Fu et al., 2020; Jia et al., 2021). Zhang et al. (2018) combined Landsat imagery and ground-measured hyperspectral data to predict the soil pH with an improved accuracy.
Estimation of soil water and organic matter content in medium and low yield fields of Ningxia Yellow River Irrigation area based on hyperspectral information
2023, Chinese Journal of Applied EcologyInversion of soil water and salt information based on UAV hyperspectral remote sensing and machine lear⁃ ning
2023, Chinese Journal of Applied Ecology