Search

One reason for soil degradation is salinization in inland dryland, which poses a substantial threat to arable land productivity. Remote-sensing technology provides a rapid and accurate assessment for soil salinity monitoring, but there is a lack of high-resolution remote-sensing spatial salinity estimations. The PlanetScope satellite array provides high-precision mapping for land surface monitoring through its 3-m spatial resolution and near-daily revisiting frequency. This study's use of the PlanetScope satellite array is a new attempt to estimate soil salinity in inland drylands. We hypothesized that field observations, PlanetScope data, and spectral indices derived from the PlanetScope data using the partial least-squares regression (PLSR) method would produce reasonably accurate regional salinity maps based on 84 ground-truth soil salinity data and various spectral parameters, like satellite band reflectance, and published satellite salinity indices. The results showed that using the newly constructed red-edge salinity and yellow band salinity indices, we were able to develop several inversion models to produce regional salinity maps. Different algorithms, including Boruta feature preference, Random Forest algorithm (RF), and Extreme Gradient Boosting algorithm (XGBoost), were applied for variable selection. The newly constructed yellow salinity indices (YRNDSI and YRNDVI) had the best Pearson correlations of 0.78 and −0.78. We also found that the proportions of the newly constructed yellow and red-edge bands accounted for a large proportion of the essential strategies of the three algorithms, with Boruta feature preference at 80%, RF at 80%, and XGBoost at 60%, indicating that these two band indices contributed more to the soil salinity estimation results. The best PLSR model estimation for different strategies is the XGBoost-PLSR model with coefficient of determination (R2), root mean square error (RMSE), and ratio of performance to deviation (RPD) values of 0.832, 12.050, and 2.442, respectively. These results suggest that PlanetScope data has the potential to significantly advance the field of soil salinity research by providing a wealth of fine-scale salinity information. [ABSTRACT FROM AUTHOR]

• Compared to SNV, FOD is particularly effective at detecting subtle spectral signal changes; • The combined dimensionality reduction method (SPA-OBCA) can effectively identify soil salinity sensitive bands, enhancing the accuracy of salinization level assessment; • Compared to RF, SOA-RF performs better in evaluating soil salinity in mulched cotton fields; • A soil salinity map with a 5 cm resolution allows for the possibility of revealing subtle spatial variations in soil salinity; Xinjiang is a crucial production base for mulched cotton fields but faces increasingly severe challenges of soil salinization. Under the current drive towards precision agriculture, high precision monitoring of agricultural conditions at the field scale has become critical. Traditional remote sensing technologies are limited by insufficient spatial and spectral resolution, making field monitoring under mulched conditions exceptionally challenging. This study introduces a novel approach combining drone-based hyperspectral remote sensing with the Seagull Optimization Algorithm-Enhanced Random Forest Model (SOA-RF). After employing the preprocessing techniques of standard normal variable (SNV) and fractional order derivative (FOD), we employed the model to assess the level of field-scale salinization under optimization of feature selection using separate dimensionality reduction algorithms (sequential projection algorithm (SPA), optimal band combination analysis (OBCA)) and a combined approach (SPA-OBCA). The following main results were found: (1) SNV and FOD enhanced the spectral characteristics of cotton canopies, with the FOD technique performing better in detecting subtle signal changes in the positive and negative peaks of the spectrum. (2) SPA and OBCA were effective in screening the soil salt-sensitive bands in cotton canopies but the combined SPA-OBCA method provided a better selective preference for hyperspectral bands, with an optimal correlation of 0.926. (3) The model combined the advantages of SOA with the power of Random Forest to optimize the parameters and improve the estimation accuracy (R2 = 0.925, RPD = 4.128). We used unmanned aerial vehicle hyperspectral imagery to create a 5-cm resolution map of soil salinity distribution in cotton fields that allowed detection of differences in soil salinity between mulches, thus providing a scientific basis for soil salinity management and the development of precision agriculture. [ABSTRACT FROM AUTHOR]

Accurate assessment of the spatial distribution and severity of soil salinization has long plagued local governments and researchers in the arid parts of Xinjiang Uygur Autonomous Region (XJUAR). The emergence of machine learning has brought hope to this research field, such as Random Forest (RF) and Stochastic Gradient Treeboost (SGT),however, which are few applications to the quantitative assessment of soil salinization. Therefore, in order to evaluate the accuracy level of the two algorithms for predicting soil salinity, twenty-seven environmental subsets were designed. Each data set is calculated using both RF and SGT to produce an optimal set of variables. The simulation results from 70.37% (19/27) of the subsets showed that the predicted value of soil salinity from SGT is closer to the observed value than is that from RF. The statistics of all datasets showed that the average values of R2 value for RF and SGT were 0.38 and 0.40, the average Root Mean Squared Error (RMSE) value were 28.59 and 27.46, and the Ratio of Prediction to Deviation (RPD) averages were 1.20 and 1.24, respectively. The important dominant factor were topographic variables with coarse resolution, temperature and vegetation indices, land use and landform. [ABSTRACT FROM AUTHOR]

[Display omitted] • SNV pre-processing amplified the absorption characteristics of the spectrum. • Two variable selection algorithms were used to extract spectral variables. • OBCA improved the correlation of spectra with soil salinity vs. CARS. • RF has better stability in predicting soil salinity than PLSR. Hyperspectral remote sensing by unmanned aerial vehicle (UAV) is an important technical tool for rapid, accurate, and real-time monitoring of soil salinity in arid zone agroecosystems. However, the key to effective soil salinity (electrical conductivity, EC) prediction by UAV visible and near-infrared (Vis-NIR) spectroscopy depends on the selection of effective features selection techniques and robust prediction characteristics algorithms. Therefore, in this study, two advanced feature selection methods and two commonly used modeling methods were applied to predict and characterize the spatial patterns of soil salinity (EC). The aim of this study was to explore the predictive performance of different feature band selection methods and to identify a robust soil salinity mapping strategy. The results demonstrated that standard normal variate (SNV) pre-processing broadened the absorption characteristics of the spectrum. Compared with competitive adaptive reweighted sampling (CARS), the optimal band combination algorithm (OBCA) strengthened the correlation with soil salinity and had a higher variable importance in the modeling. Random forest (RF) was more stable in mapping the spatial pattern of surface soil salinity compared to the partial least squares regression model (PLSR). Our results confirm the effectiveness of OBCA and RF in the developing UAV remote sensing models for surface soil salinity estimation and mapping. [ABSTRACT FROM AUTHOR]

• GMM can distinguish similar spectral features in vegetation. • After spectral feature partition, the difference modeling is superior to whole-sample modeling. • UAV hyperspectral can indirectly reveal salinization patterns by inversion of saline vegetation SPAD. The chlorophyll content of saline vegetation can indirectly reflect salinization. Rapid and non-destructive capture of chlorophyll content of saline vegetation at a regional scale is essential for saline soil improvement and sustainability of saline agriculture. However, traditional soil–plant analyzer development (SPAD) monitoring based on SPAD-502 is carried out at the leaf scale, which does not allow rapid access to SPAD information for the whole region. In this study, we proposed that post-hyperspectral classification based on spectral differences that could contribute to enhanced estimation of SPAD in saline vegetation. To test this proposal, we partitioned the hyperspectral images using a Gaussian mixture model. Then, estimation models based on spectral partition and full-sample SPAD were developed using in situ observation data and a random forest model, respectively. Finally, the unmanned aerial vehicles (UAV) hyperspectral images were used as the input data source to digitally map the SPAD of saline vegetation in the region, using the prediction model. The results indicated that there were significant intensity and shape differences in the spectral reflectance characteristics of saline vegetation under different clusters. The SPAD prediction model, based on spectral feature partition, performed significantly better than the full sample. The SPAD maps of saline vegetation before and after clustering displayed similar spatial distribution models, but the prediction uncertainty of the models, based on spectral feature partition, was relatively low. Our results confirm the effectiveness and stability of UAV hyperspectral and spectral partition-based modeling in developing SPAD spatial distribution estimation models for saline vegetation. [ABSTRACT FROM AUTHOR]