Figure 3 - uploaded by Yue Jibo
Content may be subject to copyright.
Design of four field artificial scenes. Soil moisture of four scenes are: scene 1: 12.9%, scene 2: 22.1%, scene 3:
Source publication
Crop residue deposited on the soil surface helps protect against water and wind erosion, improve soil quality and increase soil organic matter and soil carbon storage. Linear spectral mixture analysis (LSMA) is an important technique in field crop residue estimation. Traditionally, only one single or fixed standard crop residue and soil endmember s...
Citations
... The concomitant changes in SM, SBW, and spectral reflectance exhibit a complex relationship. Generally, increases in SM, SBW, and RMSH decrease the spectral reflectance, showing a coupling effect [27,28]. Noteworthily, the effect of the SOM content on the soil spectrum is far weaker than that of soil physical properties [29]. ...
Soil organic matter (SOM) is important for the global carbon cycle, and hyperspectral remote sensing has proven to be a promising method for fast SOM content estimation. However, because of the neglect of the spectral response of soil physical properties, the accuracy and spatiotemporal transferability of the SOM prediction model are poor. This study aims to improve the spatiotemporal transferability of the SOM prediction model by alleviating the coupling effect of soil physical properties on spectra. Based on satellite hyperspectral images and soil physical variables, including soil moisture (SM), soil surface roughness (root-mean-square height, RMSH), and soil bulk weight (SBW), a soil spectral correction model was established based on the information unmixing method. Two important grain-producing areas in Northeast China were selected as study areas to verify the performance and transferability of the spectral correction model and SOM content prediction model. The results showed that soil spectral corrections based on fourth-order polynomials and the XG-Boost algorithm had excellent accuracy and generalization ability, with residual predictive deviations (RPDs) exceeding 1.4 in almost all the bands. In addition, when the soil spectral correction strategy was adopted, the accuracy of the SOM prediction model and the generalization ability after the model migration were significantly improved. The SOM prediction accuracy based on the XG-Boost-corrected spectrum was the highest, with a coefficient of determination (R2) of 0.76, a root-mean-square error (RMSE) of 5.74 g/kg, and an RPD of 1.68. The prediction accuracy, R2 value, RMSE, and RPD of the model after the migration were 0.72, 6.71 g/kg, and 1.53, respectively. Compared with the direct migration prediction of the model, adopting the soil spectral correction model based on fourth-order polynomials and XG-Boost reduced the RMSE of the SOM prediction results by 57.90% and 60.27%, respectively. This performance comparison highlighted the advantages for considering soil physical properties in regional-scale SOM predictions.
... The concomitant changes in SM, SBW, and spectral reflectance exhibit a complex relationship. Generally, the increase in SM, SBW, and RMSH decreases the spectral reflectance, showing a coupling effect [27,28]. Noteworthy, the effect of SOM content on the soil spectrum is far weaker than that of soil physical properties [29]. ...
Soil organic matter (SOM) is important for the global carbon cycle, and hyperspectral remote sensing has proven a promising method for fast SOM content estimation. However, soil physical properties significantly affect the sensitivity of satellite hyperspectral imaging to SOM, leading to poor generalization ability of the estimation model. This study aims to improve the spatiotemporal transferability of the SOM prediction model by alleviating the coupling effect of soil physical properties on the spectra. Based on satellite hyperspectral images and soil physical variables, including soil moisture (SM), soil surface roughness (root mean squared height, RMSH), and soil bulk weight (SBW), a soil spectral correction strategy was established based on the information unmixing method. Two important grain-producing areas in Northeast China were selected as study areas to verify the performance and transferability of the spectral correction model and SOM content prediction model. The results showed that soil spectral corrections based on fourth-order polynomials and the XG-Boost algorithm had excellent accuracy and generalization ability, with residual predictive deviations (RPD) exceeding 1.4 in almost all bands. In addition, when the soil spectral correction strategy was adopted, the accuracy of the SOM prediction model and the generalization ability after model migration were significantly improved. The SOM prediction accuracy based on the XG-Boost corrected spectrum was the highest, with a coefficient of determination (R2) of 0.76, root mean square error (RMSE) of 5.74 g/kg, and RPD of 1.68. The prediction accuracy, R2, RMSE, and RPD of the model after migration were 0.72, 6.71 g/kg, and 1.53, respectively. Compared with the direct migration prediction of the model, adopting the soil spectral correction strategy based on fourth-order polynomials and XG-Boost reduced the RMSE of the SOM prediction results by 57.90% and 60.27%, respectively. The performance comparison highlighted the advantages of considering soil physical properties in regional-scale SOM prediction.
... Pacheco et al. (2010) tested the potential of using multispectral remote-sensing spectral reflectance and LSMA to map cropland RRC. Yue et al. (2019b) developed a dynamic soil endmember spectrum-selection approach for LSMA-based RRC mapping. In addition, several LSMA-based RRC mapping methods have been developed using remote-sensing CRSIs. ...
Accurate determination of rice residue cover (RRC) can improve the monitoring of tillage information. Currently, the accurate determination of RRC using optical remote sensing is hindered by variations in cropland moisture and cover of following crops. The fractional cover (FC) of the soil (fS), crop (fC), and crop residue (fCR) changes (fS + fC + fCR = 1) after the following crop is planted, which increases the difficulty of remote-sensing RRC estimation. Cropland soil moisture and crop residue moisture affect the values of cropland and crop residue spectral indices (CRSIs), thereby reducing the accuracy of remote-sensing RRC estimation. Deep learning techniques (e.g., convolutional neural networks [CNN] and transfer learning [TL]) have been proven to extract the deep features of input images with distortion invariance, such as displacement and scaling, which are similar to moisture and the following crop effects on remote-sensing CRSIs.
This study aimed to evaluate the combined use of deep features of cropland spectra extracted by deep learning techniques to estimate the cropland RRC under the effects of variations in cropland moisture and cover of the following crops. This study proposes an RRCNet CNN that fuses deep and shallow features to improve RRC estimation. A PROSAIL radiative transfer model was employed to simulate a cropland “soil–crop–crop residue” mixed spectra dataset (n = 103,068), considering the variations in cropland moisture and the cover of the following crop. The RRCNet was first pre-trained using the simulated dataset, and then the knowledge from the pre-trained RRCNet was updated based on field experimental FCs, RRCs, and Sentinel-2 image spectra using the TL technique. Our study indicates that RRCNet can incorporate shallow and deep spectral features of cropland “soil–crop–crop residue” mixed spectra, providing high-performance FCs and RRC estimation. The FCs and RRC estimates from RRCNet + TL (FCs: R2 = 0.939, root mean squared error (RMSE) = 0.071; RRC: R2 = 0.891, RMSE = 0.083) were more accurate than those from CRSI + multiple linear regression, CRSI + random forest, and CRSI + support vector machine (FCs: R2 = 0.877–907, RMSE = 0.086–0.101; RRC: R2 = 0.378–0.714, RMSE = 0.133–0.229). We mapped the multistage RRC based on Sentinel-2 multispectral instrument (MSI) images and RRCNet. Tillage information can be inferred from RRC and RRC difference maps changes.
... Of these regions, rice residues are often returned to the field prior to planting wheat, leading to variable backgrounds comprising soil, residue and residue-soil mixtures. The rice residue causes changes to the spectral properties of the background, especially to the spectral shape (Yue et al., 2020;Yue et al., 2019b), and consequently to the soil line and VI-LAI relationships (Biard and Baret, 1997;Zhao et al., 2012). As reported by a recent study (Gao et al., 2022), the crop residue has a considerable effect on remotely measured wheat canopy spectra, and clearly affects the accuracy of existing VI-LAI models. ...
... With a zenith angle of 0 • , the spectrometer sensor uses a 25 • fore-optic mounted 1.2 m above the field, resulting in a field of view of 0.53 m diameter. 10 spectra were taken at the center of each dry sampling position (Yue et al., 2019). A white standardized panel (with 99% reflectance) of 40 cm × 40 cm was used to calibrate the spectrometer before each measurement. ...
A R T I C L E I N F O Keywords: Three dimensional index Normalized difference index Soil organic matter Visible and near-infrared spectroscopy Spectral pre-processing A B S T R A C T Accurate, rapid, and non-destructive estimation of soil organic matter (SOM) is crucial for soil fertility diagnosis and precision farming. Due to the complicated and unstable spectral characteristics of SOM, few SOM spectral indexes have been proposed and widely used. In this paper, a new dynamic normalized difference index (DNDI) was proposed and constructed to estimate SOM using visible and near-infrared spectroscopy. A dynamic correction factor α was used to adjust the optimal wavelength range to obtain more robust features of SOM. Different spectral pre-processing methods were applied and compared. The support vector machine (SVM) model and Partial least square regression (PLSR) model were calibrated based on DNDI and applied to estimate SOM. To this end, a total of 111 soil samples were collected in the southern coastal plain of Laizhou Bay. The results showed that the DNDI index constructed by wavelength optimization could have a higher correlation with SOM than the two-dimensional normalized difference index (NDI). DNDI had the maximum correlation of 0.88 from the first derivative of reflectance, and the NDI correlations were most improved by standard normal variate transform (SNV), with the maximum correlation reaching 0.81. For SOM estimation models, DNDI exhibited better model performance, yielding a validation R 2 , RMSE, and RPD of 0.78, 0.17 g⋅kg − 1, and 2.01, respectively. Our algorithm has strong application potential for estimating other soil properties and enhancing the application of ground-and satellite-based sensing.
... Recently, the potential relationship between spectral values and soil salinity has been the basic theory for estimating soil salinity by the hyperspectral technique [7][8][9]. However, mixed hyperspectral limits the accuracy of soil salt content (SSC) assessments in areas where the soil surface is partially covered with vegetation [10]. ...
Hyperspectral technology has proven to be an effective method for monitoring soil salt content (SSC). However, hyperspectral estimation capabilities are limited when the soil surface is partially vegetated. This work aimed to (1) quantify the influences of different fraction vegetation coverage (FVC) on SSC estimation by hyperspectra and (2) explore the potential for a non-negative matrix factorization algorithm (NMF) to reduce the influence of various FVCs. Nine levels of mixed hyperspectra were measured from simulated mixed scenes, which were performed by strictly controlling SSC and FVC in the laboratory. NMF was implemented to extract soil spectral signals from mixed hyperspectra. The NMF-extracted soil spectra were used to estimate SSC using partial least squares regression. Results indicate that SSC could be estimated based on the original mixed spectra within a 25.76% FVC (R 2 cv = 0.68, RMSE cv = 5.18 g·kg −1 , RPD = 1.43). Compared with the mixed spectra, NMF extraction of soil spectrum improved the estimation accuracy. The NMF-extracted soil spectra from FVC below 63.55% of the mixed spectra provided acceptable estimation accuracies for SSC with the lowest results of determination of the estimation R 2 cv = 0.69, RMSE cv = 4.15 g·kg −1 , and RPD = 1.8. Additionally, we proposed a strategy for the model performance investigation that combines spearman correlation analysis and model variable importance projection analysis. The NMF-extracted soil spectra retained the sensitive wavelengths that were significantly correlated with SSC and participated in the operation as important variables of the model.
... MSI plays a vital role in the identification of moisture contents on land surface features in remote sensing data (Yue et al., 2019). SAVI SAVI = (NIR−Red) (NIR+Red+L) (1 + L) SAVI is a modified NDVI index that adjusted the soil brightness influence on the NDVI results when the vegetation cover is low (Huete, 1988). ...
The 2005 Kashmir earthquake has triggered widespread landslides in the Himalayan mountains in northern Pakistan and surrounding areas, some of which are active and are still posing a significant risk. Landslides triggered by the 2005 Kashmir earthquake are extensively studied; nevertheless, spatio-temporal landslide susceptibility assessment is lacking. This can be partially attributed to the limited availability of high temporal resolution remote sensing data. We present a semi-automated technique to use the Sentinel-2 MSI data for co-seismic landslide detection, landslide activities monitoring, spatio-temporal change detection, and spatio-temporal susceptibility mapping. Time series Sentinel-2 MSI images for the period of 2016–2021 and ALOS PALSAR DEM are used for semi-automated landslide inventory map development and temporal change analysis. Spectral information combined with topographical, contextual, textural, and morphological characteristics of the landslide in Sentinel-2 images is applied for landslide detection. Subsequently, spatio-temporal landslide susceptibility maps are developed utilizing the weight of evidence statistical modeling with seven causative factors, i.e., elevation, slope, geology, aspect, distance to fault, distance to roads, and distance to streams. The results reveal that landslide occurrence increased from 2016 to 2021 and that the coverage of areas of relatively high susceptibility has increased in the study area.
... Remote sensing (RS) allows CRC to map large areas rapidly (Zheng et al., 2014). At present, multispectral and hyperspectral RS data have been used to map CRC (Yue et al., 2017(Yue et al., , 2019a. For optical RS, the biggest challenge is the capacity to discriminate the crop residues from the bare soil (Daughtry, 2001). ...
Crop residues are effective for the prevention of soil erosion. The crop residue cover (CRC) can be mapped by remote sensing. Different morphologies of crop residue will affect the spectral reflectance, reducing the accuracy of CRC estimation by multispectral data. However, the influence of residue morphology is not fully considered on the accuracy of CRC mapping using satellite images. In addition, the spectral indices are easily saturated and less sensitive to high-density areas of crop residue. This study selected four maize planting sites to obtain hyperspectral reflectance and unmanned aerial vehicle (UAV) images. The effects of CRC and residue morphology on spectral reflectance were analyzed, and a new Residue Adjust Normalized Difference Residue Index (RANDRI) was proposed. UAV images were used to extract ground CRC data for training a CRC prediction model based on Sentinel-2 MSI data. Finally, the piecewise prediction model based on different residue indices was used to map CRC. The study results highlighted a linear relationship between the reflectance intersection of shortwave infrared 2 reflectance and red edge 3 of Sentinel-2 MSI with different residual morphologies, called the residue line. The model accuracy of RANDRI optimized by residual line parameters was better than that of the Normalized Difference Residue Index and Soil Adjust Normalized Difference Residue Index (SANDRI) in the high-density area of crop residue. RANDRI can weaken the influence of residue morphologies on modeling accuracy. The CRC spatial distribution by the piecewise SANDRI+RANDRI model was more consistent with CRC measured than that of the RANDRI models individually. The determination coefficient of the piecewise model was 0.82, and the relative error was 10.66%. The piecewise model can effectively improve the anti-saturation ability of the spectral indices. We suggest a rapid and accurate approach for monitoring the CRC and provide a more suitable CRC mapping strategy for high-density areas of crop residue using multispectral remote sensing data.
... Quemada and Daughtry (2016) used a water index to correct the effect of cropland moisture on NDTI, thereby providing more accurate and reliable f cr estimates. Yue et al. (2019b) used field soil moisture measurements and a Lobell model to correct for the effect of moisture on cropland surface spectra, thereby improving the estimation accuracy of rice f cr for field soil moisture content lower than 20%. Table 2. DFI= dead fuel index. ...
Accurate estimations of soil cover parameters (fractional coverage of crop residue [fcr], crop [fc], and bare soil [fbs]) in croplands can assist in decision-making for agricultural management organisations and are important for understanding the interrelations between global changes and terrestrial ecosystems. In recent decades, optical remote sensing has been widely used for mapping fc, fcr, and fbs. The crop residue and soil spectra decrease rapidly with increasing cropland moisture primarily because of water absorption in the near-infrared and shortwave infrared bands. This leads to a decrease in the denominator of current normalized crop residue spectral indices (shortwave infrared [SWIR] band 1 + band 2), thereby causing them to increase with increasing cropland moisture. This study revealed that the slopes of the soil and crop residue spectra between SWIR1 and SWIR2 were less affected by cropland moisture changes. Based on this feature, we propose a shortwave infrared angle (SWIRA) index and jointly use it with normalized difference vegetation index (NDVI) to estimate and produce multi-temporal maps of fc, fcr, and fbs using broadband Sentinel-2 MSI images. The proposed SWIRA is the slope of the spectrum between SWIR1 and SWIR2. This study (i) uses laboratory-based spectral reflectance to evaluate the performance of the SWIRA and (ii) further evaluates the performance of the SWIRA–NDVI method using multi-temporal Sentinel-2 MSI images. Our results indicate that (i) cropland moisture has a considerably smaller effect on SWIRA than it does on current broadband normalized crop residue spectral indices, and (ii) maps of the estimated fc, fcr, and fbs values can be used to improve crop growth and decision-making, for example, by providing harvest maps and soil tillage intensity maps.
... In the pixel decomposition method, the spectral characteristics of vegetation and soils depend on the water content of the endmember, which affects the calculation accuracy. Yue (Yue et al., 2019) proposed the soil moisture (SM) dynamic detection model, whereby the SM of each pixel is detected, providing a more accurate spectrum. The spectra of soil endmembers can be derived with a SM automatic detection system. ...
... The DDPM was proposed in 2019 to estimate RRC for a simple paddy field scenario (soil-rice residue), achieving good accuracy (Yue et al., 2019). The DDPM method is also based on LSMA, which considers that the pixel consists of two kinds of endmembers and considers the influence of soil water content. ...
... In addition to using the DQPM to estimate the RRC, we also used the DDPM and the SQPM to estimate the RRC. The DDPM is most accurate with bands B1 and B4 (Yue et al., 2019) (Fig. 10). The MRE and RMSE from the DQPM were 0.24 and 0.14, respectively, which are lower than the MRE and RMSE of 0.51 and 0.21 from the DDPM and 0.43 and 0.17 from the SQPM, respectively. ...
Crop residues left on the field after harvest increase soil organic matter content and improve soil quality. Linear spectral mixture analysis (LSMA) is an important technique for calculating crop residue cover. Traditionally, farmland has been considered to be composed solely of soil and crop residue endmembers. But rice paddy fields are often more complex than other fields. In the decomposition of pixels reflectance, leaving out potential endmembers greatly increases the variability of existing endmember reflectance. The error is then transferred to the rice residue endmember. In this paper, a dynamic-quadripartite pixel model (DQPM) is proposed to adapt LSMA to calculate rice residue cover (RRC) in complex paddy fields. This method considers that pixels in paddy fields are composed of four endmembers: soil, rice residues, green moss and white moss. With the approach, soil moisture can be calculated to automatically correct the reflectance of the soil endmembers of each pixel. The calculation results of our model were verified with field data and compared with those from the static-quadripartite pixel model (SQPM) without considering soil moisture (SM) content and with the dynamic-dimidiate pixel model (DDPM). Results confirm the feasibility of DQPM. The results show that with four endmembers, RRC has a large range of improved computational accuracy with DQPM compared with SQPM and DDPM. DDPM has a large error under 1% < SM < 3% (gravimetric water content) and 60% < RRC < 70%. Only when SM is near 0.3, SQPM can achieve good accuracy. Among the three models, DQPM has the best robustness under different soil moisture and RRC scenarios. Therefore, our proposed method is useful for calculating RRC in complex paddy field scenarios.
