Search

Soil organic matter (SOM) is an essential component of soil and is crucial for increasing agricultural production and soil fertility. The combination of hyperspectral remote sensing and deep learning can be used to predict the SOM content efficiently, rapidly, and cost-effectively on various scales. However, determining the optimal groups, inputs, and models for reducing the spatial heterogeneity of soil nutrients in large regions and to improve the accuracy of SOM prediction remains a challenge. Hyperspectral reflectance data from 1477 surface soil samples in Northeast China were utilized to evaluate three grouping methods (no groups (NG), traditional grouping (TG), and spectral grouping (SG)) and four inputs (raw reflectance (RR), continuum removal (CR), fractional-order differentiation (FOD), and spectral characteristic parameters (SCPs)). The SOM prediction accuracies of random forest (RF), convolutional neural network (CNN), and long short-term memory (LSTM) models were assessed. The results were as follows: (1) The highest accuracy was achieved using SG, SCPs, and the LSTM model, with a coefficient of determination (R2) of 0.82 and a root mean squared error (RMSE) of 0.69%. (2) The LSTM model exhibited the highest accuracy in SOM prediction (R2 = 0.82, RMSE = 0.89%), followed by the CNN model (R2 = 0.72, RMSE = 0.85%) and the RF model (R2 = 0.69, RMSE = 0.91%). (3) The SG provided higher SOM prediction accuracy than TG and NG. (4) The SCP-based prediction results were significantly better than those of the other inputs. The R2 of the SCP-based model was 0.27 higher and the RMSE was 0.40% lower than that of the RR-based model with NG. In addition, the LSTM model had higher prediction errors at low (0–2%) and high (8–10%) SOM contents, whereas the error was minimal at intermediate SOM contents (2–8%). The study results provide guidance for selecting grouping methods and approaches to improve the prediction accuracy of the SOM content and reduce the spatial heterogeneity of the SOM content in large regions. [ABSTRACT FROM AUTHOR]

• The effects of spectral similarity and soil classes on the prediction of SOM were considered. • Stratification strategy of finer soil hierarchical can be constructed for SOM prediction. • The input variables describing the spectral shape can be used for high-accuracy SOM prediction. Whether a finer soil classification hierarchical stratification strategy and the spectral characteristic parameters (SCPs) that describe the shape of the spectral curve can be used to improve the prediction accuracy of soil organic matter should be clarified. We measured the visible, near-infrared and shortwave infrared (VIS-NIR-SWIR, 400 – 2500 nm) spectral reflectance of 322 topsoil samples. The spectral reflectance was converted to continuum removal curves, and then, SCPs were extracted based on the curves. According to the results of the Second National Soil Survey of China, the samples were divided into 4 great groups or 8 genus, and a variety of stratification strategies were constructed based on great group (GR-S), genus (GE-S), spectral similarity (SS-S) and decision tree model (DT-S). A local random forest model was established to evaluate the performance of different stratification strategies and input variables. Our results are described as follows: (1) In different stratification strategies, the SOM prediction model based on DT-S exhibits the highest accuracy, followed by the SOM prediction models based on GE-S and SS-S; the SOM prediction model based on GR-S exhibits the lowest accuracy; (2) among the different input variables, the root mean squared error (RMSE) and coefficient of determination (R2) of the best SOM model predicted by SCPs are 5.18 g kg−1 and 0.89, respectively. Compared with the original reflectance based on the nonstratified strategy, the RMSE decreases by 4.88 g kg−1 and R2 increases by 0.32. The study results highlight the advantages of refining the soil hierarchy, which is helpful for identifying the differences in soils at the regional scale and analysing the relationship between stratification results and the characteristics of the soil environment to obtain a highly accurate prediction model. [ABSTRACT FROM AUTHOR]

• Multi-sensor comparison for SOM mapping. • Landsat-8 is currently the best choice for SOM mapping. • Temporal resolution crucial for accurate SOM prediction. • Similar spatial trends in different SOM products. Using remote sensing images to predict the spatial distribution of SOM content is a classic digital soil mapping problem. The purpose of this study is to compare the SOM mapping performance of Landsat 8 OLI, Landsat 7 ETM +, Sentinel-2 MSI and MODIS. We take the dry land in the black soil area of Northeast China as the study area and 791 soil samples were collected in the field to directly measure the SOM content, which serves as the evaluation dataset. We used four images to synthesize soil images for May over multi-year periods. May was identified as the best time of year to image the bare soil in Northeast China. All images used for the median synthesis consisted of cloud masked datasets acquired in May. Then, the image band, spectral index were used as inputs, and the SOM mapping results derived from the different satellite images was evaluated using the random forest regression algorithm. In addition, we use simulation methods to evaluate the impact of temporal, spatial, and spectral differences on SOM mapping. The results show that (1) Landsat-8 is currently the best choice for SOM mapping, Sentinel-2 images have the most potential for future SOM mapping, Landsat-7 and MODIS images can provide support for determining the historical spatial distribution of SOM content; (2) improving the temporal, spatial, and spectral resolution of images can improve the accuracy of SOM prediction, and temporal resolution is the most important because higher temporal resolution makes the selection of image pixels more conducive to accurate SOM prediction, thereby reducing the impact of the field environment; and (3) there is little difference in the spatial distribution trends and average values identified from the SOM products obtained from the four satellites, and can be generalized in some application scenarios. [ABSTRACT FROM AUTHOR]

[Display omitted] • The accuracy of soil organic matter (SOM) mapping using images of different time periods varies greatly. • May is the best time for SOM mapping in Songnen Plain. • The performance of SOM mapping using median synthetic images are the best. • More years of synthetic images of bare soil period can obtain more stable bare soil pixels. • The R2 of SOM prediction using Landsat 8 median synthetic image in the best time window is 0.705. Accurate assessment of the spatial distribution of soil organic matter (SOM) is of great significance for regional sustainable development, especially in fertile black soil areas. The present study proposed a regional-scale high spatial resolution (30 m) SOM mapping method based on multitemporal synthetic images. The study area is located on the Songnen Plain of Northeast China. First, all available Landsat 8 surface reflectance (SR) data during the bare soil period (April and May) from 2014 to 2019 in the study area were screened in the Google Earth Engine (GEE), and the cloud mask was constructed. The median, average, maximum, and minimum values of the image set were synthesized according to single-year multimonth, multiyear single-month and multiyear multimonth time ranges, and the spectral index of the synthesized image was constructed. Second, the bands and spectral indices of different synthetic images were used as input to establish a random forest (RF) model of SOM prediction, and the accuracies of different spatial prediction models of SOM were compared to evaluate the optimal regional remote sensing prediction model of SOM. The following results were show. 1) The use of the spectral index combined with the image band as input had a greater improvement in the accuracy of SOM prediction than the use of only the image band. 2) Compared to the average, maximum and minimum synthesized images, the median synthesized image had higher accuracy in SOM prediction. 3) More years of synthesized images provided more robust SOM prediction results. 4) May was the best time window for SOM mapping on the Songnen Plain. This study presents a large-scale and high spatial resolution SOM mapping method that is suitable for black soil areas in Northeast China and extends the application of GEE in digital soil mapping. [ABSTRACT FROM AUTHOR]

Whether spectral feature parameters (SFPs) can be effectively used to predict soil organic matter (SOM), and its spatial transferability was tested for different regions. In this study, a hybrid model was proposed based on SFPs and local regression method. The topsoil spectra of 221 soil samples from Sanjiang Plain and 187 soil samples from Nong'an county in Northeast China were re-sampled (at 0.01 µm intervals) and converted to the first-derivative curves and CR curves. Two SFPs were extracted on reflectance curves (RSFPs), which were defined as curve length (L c) and area (A c) between two absorption positions. Local random forest models with k-means clustering were built using the RSFPs and first-derivative spectra for evaluating the feasibility of RSFPs in these two study areas. After analyzing the results of Chinese soil samples, this method was also applied to Land Use/Land Cover Area Frame Survey (LUCAS) topsoil database in order to evaluate the feasibility of RSFPs outside China. Our results revealed these: (1) high correlation between the RSFPs of soil samples with spectral characteristics of sandy soils and SOM; (2) the importance of L c relative to A c in SOM prediction is higher; and (3) SOM prediction using RSFPs is comparable to the first-derivative spectra, and the best result has an R2 of 0.76 and an RMSE of 7.43 g kg−1. Our results suggest that RSFPs can be used to predict SOM and have good spatial transferability after applying in two study areas. RSFPs with specific geometric meaning have high potential to predict other soil properties in Northeast China. Our results also provide a reference for SOM mapping using hyper-spectral satellites. • Visible spectra are useful for predicting soil organic matter. • Local regression after k-means clustering was used. • Extract new spectral parameters (L c and A c) to predict soil organic matter. • Evaluate the spatial transferability of the new parameters. [ABSTRACT FROM AUTHOR]