| Soil is an important part of nature,and the spatiotemporal soil properties(such as nutrients,heavy metals,and physical properties)have an important impact on agricultural production and the ecological environment.Thus,understanding the spatial–temporal distribution,characteristics,and influence mechanism of soil properties is of great significance to the improvement of crop quality and yield along with agricultural sustainable development.In the current related research,the analysis of the spatial and temporal characteristics of soil organic matter(SOM)is mainly based on the spatial distribution data of regional SOM at the first and last periods,while the analysis of the continuous spatio-temporal variation process has been rarely studied.Moreover,the spatial distribution data of the first and last periods cannot fully elucidate the spatio-temporal distribution characteristics of regional SOM.In addition,the existing studies are commonly limited in terms of the spatiotemporal prediction of the future distribution of soil properties.Thus,the present study took Zigui County as the research area and the spatio-temporal soil sample points collected from 2006 to 2018 as the main data.The spatio-temporal distribution of regional SOM was obtained using the spatio-temporal geostatistical method.Then,the spatio-temporal differentiation characteristics and their influencing factors were revealed through various spatiotemporal analysis methods.Finally,the spatiotemporal distribution of SOM in the next five years was predicted.This study has the following contributions:(1)The spatiotemporal Moran index is proposed,and the spatiotemporal autocorrelation of SOM at different spatiotemporal distance thresholds was calculated.The results show that the spatiotemporal Moran indices at all spatiotemporal distance thresholds are greater than 0 and significant within the spatiotemporal range of 5 km and 10 years,indicating the significant positive spatiotemporal autocorrelation of SOM in Zigui County from 2006 to 2018.In addition to a new hybrid theoretical variogram model that integrates an exponential model at the spatial scale and a hole model at the temporal scale,the spatiotemporal ordinary kriging(STOK)method was employed to determine the spatiotemporal SOM distribution in Zigui County from 2006 to 2018.The results indicate the following.(1)The fitted hybrid theoretical variogram model could suitably represent the characteristics of the ST experimental variogram.(2)A series of numerical comparisons of STOK versus spatial ordinary kriging(SOK)indicates that the STOK method performs better than the SOK method on few soil samples because the STOK results exhibit a higher estimation accuracy and a more stable estimation variance.Moreover,the spatial smoothness and variability of the STOK results are nearer those of the soil samples than those of the SOK results.Then,a temporally continuous and spatially staggered ST soil sampling proposal was formulated for the continuous long-term monitoring of soil properties.(2)On the basis of the spatiotemporal distributions data of SOM obtained by the STOK method,the spatiotemporal distribution and change characteristics of SOM in Zigui County from 2006 to 2018 were revealed by using various spatiotemporal analysis methods.The main results are as follows.The overall mean value of SOM in Zigui County is 21.3 g/kg,and the spatial distribution trends of the multi-year mean values are high in south and low in the north,and the multi-year mean value of SOM is negatively correlated with its temporal fluctuation.Several random indicator functions were defined to evaluate the uncertainty characteristics of the spatio-temporal distribution of SOM.The results show that the SOM contents in the spatio-temporal domain are mostly between 15 and 34 g/kg,and the probability of exceeding the threshold at two spatio-temporal locations at different distances decreases sharply with the increase of the threshold.Meanwhile,the temporal change of SOM in Zigui County tended to be coordinated in the entire county,rather than only in some local areas;From2006 to 2018,the average content of SOM in Zigui County experienced a process of first decreasing,then increasing and then decreasing with 2011 and 2015 be the key inflection points.The trend test results based on the Mann–Kendall method showed that from 2006 to 2018,in Zigui County,only 15.05% and 32.05% of the regional SOM showed a significant downward trend and a significant upward trend,respectively,There is no obvious trend in other regions.However,from 2006 to 2018,the area where the concentration of SOM decreased and increased significantly accounted for 37.2%and 44.57% of the entire area of Zigui county.(3)According to the spatiotemporal distribution characteristics of SOM and taking topographic factors,soil type,land cover,vegetation cover,and township division as independent variables,the influencing factors and their contribution to the spatiotemporal variation of SOM in Zigui County were determined.The results show that township division,soil type,elevation,the NDVI,and land cover have significant effects on the spatial differentiation and spatiotemporal change of SOM in Zigui County.All the factors could explain 61.1% of the spatial differentiation of SOM’s multi-year mean and 34.6% of the spatiotemporal variation of SOM from 2006 to 2018.Furthermore,township division,soil type,and elevation the three factors with the strongest interpretation ability.Therefore,they are the dominant factors leading the spatial differentiation and spatiotemporal change of SOM in Zigui County.The uncertainty of the dependent variables and the lack of factor data in small scales are the main reasons for the low interpretation ability of the spatiotemporal change of SOM.Finally,using the spatiotemporal regression Kriging method and comprehensively considering the spatiotemporal variogram of SOM and the influence mechanism of various factors,three scenario models,namely,yield increase,flat harvest,and yield reduction,were set to predict the spatiotemporal distribution of SOM in Zigui County from 2019 to 2023.The results show that under the yield increase and yield reduction scenarios,the average annual change rates of SOM in the entire study area from 2019 to 2023 are 1.16% and-1.47%,respectively,while the temporal change of SOM is small under the flat harvest scenario.The main innovations of this study are in the following three aspects.The first is the spatiotemporal modeling and interpolation of regional SOM through spatiotemporal geostatistics.The application of the spatiotemporal geostatistics method and the innovation along with elaboration of a series of technical details,such as the spatiotemporal Moran index,the construction of a spatiotemporal theoretical variogram model,the fitting of the spatiotemporal theoretical variogram model based on the genetic algorithm,and the method of determining spatiotemporal adjacent points,provide an application paradigm for the spatiotemporal modeling and interpolation of geographical attributes.Second,a series of spatiotemporal analysis methods were applied on the spatiotemporal data of SOM to reveal their spatiotemporal distribution characteristics,including the trend of SOM based on the Mann–Kendall method;the evaluation of the significant increase or decrease of the SOM content,the estimated value,and the estimated variance;the extraction of the spatiotemporal uncertainty characteristics of SOM on the basis of spatiotemporal random indicator functions;and the use of a geographic detector,correlation analysis,analysis of variance,and regression analysis to determine the influencing factors of the spatiotemporal variation of SOM.Finally,the spatiotemporal regression Kriging method was used to predict the future spatiotemporal distribution of SOM according to the influence mechanism of SOM’s spatiotemporal change and the different scenarios in the study area.The above innovations can provide a reference for the spatiotemporal modeling and analysis of geographical attributes. |
PDF Full Text Request