CN115630567A - Coastal zone soil organic carbon reserve simulation and prediction method - Google Patents
Coastal zone soil organic carbon reserve simulation and prediction method Download PDFInfo
- Publication number
- CN115630567A CN115630567A CN202211202453.3A CN202211202453A CN115630567A CN 115630567 A CN115630567 A CN 115630567A CN 202211202453 A CN202211202453 A CN 202211202453A CN 115630567 A CN115630567 A CN 115630567A
- Authority
- CN
- China
- Prior art keywords
- soil
- model
- organic carbon
- coastal zone
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Geometry (AREA)
- Development Economics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Medical Informatics (AREA)
Abstract
The invention discloses a coastal zone soil organic carbon reserve simulation and prediction method, which carries out coastal zone soil organic carbon reserve simulation and prediction by constructing a coastal zone soil organic carbon reserve estimation and prediction enhancement regression tree model. According to the coastal zone soil organic carbon reserve simulation and prediction method, data such as high spatial resolution, open-source Sentinel optics and radar remote sensing are used as input data, simulation monitoring of 10m spatial resolution of the coastal wetland in a large regional scale is achieved, the average correlation coefficient of ten layers of cross validation of the model reaches 0.8, and the accuracy is remarkably superior to that of soil carbon reserve estimation based on a conventional survey sampling point according to an ecological system type statistical method and a spatial interpolation method. The method realizes the high spatial resolution, high precision, low cost and rapid simulation and prediction of the soil carbon reserves in the coastal zone which is difficult to sample and investigate on the spot and has broken space.
Description
Technical Field
The invention relates to the technical field of multi-source satellite remote sensing ecosystem monitoring, in particular to a coastal zone soil organic carbon reserve simulation and prediction method.
Background
The carbon stored in the soil far exceeds that in the atmosphere and organisms. More than half of the carbon in soil carbon is sequestered as organic carbon, 2 times that in the atmospheric carbon pool, and soil organic carbon is more actively involved in the carbon cycle than soil inorganic carbon (Batjes, 1996 crowther et al, 2016). Thus, soil organic carbon reserves dynamics have a significant impact on global and regional climate.
Compared with the land ecosystems such as forests, grasslands, farmlands and the like, the coastal zone ecosystem plays an important role in regional carbon circulation due to higher carbon burying rate and absorption rate, and especially has an important role in realizing the strategic target of 'double carbon' as a remarkable carbon sink ecosystem. However, due to the broken spatial distribution, poor accessibility, large spatial heterogeneity, etc., there is still a great uncertainty in estimating and predicting the carbon reserves of coastal zone ecosystems, especially in accurately estimating and predicting the organic carbon reserves of soil.
The existing estimation of soil organic carbon reserves comprises: based on an investigation sampling analysis method, an empirical statistical model or machine learning model simulation is established through sampling point test analysis data, and a process model simulation method is used. The method is limited to small-scale soil organic carbon reserve assessment and analysis based on the survey sampling method. The method for estimating the soil organic carbon reserves by establishing an empirical model or a machine learning model depends on limited sampling point observation data and single, medium and low resolution satellite remote sensing data, and the precision of the simulated monitoring of the soil carbon reserves is often greatly uncertain. The process model-based simulation of the organic carbon reserves of the soil is limited by the complex model, numerous parameters and complex optimization, has high uncertainty in regional soil organic carbon reserve estimation, and is complex and inconvenient to apply.
The satellite remote sensing earth observation technology is widely applied to the monitoring of the parameters of the geographical and ecological systems, such as earth surface and land cover, vegetation growth, surface soil moisture, soil temperature and the like, which can be directly monitored or indirectly inverted, but the satellite remote sensing earth observation technology is difficult to directly monitor or indirectly invert the soil information related to a certain depth, such as soil carbon reserve and the like. The traditional method for simulating the soil carbon reserves by sampling the test data through the sampling points and combining the medium-low resolution remote sensing data is not suitable for monitoring the soil carbon reserves of the coastal zone ecosystem. On one hand, the cost of manpower and material resources for large-scale soil carbon reserve survey is high, and the time period is long; on the other hand, the pixel of the satellite remote sensing data with the medium-low resolution ratio is not enough to distinguish small broken coastal wetlands with the size smaller than the resolution ratio, so that the simulation monitoring result of the carbon reserve of the ecosystem of the coastal zone at present has great uncertainty.
Although there are many methods for estimating the organic carbon reserves of soil, there are few effective methods for predicting the organic carbon reserves of the coastal soil under the background of future climate change. The coastal zone ecosystem is simultaneously influenced by human activities, climate changes and sea level rise caused by the human activities and the climate changes, the coastal zone ecosystem under the future climate change situation is predicted to need to simultaneously consider the influences of the climate changes, the human activities and the sea level rise, and reasonable and reliable model input variable information under the future climate change situation needs to be obtained. In summary, there are still significant challenges to the prediction of coastal zone soil organic carbon reserves in future climatic change scenarios.
Disclosure of Invention
The invention aims to solve the problems that the traditional soil organic carbon reserve estimation method by combining sampling point sampling test data with medium-low resolution remote sensing data is not suitable for high-space-time resolution monitoring of soil carbon reserve of a coastal zone ecosystem and cannot realize prediction of the soil organic carbon reserve of the coastal zone under the future climate change scene. The invention aims to provide a method for realizing high-spatial-resolution estimation of organic carbon reserves in coastal zones by combining high-spatial-resolution satellite remote sensing data acquired by current open sources with information such as weather, soil attributes, landforms and the like.
The invention also aims to provide a prediction method based on the coastal zone soil organic carbon reserve simulation method.
The technical scheme adopted for realizing the purpose of the invention is as follows:
a coastal zone soil organic carbon reserve simulation method comprises the following steps:
step 1: selecting model variables required by model construction
The model variables include: a coastal vegetation index, a normalized water body index, a chlorophyll fluorescence index, an enhanced vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an improved vegetation index, a normalized salinity index, a radar remote sensing polarization index, an elevation, a terrain humidity index, an annual average temperature, an annual precipitation, a downward solar shortwave radiation, a population, a distance to a river, a soil salinity, and a reference soil carbon reserve;
and 2, step: model variable acquisition and processing
Firstly, acquiring model variable data through a Google Earth Engine geographic science big data cloud platform and an open source basic geographic data product; then, carrying out format conversion and grid matching processing on the acquired model variable data by using GIS software;
and step 3: model training sample data set preparation
Extracting a model input variable value corresponding to the soil carbon storage sample point by using a GIS software extraction grid attribute to point tool, and acquiring a model training sample data set required by constructing a soil carbon storage model;
and 4, step 4: model variable optimization
Selecting an optimal model variable from the model variables in the step 1 by a correlation coefficient matrix analysis and expansion factor inspection method, and acquiring optimal model variable data according to the method in the step 2;
the optimal model variables include: a coastal zone vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an elevation, a distance to a river, an annual precipitation, an annual average temperature, population, soil salinity, and a reference soil carbon reserve;
and 5: construction of reinforced regression tree model for estimating and predicting organic carbon reserves of coastal zone soil
Constructing a coastal zone soil organic carbon reserve estimation and prediction enhancement regression tree model by utilizing the coastal zone spatial distribution data and the soil organic carbon sampling point data and combining the model training sample data set obtained in the step (3);
and 6: coastal zone soil organic carbon reserve simulation
Taking the optimal model variable data obtained in the step 2 as model input, and realizing the analog estimation of the organic carbon reserves of the coastal zone soil with the spatial resolution of 10m through the estimation and prediction enhancement regression tree model of the organic carbon reserves of the coastal zone soil constructed in the step 5;
and 7: statistical analysis of organic carbon reserves of coastal zone soil
And (4) carrying out statistical analysis on the total soil organic carbon storage and the average organic carbon density of the coastal zone of the area by using the spatial distribution data of the soil organic carbon storage of the coastal zone obtained in the step (6).
In another aspect of the present invention, a method for predicting organic carbon reserves in coastal zones comprises the following steps:
step 1: selecting model variables required by model construction
The model variables include: a coastal vegetation index, a normalized water body index, a chlorophyll fluorescence index, an enhanced vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an improved vegetation index, a normalized salinity index, a radar remote sensing polarization index, an elevation, a terrain humidity index, an annual average temperature, an annual precipitation, a downward solar shortwave radiation, a population, a distance to a river, a soil salinity, and a reference soil carbon reserve;
and 2, step: model variable acquisition and processing
Firstly, acquiring model variable data through a Google Earth Engine geographic science big data cloud platform and an open source basic geographic data product; then, carrying out format conversion and grid matching processing on the acquired model variable data by using GIS software;
and step 3: model training sample data set preparation
Extracting a model input variable value corresponding to the soil carbon reserve sample point by using a GIS software extraction grid attribute to point tool, and acquiring a model training sample data set required by constructing a soil carbon reserve model;
and 4, step 4: model variable optimization
Screening optimal model variables from the model variables in the step 1 by a correlation coefficient matrix analysis and expansion factor inspection method, and acquiring optimal model variable data according to the method in the step 2;
the optimal model variables include: a coastal zone vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an elevation, a distance to a river, an annual precipitation, an annual average temperature, population, soil salinity, and a reference soil carbon reserve;
and 5: construction of reinforced regression tree model for estimating and predicting soil organic carbon reserves of coastal zone
Building a coastal zone soil organic carbon reserve estimation and prediction enhancement regression tree model by utilizing the coastal zone space distribution data and the soil organic carbon sampling point data and combining the model training sample data set obtained in the step 3;
step 6: relative importance analysis and screening of optimal model variables
Determining the relative importance of each optimal model variable screened in the step 4 to the simulation and prediction of the organic carbon in the coastal zone soil through the variable importance analysis and the partial dependence graph characteristic analysis of the constructed coastal zone soil organic carbon reserve enhancement regression tree model, and dividing the optimal model variable into an important variable, a non-important variable and a stable variable;
and 7: obtaining and processing optimal model variable data
The stable variable adopts the existing data;
adopting the years of average data of the prior satellite remote sensing for the non-important variables except the stable variable;
MAT and MAP data in 2061-2080 years and 2081-2100 years under two situations of SSP245 and SSP585 predicted by adopting CMIP6 multi-model;
the population data is obtained by a Chinese population data set of a corresponding scene and a time period estimated by the model;
soil salinity data is obtained by simulating a multisource linear regression model established by adopting an AIC method through collected soil sample organic carbon sample point data and multisource data;
and step 8: coastal zone spatial distribution data acquisition under future climate change scene
And calculating the average sea level rising rate calculated based on the actual sea level rising observation data, and taking the seawater submergence and land migration into consideration on the basis of the existing coastal zone spatial distribution data to obtain the coastal zone spatial distribution data under the future climate change situation.
And step 9: prediction of soil organic carbon reserves of coastal zone ecosystem under future climate change scene
And (4) taking the optimal model variable data obtained in the step (7) as the input of the coastal zone soil organic carbon storage estimation and prediction enhanced regression tree model, and predicting the coastal zone soil organic carbon storage data under different temperature-rise scenes by combining the coastal zone spatial distribution data under the future climate change scene obtained in the step (8).
In the technical scheme, in the step 4, when the correlation coefficient of two variables in the correlation coefficient matrix is greater than or equal to 0.8, the variables with lower correlation coefficient with the soil organic carbon reserve in the correlation coefficient matrix are removed; and eliminating variables with VIF values larger than 0.4 in all variables.
In the technical scheme, in the step 5, the construction and simulation of the coastal zone soil organic carbon reserve simulation and prediction enhancement regression tree model are realized by using a stem function of a language R dismo package gbm.
In the technical scheme, model parameter optimization is further included;
the model parameter optimization method is characterized in that a gradual simulation test method is adopted, and when three key parameters of learning rate, tree complexity and bagging ratio are set to enable the model to be optimal, the optimal regression tree model for estimating and predicting the organic carbon storage of the coastal zone soil is determined.
In the technical scheme, the constructed and optimized coastal zone soil organic carbon reserve estimation and prediction enhanced regression tree model is verified and evaluated through ten-layer cross validation (10 fold cv).
In the above technical solution, in step 6, the relative importance of the 11 optimal model variables is as follows:
annual average temperature (19.8%), soil salinity (19.7%), distance from river (13.7%), population (13.4%), coastal vegetation index (7.6%), normalized vegetation moisture index (6.1%), elevation (5.6%), reference soil carbon reserve (4.7%), normalized vegetation index (3.6%), annual precipitation (2.9%), and improved soil vegetation index (2.8).
In the above technical solution, the dividing principle of the important variable, the non-important variable and the stable variable is as follows:
defining variables with relative importance greater than 10% as important variables; defining variables with relative importance less than 10% as non-important variables; the time-varying insignificant variable is defined as the stable variable.
In the technical scheme, the important variables comprise annual average temperature, soil salinity, distance from a river and population;
the non-important variables comprise a coastal zone vegetation index, a normalized vegetation humidity index, an elevation, a reference soil carbon reserve, a normalized vegetation index, an annual precipitation and an improved soil vegetation index;
the stability variables include distance from river, elevation, and reference soil carbon reserves.
In the above technical solution, in step 7, the multi-source linear regression model is represented by the following formula:
Salinity＝-0.0066×Elevation+8.3523×Distance-0.0173×MAP+0.3989×MAT-0.0017×Population+7.3230。
compared with the prior art, the invention has the beneficial effects that:
1. according to the coastal zone soil organic carbon reserve simulation method provided by the invention, data such as high spatial resolution, open-source Sentinel optics, radar remote sensing and the like are used as input data, simulation monitoring of 10m spatial resolution of the coastal wetland in a large regional scale is realized, the cross validation average correlation coefficient of ten layers of the model reaches 0.8, and the precision is remarkably superior to that of soil carbon reserve estimation based on a survey sampling point according to an ecological system type statistical method and a spatial interpolation method. The method realizes the high spatial resolution, high precision, low cost and rapid simulation and prediction of the soil carbon reserves in the coastal zone with difficult space crushing and on-site sampling investigation.
2. The invention provides a coastal zone soil organic carbon reserve prediction method, which aims at the problems of large calculated amount, complexity and uncertainty of a coastal zone ecological system soil carbon reserve prediction process model under different future climatic change scenes, simultaneously considers climate warming, sea level rising and human activity influence, firstly identifies main environmental influence factors of the coastal zone ecological system soil carbon reserve, constructs a soil salinity model which is based on weather, terrain, population and the like and is easy to obtain environmental and human activity element variables for simulation prediction to obtain the soil under the future climatic change scene required by the soil organic carbon reserve prediction model, and further constructs a data driving model to realize the high-precision, high-altitude spatial resolution (10 m) and high-efficiency prediction simulation of the coastal zone ecological system soil organic carbon reserve under different scenes in the future by combining with weather data and future population open source data of CMIP6 future scenes.
Drawings
FIG. 1 is a flow chart of a method for simulating the organic carbon reserves in coastal zones;
FIG. 2 is a graph showing the relationship between the soil organic carbon reserves and the independent variables of the prediction model of the soil organic carbon reserves in the coastal zone;
FIG. 3 is a diagram illustrating environmental element variable spatial distribution data;
FIG. 4 shows the results of simulation of the organic carbon reserves in the coastal zones of Tianjin;
FIG. 5 is a graph showing dependence of BRT model simulation on organic carbon reserves in coastal zones;
FIG. 6 shows the result of predicting the organic carbon reserves in the coastal zones of Tianjin;
wherein, the soil organic carbon storage capacity of 2041-2060 years under the condition of SSP245, (b) the soil organic carbon storage capacity of 2081-2100 years under the condition of SSP245, (c) the soil organic carbon storage capacity of 2041-2060 years under the condition of SSP585, and (d) the soil organic carbon storage capacity of 2081-2100 years under the condition of SSP 585.
Detailed Description
The present invention will be described in further detail with reference to specific examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1
A coastal zone soil organic carbon reserve simulation method comprises the following steps:
step 1: selecting model variables required by model construction
According to literature research and the characteristics of organic carbon in soil in coastal zones, initially selecting 19 variables which can be used for simulating the organic carbon reserves of the soil and comprise currently available open sources, high-spatial-resolution optical satellite remote sensing (Sentinel-2) and microwave (radar) satellite remote sensing (Sentinel-1), weather, soil attributes, landforms, population and other factors;
specifically, the preliminarily determined model variables include satellite remote sensing observation data, terrain data, soil attribute data, meteorological data and Population data (Population), and 19 environmental element variables in total. Wherein the optical satellite remote sensing variables comprise a seashore vegetation index (BNDVI), a normalized vegetation index (NDVI), a normalized water body index (NDWI), a chlorophyll fluorescence index (LCI), an Enhanced Vegetation Index (EVI), a normalized vegetation humidity index (NDMI), an improved soil vegetation index (SAVI) and an improved vegetation index (kNDVI) based on Sentinel-2 optical satellite remote sensing observation; the radar remote sensing data comprise polarization indexes (POL) based on Sentinel-1 radar satellite remote sensing; the topographic data comprises elevation (DEM) and topographic moisture index (TWI) and Distance from river (Distance); the soil property data includes normalized Salinity index (NDSI), soil Salinity (Salinity), and reference soil carbon reserves (SoilGrids); the meteorological data includes annual temperature average (MAT) and annual precipitation (MAP) and solar down short wave radiation (SRAD).
Step 2: model variable acquisition and processing
In the model variables, the satellite remote sensing data are acquired through a Google Earth Engine (GEE) geoscience big data cloud platform, and meteorological data, soil attribute data, landform data and population data are acquired through open-source basic geographic data products;
and performing preprocessing such as data calibration, grid size matching and the like on all acquired model variable data by using GIS software.
And step 3: model training sample dataset preparation
And extracting model input variable values corresponding to the soil carbon storage sample points by using a GIS software extraction grid attribute to point tool, and acquiring a model training sample data set required for constructing the soil carbon storage model.
And 4, step 4: model variable optimization
And (3) selecting the optimal model variable with the minimum multiple collinearity from the model variables in the step (2) through correlation coefficient matrix analysis and an expansion factor checking method.
Specifically, the screening principle is that when the correlation coefficient of two variables in the correlation coefficient matrix is greater than or equal to 0.8, the variables with lower correlation coefficient with the soil organic carbon reserve in the correlation coefficient matrix are removed; and eliminating variables with VIF values larger than 0.4 in all variables.
The screened optimal model variables comprise BNDVI, NDMI, NDVI, SAVI, DEM, distance, MAP, MAT, position, soil linearity and SoilGrids, and the total number of 11 optimal model variables.
And 5: construction, optimization and evaluation of coastal zone soil organic carbon reserve simulation and prediction enhanced regression tree model
Constructing a reinforced regression tree model (BRT model) for estimating and predicting the organic carbon reserves of the coastal zone soil based on the R language dismo package gbm.step function by using the model training sample data set prepared in the step 3;
the coastal zone spatial distribution data is selected from various conventional coastal zone data products with high spatial resolution, such as Sun et al, (2020), national Earth systems science data center (http:// www.geodata. Cn /); soil organic carbon sampling point data are obtained by literature acquisition and field sampling analysis.
The parameter optimization of the enhanced regression tree (BRT) model adopts a step-by-step simulation test method, and when three key parameters, namely the learning rate (lr), the tree complexity (tc) and the bagging ratio (bg), are optimal, the optimal BRT model is determined. And setting the model error distribution as Poisson distribution, and verifying and evaluating the constructed and optimized optimal BRT model through ten-layer cross verification (10-fold cv).
Step 6: coastal zone soil organic carbon reserve simulation
And (3) taking the 11 optimal model variable data obtained in the step (2) as input, and simulating and obtaining the 10m spatial resolution coastal zone soil organic carbon storage space distribution data through the coastal zone soil organic carbon storage estimation and prediction enhanced regression tree model constructed in the step (5).
And 7: statistical analysis of organic carbon reserves in coastal zone soil
And (5) carrying out statistical analysis on the total soil organic carbon storage and the average organic carbon density of the coastal zone of the region by using the spatial distribution data of the soil organic carbon storage of the coastal zone obtained in the step (6).
Example 2
The embodiment describes a method for predicting the organic carbon storage of the coastal soil under the future climate change situation by applying the model based on the enhanced regression tree model for estimating and predicting the organic carbon storage of the coastal soil constructed in the embodiment 1.
A high spatial resolution coastal zone soil organic carbon reserve prediction method comprises the following steps:
Step 6: relative importance analysis and screening of optimal model variables
Determining the relative importance of the optimal model variables screened in the step 3 to the simulation of the organic carbon reserves of the coastal zone soil through BRT model variable important analysis and partial dependence graph (PDPs) characteristic analysis, and dividing the optimal model variables into important variables, non-important variables and stable variables;
specifically, the relative importance of the 11 optimal model variables was, in order, 19.8% (MAT), salinity (19.7%), distance (13.7), population (13.4%), BNDVI (7.6%), NDMI (6.1%), DEM (5.6%), soilGrids (4.7%), NDVI (3.6%), MAP (2.9%), and SAVI (2.8).
Defining variables with relative importance greater than 10% as important variables; defining variables with relative importance less than 10% as non-important variables; defining a time-invariant small variable as a stable variable;
according to the above definition, the important variables include MAT, salinity, distance and position;
the non-essential variables include BNDVI, NDMI, DEM, soilGrids, NDVI, MAP and SAVI;
the stability variables include Distance, DEM, and SoilGrids.
And 7: obtaining and processing optimal model variable data
The stable variable adopts the existing data; adopting the years of average data of the prior satellite remote sensing for the non-important variables except the stable variable;
meteorological data (MAT) and Population data (Population) MAT and MAP data in 2061-2080 and 2081-2100 years under two situations of SSP245 and SSP585 estimated by adopting CMIP6 multi-model and estimated corresponding situations of (http:// www.nmic.cn /) and time slot Chinese Population data (https:// doi.org/10.7927/H4JW8BX 5);
soil Salinity data (Salinity) by combining collected soil sample organic carbon sample point data with model variable data preliminarily determined in the step 1 and adopting an AIC method (Akaike, 1981), the following optimal multisource linear regression model is constructed to simulate and obtain the soil Salinity data under the situations of SSP245 and SSP 585:
Salinity＝-0.0066×Elevation+8.3523×Distance-0.0173×MAP+0.3989×MAT-0.0017×Population+7.3230 (1)
and 8: coastal zone spatial distribution data acquisition under future climate change situation
According to the actual conditions of the simulated and predicted regions, the global coastal zone potential spatial distribution data under the future climate warming and sea level rising scenes developed by Schuerch et al. (2018) are adopted, or the possible coastal zone spatial distribution data under the future warming scenes are obtained by considering seawater submergence and land migration calculation on the basis of the existing coastal zone spatial distribution data according to the average sea level rising speed calculated on the basis of the actual sea level rising observation data.
And step 9: prediction of soil organic carbon reserves of coastal zone ecosystem under future climate change scene
And (4) taking the optimal model variable data obtained in the step (7) as input to serve as BRT model input, and predicting the organic carbon storage data of the coastal zone soil under different temperature rising scenes by combining the coastal zone spatial distribution data under the future climate change scene obtained in the step (8). Such as soil organic carbon reserve data of coastal zones in 2061-2080 and 2080-2100 in SSP245 and SSP585 scenes, and statistically analyzing relative history and current soil organic carbon spatiotemporal change.
Example 3
In this embodiment, the simulation of the soil organic carbon reserves in the ecosystem (offshore administrative district) of the coastal zone in Tianjin city is taken as an example, and the simulation and prediction of the soil organic carbon reserves in the ecosystem of the coastal zone in Tianjin city are performed according to the simulation method or the prediction method in embodiments 1 and 2.
Data acquisition and processing of optimal satellite remote sensing and environment element variables
Through literature inquiry analysis and biogeophysical and biogeochemical property analysis of the coastal zone soil organic carbon, 19 initial variables of the environmental elements, namely the model variables, which are easy to obtain, closely related to the coastal zone soil organic carbon reserves and used for simulating and predicting the coastal zone soil organic carbon reserves, are preliminarily determined, and the definition and obtaining sources of the model variables are shown in the following table:
combining literature data, actually surveying the organic carbon storage sample point data of the coastal zone soil acquired by sample collection, test analysis, and constructing a model training data sample point data set; based on a correlation coefficient matrix and a swelling factor test (VIF) method, the optimal environment element variables of the model training are screened, namely, the correlation coefficient matrix is firstly calculated according to the preliminarily determined 19 environment element variables, and for two variables with correlation coefficients larger than 0.8, the variable with the smaller correlation coefficient with the soil organic carbon storage is removed, and on the basis, the VIF test is further carried out, and the variable with the VIF larger than 4 is removed. Finally, 11 variables which are most important for predicting the organic carbon reserves of the coastal zone Soil are screened out, namely, optimal model variables, BNDVI, NDMI, NDVI, SAVI, elevation (DEM), distance, MAP, MAT, position, soil linearity and SoilGrids.
Further analysis showed that there was a significant correlation (p < 0.05) between the soil organic carbon reserves (SCT) of the shore zone ecosystem and the remaining 10 variables screened, except Distance (p = 0.057) (fig. 2). And acquiring the selected 11 model variable data based on the GEE platform and GIS software, and performing preprocessing such as data calibration, grid size matching and the like. FIG. 3 shows the processed Tianjin coast zone area 10m spatial resolution variable spatial distribution data.
According to the method for simulating the organic carbon reserves of the coastal zone soil in the embodiment 1, an enhanced regression tree model for estimating and predicting the organic carbon reserves of the coastal zone soil is constructed.
And (3) performing multiple groups of model experiments by adopting a step-by-step simulation test (parameters are changed step by step according to certain gradient intervals), determining the learning rate (lr) of the BRT model of the organic carbon in the soil in the coastal zone of Tianjin city, wherein the optimal values of the parameters of tree complexity (tc) and bagging ratio (bg) are 0.005,0.5 and 5, and determining the SCT simulation prediction optimal BRT model. The simulation result of the ten-layer cross validation evaluation model shows that the correlation coefficient between the observation of the BRT model simulation SCT sampling points reaches 0.78, and the SCT simulation precision requirement is met.
(III) simulating organic carbon reserves in coastal zone soil
And (2) taking the 11 optimal model variable spatial distribution data obtained in the step (1) as model input, and simulating to obtain soil organic carbon reserve data (shown in figure 4) of the 10m spatial resolution coastal zone ecological system in the current Tianjin city, wherein the coastal zone spatial distribution data adopts 10m spatial resolution coastal zone data of the Bohai-sea developed by Sun et al. (2020). Based on the simulation results, the organic carbon reserves and the average carbon densities of the soil in the coastal zone of Tianjin are respectively 8.3Tg and 68.8Mg ha -1 。
Prediction of organic carbon reserves In (IV) coastal zone soil
And determining the relative importance of the model input independent variable to the soil organic carbon storage quantity simulation through BRT model variable importance analysis and partial dependence graph (PDPs) characteristic analysis. The relative importance of 11 optimal model variables in soil organic carbon reserves simulation of ecosystem in coastal zone in Tianjin is shown in FIG. 5, and the relative importance is 19.8% (MAT), salinity (19.7%), distance (13.7), population (13.4%), BNDVI (7.6%), NDMI (6.1%), DEM (5.6%), soilGrids (4.7%), NDVI (3.6%), MAP (2.9%) and SAVI (2.8) in sequence. The relative importance of the annual average temperature, soil salinity, distance from the river and population to soil organic carbon reserve simulation were 19.8%,19.7%,13.7%, and 13.4%, respectively, which were significantly higher than other environmental element variables. The relative importance of BNDVI, NDMI, elevation (DEM), soilGrids, NDVI, MAP and SAVI is less than 10%. Since future meteorological data and population data can be obtained through CMIP6 and various existing future population data products, the key point for predicting the soil organic carbon storage is how to obtain soil salinity data under the future climate change situation.
The organic carbon storage of the soil in the coastal zone of Tianjin city under the future climate change situations of SSP245 and SSP585 in 2041-2060 and 2081-2100 is taken as an example.
Firstly, simulating and acquiring soil Salinity data (Salinity) under the future climate change situation by using a formula (1).
The meteorological data are obtained by average meteorological data (MAT and MAP) in 2041-2060 years and more than one year in 2081-2100 years under the situations of CMIP6 SSP245 and SSP 585. Population data in future climatic change scenarios are extracted from predicted data sets of chinese population in future climatic change scenarios developed by Chen (et al, 2020).
The DEM and SoilGrids can adopt the existing data, and the BNDVI, NDMI, NDVI and SAVI variables with relatively small importance can adopt the existing satellite remote sensing multi-year average data.
And (3) inputting the constructed coastal zone soil organic carbon storage quantity simulation prediction BRT model by using the obtained model input variable data, and simulating and predicting the soil organic carbon storage quantity in two future climate change situations of 2041-2060 years, 2081-2100 years, SSP245 and SSP 585.
And finally, estimating and predicting the organic carbon storage of the soil under the future climate change scene by combining possible coastal zone spatial distribution data under the future temperature rise scene, which is obtained by calculating the average sea level rise rate according to the actual sea level rise observation data (figure 6). The statistical analysis result shows that the total soil organic carbon reserves and the average carbon density of the Tianjin coastal zone in 2041 to 2060 and 2081 to 2100 years under the SSP245 condition are respectively 6.9Tg and 57.2Mg ha -1 5.9Tg and 48.9Mg ha -1 (ii) a Under the SSP585 situation, the total soil organic carbon reserves and the average carbon density of the Tianjin coastal zone in 2041 to 2060 and 2081 to 2100 years are respectively 7.1Tg C and 58.9Mg ha -1 5.6Tg C and 46.4Mg ha -1 。
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, many modifications and adaptations can be made without departing from the principle of the present invention, and such modifications and adaptations should also be considered as the scope of the present invention.
Claims (10)
1. A coastal zone soil organic carbon reserve simulation method is characterized by comprising the following steps: the method comprises the following steps:
step 1: selecting model variables required by model construction
The model variables include: a coastal vegetation index, a normalized water body index, a chlorophyll fluorescence index, an enhanced vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an improved vegetation index, a normalized salinity index, a radar remote sensing polarization index, an elevation, a terrain humidity index, an annual average temperature, an annual precipitation, a downward solar shortwave radiation, a population, a distance to a river, a soil salinity, and a reference soil carbon reserve;
step 2: model variable acquisition and processing
Firstly, acquiring model variable data through a Google Earth Engine geographic science big data cloud platform and an open-source basic geographic data product; then, carrying out format conversion and grid matching processing on the acquired model variable data by using GIS software;
and step 3: model training sample dataset preparation
Extracting a model input variable value corresponding to the soil carbon reserve sample point by using a GIS software extraction grid attribute to point tool, and acquiring a model training sample data set required by constructing a soil carbon reserve model;
and 4, step 4: model variable optimization
Selecting an optimal model variable from the model variables in the step 1 by a correlation coefficient matrix analysis and expansion factor inspection method, and acquiring optimal model variable data according to the method in the step 2;
the optimal model variables include: a coastal zone vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an elevation, a distance to a river, an annual precipitation, an annual average temperature, population, soil salinity, and a reference soil carbon reserve;
and 5: construction of reinforced regression tree model for estimating and predicting organic carbon reserves of coastal zone soil
Constructing a coastal zone soil organic carbon reserve estimation and prediction enhancement regression tree model by utilizing the coastal zone spatial distribution data and the soil organic carbon sampling point data and combining the model training sample data set obtained in the step (3);
step 6: coastal zone soil organic carbon reserve simulation
Taking the optimal model variable data obtained in the step 2 as model input, and realizing the analog estimation of the organic carbon reserves of the coastal zone soil with the spatial resolution of 10m through the estimation and prediction enhancement regression tree model of the organic carbon reserves of the coastal zone soil constructed in the step 5;
and 7: statistical analysis of organic carbon reserves in coastal zone soil
And (4) carrying out statistical analysis on the total soil organic carbon storage and the average organic carbon density of the coastal zone of the area by using the spatial distribution data of the soil organic carbon storage of the coastal zone obtained in the step (6).
2. A method for predicting the organic carbon reserve of coastal zone soil is characterized by comprising the following steps: the method comprises the following steps:
step 1: selecting model variables required by model construction
The model variables include: a coastal zone vegetation index, a normalized water body index, a chlorophyll fluorescence index, an enhanced vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an improved vegetation index, a normalized salinity index, a radar remote sensing polarization index, an elevation, a terrain humidity index, an annual average temperature, an annual precipitation, a downward solar short wave radiation, a population, a distance to a river distance, a soil salinity, and a reference soil carbon reserve;
step 2: model variable acquisition and processing
Firstly, acquiring model variable data through a Google Earth Engine geographic science big data cloud platform and an open-source basic geographic data product; then, carrying out format conversion and grid matching processing on the acquired model variable data by using GIS software;
and step 3: model training sample dataset preparation
Extracting a model input variable value corresponding to the soil carbon reserve sample point by using a GIS software extraction grid attribute to point tool, and acquiring a model training sample data set required by constructing a soil carbon reserve model;
and 4, step 4: model variable optimization
Screening optimal model variables from the model variables in the step 1 by a correlation coefficient matrix analysis and expansion factor inspection method, and acquiring optimal model variable data according to the method in the step 2;
the optimal model variables include: a coastal zone vegetation index, a normalized vegetation humidity index, an improved soil vegetation index, an elevation, a distance to a river, an annual precipitation, an annual average temperature, population, soil salinity, and a reference soil carbon reserve;
and 5: construction of reinforced regression tree model for estimating and predicting soil organic carbon reserves of coastal zone
Constructing a coastal zone soil organic carbon reserve estimation and prediction enhancement regression tree model by utilizing the coastal zone spatial distribution data and the soil organic carbon sampling point data and combining the model training sample data set obtained in the step (3);
and 6: relative importance analysis and screening of optimal model variables
Determining the relative importance of each optimal model variable screened in the step 4 to the simulation and prediction of the organic carbon in the coastal zone soil through the variable importance analysis and the partial dependence graph characteristic analysis of the constructed coastal zone soil organic carbon reserve enhancement regression tree model, and dividing the optimal model variable into an important variable, a non-important variable and a stable variable;
and 7: obtaining and processing optimal model variable data
The stable variable adopts the existing data;
non-important variables except the stable variable adopt the existing satellite remote sensing multi-year average data;
MAT and MAP data in 2061-2080 years and 2081-2100 years under two scenes of SSP245 and SSP585 predicted by adopting CMIP6 multi-model;
the population data is obtained by a Chinese population data set of a corresponding scene and a time period estimated by the model;
soil salinity data is obtained by simulating a multisource linear regression model established by adopting an AIC method through collected soil sample organic carbon sample point data and multisource data;
and 8: coastal zone spatial distribution data acquisition under future climate change situation
And calculating the average sea level rising rate calculated based on the actual sea level rising observation data, and taking the seawater submergence and land migration into consideration on the basis of the existing coastal zone spatial distribution data to obtain the coastal zone spatial distribution data under the future climate change situation.
And step 9: prediction of organic carbon reserves of coastal zone ecosystem soil under future climate change situation
And (4) taking the optimal model variable data obtained in the step (7) as the input of the coastal zone soil organic carbon storage estimation and prediction enhanced regression tree model, and predicting the coastal zone soil organic carbon storage data under different temperature-rise scenes by combining the coastal zone spatial distribution data under the future climate change scene obtained in the step (8).
3. The coastal zone soil organic carbon reserve simulation method according to claim 1 or the coastal zone soil organic carbon reserve prediction method according to claim 2, characterized in that: in the step 4, the screening principle is that when the correlation coefficient of two variables in the correlation coefficient matrix is greater than or equal to 0.8, the variables with lower correlation coefficient with the soil organic carbon reserve in the correlation coefficient matrix are removed; and eliminating variables with VIF values larger than 0.4 in all variables.
4. The coastal zone soil organic carbon reserve simulation method according to claim 1 or the coastal zone soil organic carbon reserve prediction method according to claim 2, characterized in that: in step 5, the coastal zone soil organic carbon reserve simulation and the construction and simulation of the prediction enhancement regression tree model are realized by using a Gbm.
5. The coastal zone soil organic carbon reserve simulation method or the coastal zone soil organic carbon reserve prediction method according to claim 4, characterized in that: optimizing model parameters;
the model parameter optimization method is characterized in that a gradual simulation test method is adopted, and when three key parameters of learning rate, tree complexity and bagging ratio are set to enable the model to be optimal, the optimal regression tree model for estimating and predicting the organic carbon storage of the coastal zone soil is determined.
6. The coastal zone soil organic carbon reserve simulation method or the coastal zone soil organic carbon reserve prediction method according to claim 5, characterized in that: and verifying and evaluating the constructed and optimized coastal zone soil organic carbon reserve estimation and prediction enhancement regression tree model through ten-layer cross verification (10 fold cv).
7. The coastal zone soil organic carbon reserve prediction method of claim 2, wherein: in step 6, the relative importance of the 11 optimal model variables is as follows:
annual average temperature (19.8%), soil salinity (19.7%), distance to river (13.7%), population (13.4%), coastal vegetation index (7.6%), normalized vegetation humidity index (6.1%), elevation (5.6%), reference soil carbon reserves (4.7%), normalized vegetation index (3.6%), annual precipitation (2.9%), improved soil vegetation index (2.8).
8. The coastal zone soil organic carbon reserve prediction method of claim 7, characterized in that: the dividing principle of the important variable, the non-important variable and the stable variable is as follows:
defining variables with relative importance greater than 10% as important variables; defining variables with relative importance less than 10% as non-important variables; the time-invariant small variable is defined as a stable variable.
9. The coastal zone soil organic carbon reserve prediction method of claim 8, wherein: the important variables include annual average temperature, soil salinity, distance from river and population;
the non-important variables comprise a coastal zone vegetation index, a normalized vegetation humidity index, an elevation, a reference soil carbon reserve, a normalized vegetation index, an annual precipitation and an improved soil vegetation index;
the stability variables include distance from river, elevation, and reference soil carbon reserves.
10. The coastal zone soil organic carbon reserve prediction method of claim 2, characterized in that: in step 7, the multi-source linear regression model is represented by the following formula:
Salinity＝-0.0066×Elevation+8.3523×Distance-0.0173×MAP+0.3989×MAT-0.0017×Population+7.3230。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211202453.3A CN115630567A (en) | 2022-09-29 | 2022-09-29 | Coastal zone soil organic carbon reserve simulation and prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211202453.3A CN115630567A (en) | 2022-09-29 | 2022-09-29 | Coastal zone soil organic carbon reserve simulation and prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115630567A true CN115630567A (en) | 2023-01-20 |
Family
ID=84904049
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211202453.3A Pending CN115630567A (en) | 2022-09-29 | 2022-09-29 | Coastal zone soil organic carbon reserve simulation and prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115630567A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116341724A (en) * | 2023-03-08 | 2023-06-27 | 黑龙江省生态气象中心 | Carbon absorption pre-estimating method based on global climate mode driven carbon circulation mechanism model |
| CN116563718A (en) * | 2023-07-11 | 2023-08-08 | 成都垣景科技有限公司 | Remote sensing mapping-based carbon reserve estimation method |
| CN116858999A (en) * | 2023-07-05 | 2023-10-10 | 中环宇恩(广东)生态科技有限公司 | Carbon sink statistical method based on monitoring and evaluating carbon sink potential of mangrove ecological system |
| CN117541928A (en) * | 2024-01-09 | 2024-02-09 | 南京信息工程大学 | Urban building material stock estimation method and system based on convolutional neural network |
| CN117911896A (en) * | 2024-02-19 | 2024-04-19 | 西北师范大学 | Grassland carbon sink monitoring method based on satellite remote sensing |
-
2022
- 2022-09-29 CN CN202211202453.3A patent/CN115630567A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116341724A (en) * | 2023-03-08 | 2023-06-27 | 黑龙江省生态气象中心 | Carbon absorption pre-estimating method based on global climate mode driven carbon circulation mechanism model |
| CN116341724B (en) * | 2023-03-08 | 2023-09-15 | 黑龙江省生态气象中心 | Carbon absorption pre-estimating method based on global climate mode driven carbon circulation mechanism model |
| CN116858999A (en) * | 2023-07-05 | 2023-10-10 | 中环宇恩(广东)生态科技有限公司 | Carbon sink statistical method based on monitoring and evaluating carbon sink potential of mangrove ecological system |
| CN116858999B (en) * | 2023-07-05 | 2024-01-16 | 中环宇恩(广东)生态科技有限公司 | Carbon sink statistical method based on monitoring and evaluating carbon sink potential of mangrove ecological system |
| CN116563718A (en) * | 2023-07-11 | 2023-08-08 | 成都垣景科技有限公司 | Remote sensing mapping-based carbon reserve estimation method |
| CN116563718B (en) * | 2023-07-11 | 2023-09-05 | 成都垣景科技有限公司 | Remote sensing mapping-based carbon reserve estimation method |
| CN117541928A (en) * | 2024-01-09 | 2024-02-09 | 南京信息工程大学 | Urban building material stock estimation method and system based on convolutional neural network |
| CN117541928B (en) * | 2024-01-09 | 2024-04-19 | 南京信息工程大学 | Urban building material stock estimation method and system based on convolutional neural network |
| CN117911896A (en) * | 2024-02-19 | 2024-04-19 | 西北师范大学 | Grassland carbon sink monitoring method based on satellite remote sensing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zeng et al. | Highland cropland expansion and forest loss in Southeast Asia in the twenty-first century | |
| Koch et al. | Revealing areas of high nature conservation importance in a seasonally dry tropical forest in Brazil: Combination of modelled plant diversity hot spots and threat patterns | |
| CN115630567A (en) | Coastal zone soil organic carbon reserve simulation and prediction method | |
| Stueve et al. | Spatial variability of biotic and abiotic tree establishment constraints across a treeline ecotone in the Alaska Range | |
| Lara et al. | Local-scale Arctic tundra heterogeneity affects regional-scale carbon dynamics | |
| Chaaban et al. | Accuracy assessment of ESA WorldCover 2020 and ESRI 2020 land cover maps for a Region in Syria | |
| Li et al. | Mapping population density distribution at multiple scales in Zhejiang Province using Landsat Thematic Mapper and census data | |
| Vitt et al. | Restoration and reclamation of boreal ecosystems: attaining sustainable development | |
| Zhou et al. | Assessing the impact of urban sprawl on net primary productivity of terrestrial ecosystems using a process-based model—a case study in Nanjing, China | |
| Roy et al. | Decadal land use and land cover classifications across India, 1985, 1995, 2005. ORNL DAAC, Oak Ridge, Tennessee, USA | |
| CN107247809A (en) | A kind of new method of artificial forest different age forest space mapping | |
| Park et al. | Improvement of satellite-based estimation of gross primary production through optimization of meteorological parameters and high resolution land cover information at regional scale over East Asia | |
| Jin et al. | Review on China’s spatially-explicit historical land cover datasets and reconstruction methods | |
| Santarsiero et al. | Analysis of the effect of soil erosion in abandoned agricultural areas: the case of NE area of Basilicata region (southern Italy) | |
| CN117493476A (en) | Runoff backtracking simulation method and system integrating physical mechanism and artificial intelligence | |
| Mahmoud | Integrating geoinformation and socioeconomic data for assessing urban land-use vulnerability to potential climate-change impacts of Abuja | |
| Sohail et al. | Spatio-temporal analysis of land use dynamics and its potential implications on land surface temperature in lahore district, punjab, pakistan | |
| Kumbhar et al. | Assessment of irrigation and agriculture potential using geospatial techniques: a case study of “Bhima-Ujjani” project | |
| Biswal et al. | Analysis of temporal and spatial changes in the vegetation density of Similipal Biosphere Reserve in Odisha (India) using multitemporal satellite imagery | |
| Shen | Multi-layer perceptron-markov chain based geospatial analysis of land use and land cover change: A case study of Stoney Creek Watershed, BC, Canada | |
| Ilies et al. | Integrating forest windthrow assessment data in the process of windscape reconstruction: Case of the extratropical storms downscaled for the Gutai Mountains (Romania) | |
| CN113392585B (en) | Method for spatialization of sensitive crowd around polluted land | |
| Liu et al. | Effects of Thermokarst Lake Drainage on Localized Vegetation Greening in the Yamal–Gydan Tundra Ecoregion | |
| Schneider et al. | Estimating relationships between snow water equivalent, snow covered area, and topography to extend the Airborne Snow Observatory dataset | |
| Gonzalez | Prediction of groundwater levels in a changing climate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |