CN112597948B - Urban land utilization change prediction method - Google Patents
Urban land utilization change prediction method Download PDFInfo
- Publication number
- CN112597948B CN112597948B CN202011601204.2A CN202011601204A CN112597948B CN 112597948 B CN112597948 B CN 112597948B CN 202011601204 A CN202011601204 A CN 202011601204A CN 112597948 B CN112597948 B CN 112597948B
- Authority
- CN
- China
- Prior art keywords
- urban land
- image
- remote sensing
- land use
- driving factor
- 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.)
- Active
Links
- 230000008859 change Effects 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000001413 cellular effect Effects 0.000 claims abstract description 19
- 230000009467 reduction Effects 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000003066 decision tree Methods 0.000 claims abstract description 15
- 238000013528 artificial neural network Methods 0.000 claims abstract description 12
- 238000012847 principal component analysis method Methods 0.000 claims abstract description 5
- 238000006243 chemical reaction Methods 0.000 claims description 28
- 238000005070 sampling Methods 0.000 claims description 13
- 230000009466 transformation Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 4
- 230000003595 spectral effect Effects 0.000 claims description 4
- 238000004088 simulation Methods 0.000 description 12
- 238000007477 logistic regression Methods 0.000 description 11
- 238000000513 principal component analysis Methods 0.000 description 5
- 230000007704 transition Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0463—Neocognitrons
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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"
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
- Y02A30/60—Planning or developing urban green infrastructure
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Educational Administration (AREA)
- Primary Health Care (AREA)
- Multimedia (AREA)
- Remote Sensing (AREA)
- Astronomy & Astrophysics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method for predicting urban land utilization change, which comprises the following steps: carrying out supervision and classification on the remote sensing image to obtain an urban land utilization classification image, and obtaining a driving factor image based on a driving factor; performing dimensionality reduction by using a principal component analysis method to obtain dimensionality reduction driving factor data and land utilization classification data; training an artificial neural network by using the dimension reduction driving factor data and the land utilization classification data training set, and taking the penultimate layer of the artificial neural network as extended characteristic data; according to the dimension reduction driving factor data and the expansion characteristic data, training a cellular automaton through a gradient lifting decision tree, and combining a limiting factor, probability enhancement, neighborhood scaling and neighborhood influence to obtain an urban land utilization change prediction model; and predicting the urban land use change according to the urban land use change prediction model. Compared with the prior art, the problem that characteristics cannot be fully used due to strong correlation among driving factors is solved, and the reliability and accuracy of results can be improved.
Description
Technical Field
The invention relates to the field of urban land use change prediction, in particular to an urban land use change prediction method.
Background
The urban growth is a process of converting the land utilization type into urban land utilization, is the image embodiment of urbanization, and has important guiding significance for predicting the urban growth on social production and development. In urban growth prediction, a space driving factor is a factor considered by most methods, and mainly selects space variables such as GDP (graphics data processing), education degree, distance to a first-level highway and the like, but for complex urban development, a model can not have a good fitting effect only through simple and few feature establishment, and the actual feature number is further rare due to strong correlation among the features.
The specification of the Chinese invention patent CN110826244A discloses a conjugate gradient cellular automata method for simulating rail transit to influence urban growth, which comprises the following steps: 1) Carrying out supervision and classification on the remote sensing images to obtain city spatial pattern graphs of initial and final years; acquiring urban rail transit and various infrastructure data; 2) Acquiring a space driving factor influencing the change of the urban pattern; 3) Sampling the city pattern diagram and the space variable to obtain effective sample points; 4) Training effective sample points by using conjugate gradients to obtain a CA conversion rule; 5) Acquiring the conversion probability of the land under different influences of urban rail transit based on the conversion rule; 6) Establishing a geographic CA model based on CG by combining the model elements of the cellular automata and the CG rules; 7) Simulating and predicting land use change by using a CACG model, and evaluating the precision; 8) And outputting and storing the result. However, the method is mainly focused on urban rail transit characteristics, and has the problem that the characteristics cannot be fully used due to strong correlation among driving factors, and finally the result lacks reliability and accuracy.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a method for predicting urban land use change.
The purpose of the invention can be realized by the following technical scheme:
a method for predicting urban land use change, the method comprising the steps of:
s1: acquiring a remote sensing image;
s2: carrying out supervision and classification on the remote sensing image to obtain an urban land utilization classification image, and obtaining a driving factor image based on a driving factor;
s3: performing dimensionality reduction on the urban land utilization classified image and the driving factor image by using a principal component analysis method to obtain dimensionality reduction driving factor data and land utilization classified data;
s4: training an artificial neural network by using the dimension reduction driving factor data and the land utilization classification data as a training set, and performing feature expansion on the penultimate layer of the artificial neural network to obtain expanded feature data;
s5: according to the dimension reduction driving factor data and the expansion characteristic data, training a cellular automaton through a gradient lifting decision tree, and combining a limiting factor, probability enhancement, neighborhood scaling and neighborhood influence to obtain an urban land utilization change prediction model;
s6: and predicting the urban land use change according to the urban land use change prediction model.
In S2, the urban land use classification image includes an initial year urban land use classification image and an end year urban land use classification image, and the process of obtaining the initial year urban land use classification image and the end year urban land use classification image includes:
s11: acquiring an initial year satellite remote sensing image and an end year satellite remote sensing image;
s13: and acquiring an initial year urban land utilization classification image and an end year urban land utilization classification image by using a spectral angle supervision classification method based on the initial year satellite remote sensing image and the end year satellite remote sensing image.
In S11, after the initial year satellite remote sensing image and the end year satellite remote sensing image are obtained, space reference unification and geometric correction are carried out on the initial year satellite remote sensing image and the end year satellite remote sensing image.
In S2, the driving factors comprise distances from land to highways, railways, subways, first-level roads, banks and universities, and driving factor images are obtained by calculating Euclidean distances based on remote sensing images.
And based on the remote sensing image, calculating by using Euclidean distance in ArcGIS to obtain a driving factor image.
And S3, sampling the urban land use classified image and the driving factor image by using random hierarchical sampling and system hierarchical sampling before reducing the dimension of the urban land use classified image and the driving factor image by using a principal component analysis method.
In S5, S5 includes the steps of:
s51: training the cellular automata through a gradient lifting decision tree according to the dimension reduction driving factor data and the expansion characteristic data to obtain a conversion rule of the cellular automata;
s52: obtaining urban land utilization conversion probability according to the conversion rule of the cellular automata;
s53: and establishing a land use change prediction model by combining the urban land use conversion probability, the limiting factor, the probability enhancement, the neighborhood scaling and the neighborhood influence.
In S6, the urban land use change prediction model is expressed as:
wherein,represents the cell state at time t +1>Represents the cell state at time t>Is a limiting factor, and the value of 0 or 1,0 indicates that the cell can not be developed into a city cell, and 1 indicates that the cell can be developed into a city cell, P d Representing a land use conversion probability based on a driving factor, device for selecting or keeping>Representing the domain influence, f is the probability of the total transformation P g A function of interest;
P g expressed as:
wherein, the value range of the time increment parameter TIP is 0.0-0.1, and the value range of the local adjustment parameter LAP is 0.5-1.0;
P d expressed as:
wherein f is 0 (x) Representing the initial learner, x i Representing the vector formed by the i-th driving factor, c mj Represents the best fit value, I is the indicator function;
Wherein, the central unit cell i does not participate in the calculation,representing the total number of city cells within the m x m neighborhood.
The total transformation probability P g And when the current type of the cells is larger than the set threshold value, converting the cells into the urban land type, otherwise, keeping the current type of the cells unchanged.
And S5, after the conversion rule of the cellular automaton is obtained, performing precision detection on the conversion rule, wherein the precision detection comprises a total precision index, a precision rate index, a recall rate index, a Kappa coefficient index, a figure goodness index and an F1 score.
Compared with the prior art, the invention has the following advantages:
(1) The problem that a plurality of driving factors cannot be fully utilized in model fitting is considered, principal component analysis and artificial neural network expansion are carried out, the plurality of driving factors can exert reasonable efficacy through feature extraction and enhancement, the problem that features cannot be fully used due to strong correlation among the driving factors is solved, and reliability and accuracy of results can be improved.
(2) By utilizing the cellular automata simulation framework, the geographic space variables can be conveniently extracted, the space sample data can be conveniently acquired, the preprocessing of other aspects can be carried out, and meanwhile, the framework can conveniently execute various cellular models and carry out precision evaluation on the simulation result.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a plot of a study area of an embodiment of the present invention;
FIG. 3 is a driver graph according to the present invention;
FIG. 4 shows real city land use change and CA of an embodiment of the invention FE-GBDT 、CA LR Comparing the transformation probability of the two models;
FIG. 5 shows real urban land distribution and CA in an embodiment of the present invention FE-GBDT 、CA LR The two models simulate the urban land distribution comparison diagram.
Detailed Description
The invention is described in detail below with reference to the figures and the specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Examples
The embodiment provides a city land use change prediction method based on feature expansion and gradient boosting decision trees, as shown in fig. 1, including the following steps:
1) Monitoring and classifying the satellite remote sensing images to obtain an initial year urban land utilization classified image and an end year urban land utilization classified image;
2) Obtaining a driving factor image based on the driving factor; carrying out random layered sampling and system layered sampling on the city land utilization classified image in the initial year, the city land utilization classified image in the end year and the driving factor image;
3) Performing data dimensionality reduction on the sampled urban land utilization classified image of the initial year, the sampled urban land utilization classified image of the final year and the driving factor image by using a Principal Component Analysis (PCA) method;
4) Performing feature expansion on the data after dimension reduction through an Artificial Neural Network (ANN);
5) According to the dimension reduction driving factor data and the expansion characteristic data, combining the limiting factor, the probability enhancement, the neighborhood scaling and the neighborhood influence,training a Cellular Automaton (CA) through a gradient lifting decision tree to obtain an urban land use change prediction model (CA) FE-GBDT A model);
6) Using CA FE-GBDT The model carries out simulation application and verification analysis on urban land utilization, and assessment is realized through indexes such as Overall Accuracy (OA) and figure goodness (FOM);
7) And outputting and storing the simulation result.
Specifically, the method comprises the following steps:
the step 1) is specifically as follows:
after an initial year satellite remote sensing image and an end year satellite remote sensing image are obtained, carrying out space reference unification and geometric correction on the initial year satellite remote sensing image and the end year satellite remote sensing image;
and acquiring an initial year urban land utilization classification image and an end year urban land utilization classification image by using a spectral angle supervision classification method based on the initial year satellite remote sensing image and the end year satellite remote sensing image which are subjected to spatial reference unification and geometric correction.
The step 2) is specifically as follows:
the driving factors include the distance from land to highway, railway, subway, first-level highway, bank and institution, the distance from land to restaurant, city and city center, and elevation, GDP, area and education level;
based on the remote sensing image, calculating by using Euclidean distance in ArcGIS to obtain a driving factor image;
and carrying out random hierarchical sampling and system hierarchical sampling on the initial year remote sensing image, the final year remote sensing image and the drive factor image, and providing reliable effective sample data for establishing an urban land use change prediction model.
The step 3) is specifically as follows:
and performing data dimension reduction on effective sample data by using a Principal Component Analysis (PCA) method, reserving important features, and removing noise and unimportant features.
The step 4) is specifically as follows:
training effective sample data after dimension reduction by an Artificial Neural Network (ANN) method, and taking the penultimate layer of the artificial neural network as an expanded feature.
The step 5) is specifically as follows:
s51: training the cellular automata through a gradient lifting decision tree according to the dimension reduction driving factor data and the expansion characteristic data to obtain a conversion rule of the cellular automata;
s52: obtaining urban land utilization conversion probability according to the conversion rule of the cellular automata;
s53: and establishing a land use change prediction model by combining the urban land use conversion probability, the limiting factor, the probability enhancement, the neighborhood scaling and the neighborhood influence.
The method for acquiring land transition probability data comprises the following steps:
assuming that s represents whether the cell state is converted, if the cell state is converted from Non-city (Non-Urban) to city (Urban) from time t to t +1, s is marked as 1; from time t to t +1, the cell state has not changed, and s is marked as 0.
The obtained urban land use change prediction model is expressed as follows:
wherein,represents the cell state at time t +1>Represents the cell state at time t>Is a limiting factor, and the value of 0 or 1,0 indicates that the cell can not be developed into a city cell, and 1 indicates that the cell can be developed into a city cell, P d Representing a land use conversion probability based on a driving factor, device for selecting or keeping>Representing a domain effect, f isTotal probability of transformation P g A function of interest;
a Time Increment Parameter (TIP) is used to counteract the decay of the transition probability based on the drive factor, and a Local Adjustment Parameter (LAP) is used to counteract the increase in neighborhood impact. Therefore, land use conversion probability (P) based on driving factor d ) And total probability of transformation (P) g ) This can be given by the following equation:
wherein, the TIP value range is 0.0-0.1, and the LAP value range is 0.5-1.0; f. of 0 (x) Represents an initial learner; x is the number of i Represents the vector formed by the ith driving factor; c. C mj Represents the best fit value; i is an indicator function.
For the evaluation of neighborhood impact, CA mostly adopts a regular neighborhood of squares or circles, e.g. a Moore neighborhood of m × m can be expressed as:
in the formula, the central unit cell i does not participate in calculation,the total number of city cells in the m × m neighborhood range is represented, and Moore 5 × 5 is selected as the cell neighborhood in this embodiment.
The restriction factor Con indicates that the cells are not able to develop and transform into urban cells due to some restriction, including large water bodies, basic farmlands, ecological conservation areas, parks, greenbelts, etc. Con can be expressed as:
Con=Bin(cell i (t)~ava) (4)
in the formula, con value 0 or 1,0 indicates that the cell can not be developed into a city cell, and 1 indicates that the cell can be developed into a city cell i (t) denotes the i-th cell, and ava denotes the last cell.
The urban land utilization change prediction model can be added with a random factor R for simulating the state transition of the cells caused by uncertain factors, for example, a certain cell is converted from a non-urban state to an urban state by improving the development probability through the random factor under the condition that no urban cell is nearby. The random factor R is expressed as:
R=1+(lnr) a (5)
in the formula, R is a random number between 0 and 1, alpha is a control parameter of a random factor R, and the value is an integer between 0 and 10.
Land use conversion probability P based on driving factor d Is the core part of the transformation rule, which represents the influence of the driving factors on land utilization and influences the cell state at the next moment in a probability manner. The decision tree algorithm has the characteristics of strong interpretability, high calculation speed and the like, but the problem of overfitting is easily caused by singly using the decision tree algorithm, the fitting capacity of a single decision tree is restrained by reducing the complexity of the decision tree, and a plurality of decision trees are integrated by a gradient lifting method, so that the overfitting problem can be solved to a great extent. The training of the gradient lifting decision tree method mainly comprises initializing a learner and iteratively training M trees, training the next tree by taking the residual error of the previous tree as a new real value of a sample, and continuously iterating the learner to finally obtain the learner.
Calculating the global transformation probability P of land utilization according to the formula (2) g . In actual calculation, parameters of the model are calculated by using Python language and CityAIModel software, and the calculation result is compared with a set threshold value P thd And comparing to judge whether the cells are transformed at one moment. When the global transformation probability of the cell i is larger than a set threshold value P thd And converting the cell into a city type, otherwise, keeping the state of the cell unchanged:
the step 6) is specifically as follows:
implementation of CA using Python language and CityAIModel software FE-GBDT The simulation and prediction process of the model selects a land use pattern of a certain year as an initial state and utilizes CA FE-GBDT The model runs for M times (the difference between the initial year and the final year) to obtain the simulation and prediction results of the land use change;
for CA LR Model and CA FE-GBDT And evaluating the simulation accuracy from multiple aspects according to the land utilization result of the model simulation.
And comparing the real land use classification image, and performing Precision calculation on the simulation result, wherein the main indexes are total Precision (OA), precision (Precision), recall (Recall), kappa coefficient, figure goodness (FoM) and F1 Score (F1 Score). The precision rate is the proportion of the land which is actually the city and simulated as the city to the land which is simulated as the city, and the recall rate is the proportion of the land which is actually the city and simulated as the city to the land of the actual city.
The step 7) is specifically as follows:
and outputting and storing the prediction result in GIS software.
The following is a specific example:
in the case of the land utilization of the Suzhou city in 2010-2020, the regional location in this case is shown in FIG. 2. To verify CA FE-GBDT Effectiveness of the model in land use Change simulation, in case of a CA model based on logistic regression (CA) LR ) As a comparison object, the method simulates the change process of the land utilization of the same-period cities, and the result shows that CA FE-GBDT Has better simulation effect than CA LR And (4) modeling. The process is as follows:
1) Firstly, selecting satellite remote sensing images of Suzhou city in 2000, 2010 and 2020, and obtaining urban land utilization classification images by using a spectral angle method. Taking a remote sensing image and a vector map (including an administrative region map, a road traffic map and the like) as basic data, and calculating data such as distances and elevations to an expressway, a railway, a subway and a first-level highway by using Euclidean distances to obtain a driving factor image;
2) Sampling the urban land utilization classified image and the driving factor image by using a random hierarchical sampling method;
3) Using Python language and CityThe AIModel software carries out the realization of a gradient lifting decision tree (FE-GBDT) and a Logistic Regression (LR) of feature expansion to obtain a land use conversion probability P based on a driving factor d And CA conversion rules. FIG. 3 shows urban land transitions under two models;
4) Utilizing the obtained land use conversion probability P based on the driving factor d And CA conversion rule, establishing CA FE-GBDT Model and CA LR A model;
TABLE 1 CA LR And CA FE-GBDT And (4) analyzing the results.
5) With 2010 status as initial value, respectively running CA FE-GBDT And CA LR Model 10 times, predicting 2020 land use change;
6) Comparing the result of the simulation prediction with the real city growth, and analyzing the changes of the Overall Accuracy (OA), precision (Precision), recall (Recall), kappa coefficient, figure goodness (FOM), and F1 score, as shown in table 1, it can be seen that the CA of the present embodiment FE-GBDT The method is obviously superior to CA LR A method;
9) And outputting and storing the visualized result.
Claims (1)
1. A method for predicting urban land use change is characterized by comprising the following steps:
s1: acquiring a remote sensing image;
s2: carrying out supervision and classification on the remote sensing images to obtain urban land utilization classification images, and obtaining driving factor images based on driving factors;
s3: reducing the dimensions of the urban land use classification image and the driving factor image by using a principal component analysis method to obtain dimension reduction driving factor data and land use classification data;
s4: training an artificial neural network by using the dimension reduction driving factor data and the land utilization classification data as a training set, and performing feature expansion on the penultimate layer of the artificial neural network to obtain expanded feature data;
s5: according to the dimension reduction driving factor data and the expansion characteristic data, training a cellular automaton through a gradient lifting decision tree, and combining a limiting factor, probability enhancement, neighborhood scaling and neighborhood influence to obtain an urban land utilization change prediction model;
s6: predicting the urban land utilization change according to the urban land utilization change prediction model;
in S2, the urban land use classification image includes an initial year urban land use classification image and an end year urban land use classification image, and the process of obtaining the initial year urban land use classification image and the end year urban land use classification image includes:
s11: acquiring an initial year satellite remote sensing image and an end year satellite remote sensing image;
s13: acquiring an initial year urban land utilization classification image and an end year urban land utilization classification image by using a spectral angle supervision classification method based on the initial year satellite remote sensing image and the end year satellite remote sensing image;
s11, after the initial year satellite remote sensing image and the end year satellite remote sensing image are obtained, carrying out spatial reference unification and geometric correction on the initial year satellite remote sensing image and the end year satellite remote sensing image;
the driving factors comprise the distances from land to highways, railways, subways, first-level highways, banks and universities, and driving factor images are calculated by utilizing Euclidean distances on the basis of remote sensing images;
based on the remote sensing image, calculating by using Euclidean distance in ArcGIS to obtain a driving factor image;
s3, sampling the urban land use classified images and the driving factor images by utilizing random hierarchical sampling and system hierarchical sampling before reducing dimensions of the urban land use classified images and the driving factor images by utilizing a principal component analysis method;
in S5, S5 includes the steps of:
s51: training the cellular automata through a gradient lifting decision tree according to the dimension reduction driving factor data and the expansion characteristic data to obtain a conversion rule of the cellular automata;
s52: obtaining urban land utilization conversion probability according to the conversion rule of the cellular automata;
s53: establishing a land use change prediction model by combining the urban land use conversion probability, the limiting factor, the probability enhancement, the neighborhood scaling and the neighborhood influence;
the urban land use change prediction model is expressed as:
wherein,represents the cell state at time t +1, is present>Represents the cell state at time t>Is a limiting factor, and the value of 0 or 1,0 indicates that the cell can not be developed into a city cell, and 1 indicates that the cell can be developed into a city cell, P d Representing a land use conversion probability based on a driving factor, device for selecting or keeping>Representing the domain influence, f is the probability of the total transformation P g A function of interest;
P g expressed as:
wherein, the value range of the time increment parameter TIP is 0.0-0.1, and the value range of the local adjustment parameter LAP is 0.5-1.0;
P g expressed as:
wherein f is d (x) Representing the initial learner, x i Representing the vector formed by the i-th driving factor, c mj Represents the best fit value, I is the indicator function;
wherein, the central unit cell i does not participate in the calculation,representing the total number of city cells within the m × m neighborhood range;
the total transformation probability P g When the current type of the cells is larger than the set threshold value, converting the cells into the urban land type, otherwise, keeping the current type of the cells unchanged;
and S5, after the conversion rule of the cellular automaton is obtained, performing precision detection on the conversion rule, wherein the precision detection comprises a total precision index, a precision rate index, a recall rate index, a Kappa coefficient index, a figure goodness index and an F1 score.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011601204.2A CN112597948B (en) | 2020-12-29 | 2020-12-29 | Urban land utilization change prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011601204.2A CN112597948B (en) | 2020-12-29 | 2020-12-29 | Urban land utilization change prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112597948A CN112597948A (en) | 2021-04-02 |
CN112597948B true CN112597948B (en) | 2023-03-28 |
Family
ID=75203855
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011601204.2A Active CN112597948B (en) | 2020-12-29 | 2020-12-29 | Urban land utilization change prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112597948B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116976526B (en) * | 2023-09-20 | 2023-12-15 | 北京师范大学 | Land utilization change prediction method coupling ViViViT and ANN |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110909924A (en) * | 2019-11-12 | 2020-03-24 | 同济大学 | City expansion multi-scenario simulation cellular automata method based on cross entropy optimizer |
CN110991262A (en) * | 2019-11-12 | 2020-04-10 | 同济大学 | Multi-bandwidth geographical weighted regression cellular automata method for ecological service value prediction |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102222313B (en) * | 2010-04-14 | 2013-05-01 | 同济大学 | Urban evolution simulation structure cell model processing method based on kernel principal component analysis (KPCA) |
CN106021751B (en) * | 2016-05-26 | 2019-04-05 | 上海海洋大学 | Littoral zone simulation of land use changes method based on CA and SAR |
CN109816581A (en) * | 2019-01-25 | 2019-05-28 | 东南大学 | A kind of urban land automatic recognition system of comprehensive industry situation big data and Form of Architecture |
CN110991497B (en) * | 2019-11-15 | 2023-05-02 | 同济大学 | BSVC (binary sequence video coding) -method-based urban land utilization change simulation cellular automaton method |
CN110826244B (en) * | 2019-11-15 | 2024-04-26 | 同济大学 | Conjugated gradient cellular automaton method for simulating influence of rail transit on urban growth |
CN111080070B (en) * | 2019-11-19 | 2023-05-02 | 同济大学 | Urban land utilization cellular automaton simulation method based on space errors |
CN111274905A (en) * | 2020-01-16 | 2020-06-12 | 井冈山大学 | AlexNet and SVM combined satellite remote sensing image land use change detection method |
CN112131731B (en) * | 2020-09-15 | 2022-06-14 | 同济大学 | Urban growth cellular simulation method based on spatial feature vector filtering |
-
2020
- 2020-12-29 CN CN202011601204.2A patent/CN112597948B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110909924A (en) * | 2019-11-12 | 2020-03-24 | 同济大学 | City expansion multi-scenario simulation cellular automata method based on cross entropy optimizer |
CN110991262A (en) * | 2019-11-12 | 2020-04-10 | 同济大学 | Multi-bandwidth geographical weighted regression cellular automata method for ecological service value prediction |
Also Published As
Publication number | Publication date |
---|---|
CN112597948A (en) | 2021-04-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11657708B2 (en) | Large-scale real-time traffic flow prediction method based on fuzzy logic and deep LSTM | |
CN110991497B (en) | BSVC (binary sequence video coding) -method-based urban land utilization change simulation cellular automaton method | |
CN110909924B (en) | Urban expansion multi-scenario simulation cellular automaton method based on cross entropy optimizer | |
CN109359166B (en) | Space growth dynamic simulation and driving force factor contribution degree synchronous calculation method | |
CN108629978A (en) | A kind of traffic trajectory predictions method based on higher-dimension road network and Recognition with Recurrent Neural Network | |
CN106228125A (en) | Method for detecting lane lines based on integrated study cascade classifier | |
CN108090624B (en) | Urban ecological safety simulation and prediction method for improving cellular automaton | |
CN110991262A (en) | Multi-bandwidth geographical weighted regression cellular automata method for ecological service value prediction | |
CN113222316A (en) | Change scene simulation method based on FLUS model and biodiversity model | |
CN110826244B (en) | Conjugated gradient cellular automaton method for simulating influence of rail transit on urban growth | |
CN111814596A (en) | Automatic city function partitioning method for fusing remote sensing image and taxi track | |
CN113836999A (en) | Tunnel construction risk intelligent identification method and system based on ground penetrating radar | |
CN110675626A (en) | Traffic accident black point prediction method, device and medium based on multidimensional data | |
Zhang et al. | Using street view images to identify road noise barriers with ensemble classification model and geospatial analysis | |
CN114283285A (en) | Cross consistency self-training remote sensing image semantic segmentation network training method and device | |
CN104850677A (en) | Deep learning cellular automaton simulation analysis method based on geographic information system | |
CN112597948B (en) | Urban land utilization change prediction method | |
CN116662468A (en) | Urban functional area identification method and system based on geographic object space mode characteristics | |
CN105243503A (en) | Coastal zone ecological safety assessment method based on space variables and logistic regression | |
CN117556197A (en) | Typhoon vortex initialization method based on artificial intelligence | |
CN117238126A (en) | Traffic accident risk assessment method under continuous flow road scene | |
CN105404858A (en) | Vehicle type recognition method based on deep Fisher network | |
CN105335758A (en) | Model identification method based on video Fisher vector descriptors | |
Chao et al. | A spatio-temporal neural network learning system for city-scale carbon storage capacity estimating | |
CN109636194B (en) | Multi-source cooperative detection method and system for major change of power transmission and transformation project |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |