CN109002621A - A kind of mean height and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method for taking neighborhood and geographical difference into account - Google Patents
A kind of mean height and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method for taking neighborhood and geographical difference into account Download PDFInfo
- Publication number
- CN109002621A CN109002621A CN201810829343.7A CN201810829343A CN109002621A CN 109002621 A CN109002621 A CN 109002621A CN 201810829343 A CN201810829343 A CN 201810829343A CN 109002621 A CN109002621 A CN 109002621A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- bottom class
- tree
- forest
- 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.)
- Granted
Links
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
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of mean heights and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method for taking neighborhood and geographical difference into account, including data are extracted, geographical area statistics, are established hierarchical data sets, model training, model accuracy amendment, model calculating and checked the several links of overall data.The invention proposes one with growing environment, tree species, based on management state and all kinds of survey datas, to with history survey data and lacking the modeling process that the not applicable tree species of growth model or existing growth model combine existing common all kinds of growth models, and application space range statistics, space interpolation, Geographical Weighted Regression, machine learning and ArcGIS software building model, it can be according to the region bottom class dominant tree mean stand height of data cases self-teaching and optimization and mean DBH increment calculation method, can at the regional level in establish computation model for target bottom class dynamic the mean stand height of bottom class calculated with the diameter of a cross-section of a tree trunk 1.3 meters above the ground, reliable data supporting can be provided for the calculating of forest reserves.
Description
Technical field
The present invention relates to Stand Growth studying technological domain, specific field is a kind of standing forest for taking neighborhood and geographical difference into account
Mean height and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method.
Background technique
In recent years, as many as Stand Growth Model quantity, range are wide, have reached unprecedented degree.From the mark of early stage
The individual growing model of quasi- yield table till now has gradually formed a continuous Unified Model body.Due to establishing model
Purpose it is different, the mathematical method of tectonic model is also different.Burkhart and Avery, Tang Shouzheng are according to the scale of model and pre-
Estimate result and Stand Growth Model is divided into following three classes: the Whole stand model based on standing forest general characteristic target variable
(growth model that can be divided into variable density model and averag density standing forest) is distributed with the diameter rank that forest grade is basic analogue unit
Model and the individual growing model (can be divided into distance dependent and apart from unrelated two class) based on Single tree growth information.Mesh
Before, Stand Growth and Volume model are gradually horizontal from stand level model to the horizontal model of diameter grade, single wood.Initially, simulating forest
Dynamic tool is chart, using it is most common be exactly yield table.Earliest yield table is German silviculturist Reanmur in the world
In the yield table that 1721 deliver, the concern of various countries forestry researcher is caused, the country such as subsequent Russia, Japan, U.S. is also sent out
The a series of yield table of table.It is the Different forest stands for recording a certain region respectively in table form, locality class in different lifes
A kind of method of long stage stand facters.With the continuous development of mathematical formulae, nineteen thirty-seven Mackinlley et al. is for the first time returning
Analytical technology is returned to be applied in the research of Forest Growth and harvest.The application of regression analysis technique has forest litterfall analogy method
Important breakthrough, yield table are changed into variable density yield table from pervious constant density yield table, and the precision of yield table also has
It is improved, while this new model of Forest Growth function occurs and carrying out simulating forest dynamic change.Hereafter a large amount of scholar is based on
Homing method combines different parameters, and 1964, J.H.smith and Newham established the Individual tree model with distance dependent.
Buckman establishes regression equation using the density of crop for the first time, directly estimates stand growth.Bailey was established with age, close
Degree, site index, stand type and thinning intensity are the Slash Pine Plantations strain number compatibility of input variable.
Pienaar establishes the growth of Zululand Slash Pine Plantations and yield forecast model.Khatouri utilizes Morocco's Chinese mugwort Gustaf Dyrssen
One first step of non linear equation group of the temporary sample plot data mining of woods deodar, for predicting the diameter distribution and its growth of deodar
And harvest.Wenk studies and compares growth and the yield forecast model of pure forest and mixed forest.Wimberly uses regression analysis
Method establish fir and distance dependent and with apart from unrelated single wooden basal area growth equation.Fernando Castedo
Dorado is directed to the pine artificial forest of Galicia (the Spain northwestward), fixed and random by the estimation of mixed model technology
Parameter constructs Generalized Tree height-diameter of a cross-section of a tree trunk 1.3 meters above the ground model based on Schnute function.Huuskonen is Finland with resetting sample data
Young age pinus sylvestris var. mongolica establishes nonlinear mixed-effect model, analyzes the Dominant height and average diameter of pinus sylvestris var. mongolica, and is applicable in non-
Linear assembly language constructs basal area and timber volume model.Coble is directed to the torch pine and wet-land pine tree of eastern Texas,
Develop the year model of growth of a kind of predictable individual tree survival and growth in thickness.Sghaier utilizes Richards,
Four difference equations that the basic function of Lundqvist, Hossfeld IV and Weibull obtain, use the growth unrelated with the age
Function formula is tested, establish Tunisia northeast arbor-vitae wildwood with apart from unrelated single wooden diameter of a cross-section of a tree trunk 1.3 meters above the ground model of growth.Two
After 80th century of tenth century, substantially seldom occur that there is the new model of breakthrough meaning, most of is all to existing model
Improvement, Trimble and Shriner have only just summed up 100 Stand Growth Models in US range.
It starts late to the research of forest, Stand Growth and Volume model in China.Just there is yield table in the forties in last century,
Just start the simulation of system research Plantation Growth, a large amount of scholar and researcher to the latter stage eighties to manage from the growth of standing forest
By setting out, density is introduced, site index, Advantage height, interaction, human activity factor and environmental factor between tree species are established
Relevant full standing forest dynamic model.The mid-80 beginning researcher studies single wooden model using the part tree species in the north
Type.In recent years, also someone to the southern main timber tree species such as progress of China fir, masson pine, wet-land pine tree, natural spruce forest list
The research of the wooden growth model, Cheng Guangming connect the reset sample trees increment determination data in clear permanent sample plot using the forest reserves, with
Volume increment percent P is dependent variable, and age A and diameter of a cross-section of a tree trunk 1.3 meters above the ground D are auxiliary variable, selects P=a DbAcFor basic model, using gradually returning
Technology is returned to establish the volume increment percent model of variable element, and with modified simplex method Optimal Parameters.Chen Wenxiong uses Lianjiang County
The even reset sample trees of the Slash Pine Plantations permanent sample plot in clear data, using age, the diameter of a cross-section of a tree trunk 1.3 meters above the ground as independent variable, using Richard's equation
Volume increment percent model is constructed, optimizes fitting with Immune Evolutionary Algorithm.These research work all have highly important meaning
Justice, it lays the foundation for popularization and application of the individual growing model in China.
Since the purpose for establishing model is different, the mathematical method of tectonic model is also different, and traditional method uses mostly
Be rely on mathematical equation statistical method and a large amount of sample data, these method operations are cumbersome, using personnel need to have compared with
Strong mathematical modeling and misarrangement ability.With the development of machine learning, more and more researchers are introduced into standing forest diameter point
In the research of cloth, but existing model and method also has the disadvantage that the value volume and range of product of model existing first is various, but
It is that almost all of model does not all account for other management levels (such as apply fertilizer, foster) and standing forest itself in addition to intermediate cutting
Influence of the growth rhythm to Stand Growth；Secondly at present it is most research both for management common tree (such as
China fir, masson pine, larch etc.), modeling process needs a large amount of sample and sample trees data, and the collection of data needs to expend a large amount of
Human cost and goods and materials cost, it is difficult to cover all areas and tree species, and simultaneously because model the region limitation of selected sample trees
Property, cause the model established that cannot simply be applied to other regions and tree species；There are also a bit, due to the growth of forest have compared with
Strong space-time characteristic, it is therefore desirable to the model for being suitable for target area space-time characteristic is established, and existing model is mostly with sample early period
Ground sample trees data (before 20,30,50 years) establish the model estimated whole numerical value from global feature, and have ignored
The otherness of bottom class's individual, and model lacks self-correction and design seismic wave Data Matching and adaptive ability.
Summary of the invention
The purpose of the present invention is to provide a kind of mean height for taking neighborhood and geographical difference into account and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method,
To solve the problems mentioned in the above background technology.
To achieve the above object, the present invention provides following calculation method, including the following steps:
1) data are extracted:
A, using forest inventory control platform software, previous year forest reserves bottom class data are replicated, create field backup
Bottom class's mean height, mean DBH increment, and bottom class's mean height, mean DBH increment are emptied, bottom class's data are decomposed into and have nature growth mould
The tree species bottom class of type and the tree species bottom class for lacking nature growth model carry out data update using corresponding model, will lack certainly
Right growth model bottom class data establish level figure layer by tree species, age, county and forest farm range, forest ownership, and the age is corresponding
Add 1, establishes target tree species to be calculated bottom class distributed data collection T0；
B, according to T0Middle tree species, age, county and forest farm range, forest ownership are from previous year forest reserves bottom class data
Corresponding bottom class's data are extracted respectively, establish corresponding figure layer data set S0；
C, annual cutting area data D is arranged1, and tree height, the diameter of a cross-section of a tree trunk 1.3 meters above the ground are corrected or backfilled according to age, tree species, bottom class number
To S0Corresponding bottom class；
2) geographic area counts:
A, using the application region statistical function of ArcGIS software to T0With S0Zhong Ge bottom class is by graphics field statistics soil point
Cloth data D3, bioclimate variable data D4, solar radiation quantity D5-2, terrain information D5-3Each achievement data, in T0With S0In
Increasing corresponding field newly will be in the corresponding newly-built field of value filling；
B, data D is fostered in management2Forest ownership is used in the case where lacking Operation Measures data；
C, basin classification data D5-1According to each basin distributed data that hydrological analysis tool obtains, by by west in region
To east, from north orientation southing row Unified number, by number result assignment to T0With S0In the basin field of data set；
3) hierarchical data sets are established:
Using the selection function in ArcGIS software by tree species, age, forest ownership by T0With S0Age-based level layering,
Establish T1With S1Data set, while verifying survey data if there is previous year field operation, then this partial data is accordingly also added
To S1In data set；
4) model training:
For S1Each tree species, annual bottom class data Layer, carry out model training and accuracy checking respectively in data set, save full
The model of sufficient required precision simultaneously calculates corresponding T with the model1Bottom class's figure layer data in data set carry out step 6) model meter
It calculates, the model for being unsatisfactory for required precision carries out the amendment of step 5) model accuracy；
Model training mode has following three kinds of training methods available according to the demand of user:
A. based on by Spatial Variability with correlation, passed through according to target tree species, age, working group in investigation tree species
Semivariable function calculates the high correlation distance spatially of the diameter of a cross-section of a tree trunk 1.3 meters above the ground, tree, calculates neighbouring specified quantity (generally by the distance
5,10,15 or all) bottom class to the weighing factor of target bottom class to be calculated (anti-distance, Gauss distance, exponential distance etc.), most
It is common to calculate afterwards according to the diameter of a cross-section of a tree trunk 1.3 meters above the ground of neighbouring bottom class, the high diameter of a cross-section of a tree trunk 1.3 meters above the ground, the tree high level for calculating target bottom class with the weight sum of products of tree
Method has the methods of GWR, Ke Lvjin interpolation, geographical recurrence；
B. the total factor regression model based on machine learning: the growth of forest is more by natural environment, tree species, business activities etc.
The influence of kind factor, under different space-time conditions, influence of each factor to forest is different, therefore can be with the analysis side of total factor
Method is analyzed by factor correlativity, is removed the high factor of correlation, is passed through the machines such as decision tree, random forest, artificial neural network
Device learning method carries out model training to area data, according to pre-set required precision, correlation model is saved, for not
The tree species for meeting required precision establish corresponding Sub Data Set, then proceed by the amendment of step 5) model accuracy；
The method that c.b is combined with a: comprehensive analysis is carried out to factor element by Random Forest model, is returned according to model
The preceding 5-10 factor may be selected as empty to improve calculating speed and performance in important (Importance) the property sequencing table of the factor returned
Interpolation variable selects collaboration Ke Lvjin model to carry out the diameter of a cross-section of a tree trunk 1.3 meters above the ground, the interpolation of height increment, distribution map is generated, according to interpolation knot
Fruit using area statistical function calculates the average value of bottom class to be updated, median etc., provides the diameter of a cross-section of a tree trunk 1.3 meters above the ground, tree Gao Gengxin basic data；
5) model accuracy is corrected:
From the history bottom class distributed data for the tree species for being unsatisfactory for modeling accuracy requirement, random or system extract 40% it is small
Class's data generate field operation supplement survey data, carry out tree height, diameter of a cross-section of a tree trunk 1.3 meters above the ground field operation supplement survey, survey data is back to S1Data set
In, then carry out step 4) model training to model accuracy again and reaches requirement.The partial data will also enter next year simultaneously
It spends in model training；
6) model calculates:
Applying step 4) model training model result to T1Corresponding tree species in data set, bottom class's number corresponding to the age
According to being calculated, abnormal data is checked and is corrected；
7) performance data is checked:
Check data processing precision, the data that precision is unsatisfactory for requiring then carry out the amendment of step 5) model accuracy, and precision is full
Foot requires that the Update order of the predict function of result application model or SQL statement is then updated to T0, then generate final
Annual update performance data T3。
Preferably, the hierarchical data sets referred in the step 3) can may be virtual figure layer for solid data figure layer.
Compared with prior art, the beneficial effects of the present invention are: the present invention by building one with growing environment, tree species,
Based on management state and all kinds of survey datas, to history survey data (a kind of, two classes, cutting area) and shortage life
The tree species that long model or existing growth model are not suitable for, in conjunction with the modeling process of existing common all kinds of growth models, using sky
Between range statistics, space interpolation, Geographical Weighted Regression, machine learning and ArcGIS software, can according to data cases self learn
The region bottom class dominant tree mean stand height and mean DBH increment calculation method with optimization are practised, it can be interior small for target at the regional level
Class's dynamic is established computation model and is calculated with the diameter of a cross-section of a tree trunk 1.3 meters above the ground the mean stand height of bottom class, can provide reliably for the calculating of forest reserves
Data supporting.
Detailed description of the invention
Flow diagram of the Fig. 1 by the calculation method proposed of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Referring to Fig. 1, the present invention provides following calculation method, including the following steps:
1) data are extracted:
A, using forest inventory control platform software, which is the common basic software of existing forestry research, or is used
Other softwares with same function can also, replicate previous year forest reserves bottom class data, it is average to create field backup bottom class
High, mean DBH increment, and bottom class's mean height, mean DBH increment are emptied, bottom class's data are decomposed into the tree species for having nature growth model
Bottom class and the tree species bottom class for lacking nature growth model carry out data update using corresponding model, will lack nature and grow mould
Type bottom class data establish level figure layer by tree species, age, county and forest farm range, forest ownership, and the age is added 1 accordingly, is established
Target tree species to be calculated bottom class distributed data collection T0；
B, according to T0Middle tree species, age, county and forest farm range, forest ownership are from previous year forest reserves bottom class data
Corresponding bottom class's data are extracted respectively, establish corresponding figure layer data set S0；
C, annual cutting area data D is arranged1, and tree height, the diameter of a cross-section of a tree trunk 1.3 meters above the ground are corrected or backfilled according to age, tree species, bottom class number
To S0Corresponding bottom class；
2) geographic area counts:
A, using the application region statistical function of ArcGIS software to T0With S0Zhong Ge bottom class is by graphics field statistics soil point
Cloth data D3, bioclimate variable data D4, solar radiation quantity D5-2, terrain information D5-3Average value, maximum value, minimum value, in
Each achievement data such as digit, in T0With S0In increase corresponding field newly will be in field corresponding to value filling；
B, data D is fostered in management2Forest ownership is used in the case where lacking Operation Measures data；
C, basin classification data D5-1According to each basin distributed data that hydrological analysis tool obtains, by by west in region
To east, from north orientation southing row Unified number, by number result assignment to T0With S0In the basin field of data set；
3) hierarchical data sets are established:
Using the selection function in ArcGIS software by tree species, age, forest ownership by T0With S0Age-based level layering,
Establish T1With S1Data set, while verifying survey data if there is previous year field operation, then this partial data is accordingly also added
To S1In data set；
4) model training:
For S1Each tree species, annual bottom class data Layer, carry out model training and accuracy checking respectively in data set, save full
The model of sufficient required precision simultaneously calculates corresponding T with the model1Bottom class's figure layer data in data set carry out step 6) model meter
It calculates, the model for being unsatisfactory for required precision carries out the amendment of step 5) model accuracy；
Model training is segmented into three kinds of training methods:
A. bottom class's mean DBH increment based on Spatial Interpolation Method/high calculation method of tree: using Spatial Variability with correlation as base
Plinth calculates the diameter of a cross-section of a tree trunk 1.3 meters above the ground, tree height spatially by semivariable function according to target tree species, age, working group in investigation tree species
Correlation distance, calculate the bottom class of neighbouring specified quantity (generally 5,10,15 or all) to mesh to be calculated by the distance
Mark the weighing factor (anti-distance weighting, Gauss distance weighting, exponential distance weight etc.) of bottom class, the last chest according to neighbouring bottom class
The high diameter of a cross-section of a tree trunk 1.3 meters above the ground, the tree high level that target bottom class is calculated with the weight sum of products of diameter, tree, common calculation method have GWR, Ke Lvjin to insert
Value, it is geographical the methods of return, be described below and of the invention applied a kind of carry out that the diameter of a cross-section of a tree trunk 1.3 meters above the ground, tree are high to insert based on common stock Krieger
It is worth calculation process and related mathematical formulae, it is as follows calculates basic step:
(1) the distance between known bottom class's (using mass center as calculating position) is calculated, Ordinary Kriging Interpolation mean DBH increment, tree are solved
High interpolation equation group needs first to construct
Above formula is reduced to γ × λ=g
Coefficient matrix and right side array, i.e., first seek variation γijWith gi0, and variogram is between bottom class
Distance dependent, and it is unrelated with the coordinate of bottom class, therefore can first calculate the distance between each bottom class；
(2) by sequence is lined up from small to large, then distance grouping will first be calculated resulting distance value by Euclidean distance formula
Distance value is divided into several groups, every group includes a certain number of distance values；
Euclidean distance calculation formula:
(3) it is calculated using fitting variogram γ (h)-h.For every group of distance value, average distance is calculatedAnd press formula
The estimated value of the dynamic difference in group function is calculated, then selects certain variogram theoretical model may be selected spherical
Model, then Function Fitting, finds out model parameter, to obtain the expression formula of variogram γ (h)；
Spherical model:
(4) array of calculating formula γ × λ=g coefficient matrix and right side, will corresponding known bottom class and bottom class to be calculated it
Between distance value substitute into variogram γ (h) in, calculate all variation γ in formulaij；
(5) estimated value for calculating bottom class solves λ=g × γ for each bottom class to be calculated-1, all correspondences can be obtained
Distance weighting coefficient lambdan, then according toMean DBH increment, tree height can be acquired to be calculated
Bottom class x0The estimator at place, since γ × λ=g coefficient matrix λ is not change with bottom class's variation is calculated, for
For all calculating bottom classes in target area, coefficient matrix can be formed once, but the array on right side is different because of bottom class to be calculated
And it is different, therefore, each estimation bottom class is needed individually to calculate；
B. the total factor regression model based on machine learning: the growth of forest is more by natural environment, tree species, business activities etc.
The influence of kind of factor, under different space-time conditions, influence of each factor to forest is different, therefore the present invention is with the analysis of total factor
Method is analyzed by factor correlativity, passes through decision tree, random forest, artificial neural network removing the high factor of correlation
Equal machine learning methods carry out model training to area data, according to pre-set required precision, save correlation model, right
In the tree species for being unsatisfactory for required precision, corresponding Sub Data Set is established, the amendment of step 5) model accuracy is carried out, the present invention is described below
Applied in it is a kind of based on random forest method carry out regression model foundation process and relevant calculation formula:
(1) according to step 3) establish hierarchical data sets as a result, carrying out autocorrelation analysis to all factor rejects VIF
> 10 factor, wherein VIF=1/1-R2, R2It is with XjThe coefficient of multiple determination that other independents variable are returned when for dependent variable, finally
The factor of reservation is influenced and different by tree species and region, therefore not tired is here stated；
(2) Bootstraping method is concentrated use in from original training puts back to sampling at random and select m bottom class, carry out altogether
N_tree sampling, generates n_tree trained bottom class data set；
(3) for n_tree trained bottom class data set, n_tree decision-tree model is respectively trained；
(4) for single decision-tree model, it is assumed that training bottom class has N number of feature, therefrom randomly chooses n feature and carries out
Decision tree building selects best feature than, gini index according to information gain, information gain when each decision tree nodes divide
It is divided；
Wherein, comentropy calculation formula:
Information gain: IG (X)=H (c)-H (c | X)
Information gain ratio are as follows: and gr=H (c)-H (c | X) H (X)
In formula: p (Xi) it be some feature value is XiProbability.
(5) each tree all go down so always by division, until all trained bottom classes of the node belong to same class.Certainly
Beta pruning is not needed in the fission process of plan tree；
(6) more decision trees of generation are formed into random forest, the mean value for setting predicted value by more determines final to be calculated
The result of the mean stand height of bottom class, the diameter of a cross-section of a tree trunk 1.3 meters above the ground；
The method that c.b is combined with a: comprehensive analysis is carried out to factor element by Random Forest model, is returned according to model
The preceding 5-10 factor may be selected as empty to improve calculating speed and performance in important (Importance) the property sequencing table of the factor returned
Interpolation variable selects collaboration Ke Lvjin model to carry out the diameter of a cross-section of a tree trunk 1.3 meters above the ground, the interpolation of height increment, distribution map is generated, according to interpolation knot
Fruit using area statistical function calculates the average value of bottom class to be updated, median etc., provides the diameter of a cross-section of a tree trunk 1.3 meters above the ground, tree Gao Gengxin basic data；
5) model accuracy is corrected:
From the history bottom class distributed data for the tree species for being unsatisfactory for modeling accuracy requirement, random or system extract 40% it is small
Class's data generate field operation supplement survey data, carry out tree height, diameter of a cross-section of a tree trunk 1.3 meters above the ground field operation supplement survey, survey data is back to S1Data set
In, then carry out step 4) model training to model accuracy again and reaches requirement.The partial data will also enter next year simultaneously
It spends in model training；
6) model calculates:
Applying step 4) model training model result to T1Corresponding tree species in data set, bottom class's number corresponding to the age
According to being calculated, abnormal data is checked and is corrected；
7) performance data is checked:
Check data processing precision, the data that precision is unsatisfactory for requiring then carry out the amendment of step 5) model accuracy, and precision is full
Foot requires that the Update order of the predict function of result application model or SQL statement is then updated to T0, then generate final
Annual update performance data T3。
Specifically, the hierarchical data sets referred in the step 3) can may be void for solid data figure layer
Quasi- figure layer, according to First Law of Geography, according to the different demands and data cases of user, the setting of the hierarchical data sets
Method has following two:
A. by T0In each tree species using bottom class as minimum calculation unit, for certain tree species by the range in block -> forest farm -> at county level
Bottom class's data retrieval is carried out, until data volume meets pre-set bottom class's quantity, establishes corresponding data set；
B. geo-relevance is pressed, is calculated according to target tree species, age, working group in investigation tree species by semivariable function
The high correlation distance spatially of the diameter of a cross-section of a tree trunk 1.3 meters above the ground, tree out, selects same type history bottom class data according to this distance, establishes data set.
Specifically, all data needed for the calculation method can be obtained according to following sources or mode:
Forest resources inventory and planning data: service area annual forest inventory for planning and designing performance data recently,
Including spatial position, engineering type, class, the seed of forest, ownership, dominant tree, age, mean DBH increment, mean stand height, the gradient, slope
To in-place inventories factor datas such as, slope position, associated species；
Annual cutting area data D1: the every wooden dipping data of bottom class's study plot in Felling area design investigation, and ask calculation
Bottom class's mean stand height, mean DBH increment mainly include the factors such as tree species, bottom class's number, tree height, the diameter of a cross-section of a tree trunk 1.3 meters above the ground, age；
Data D is fostered in management2: deriving from management file data, including manage intensity, Operation Measures etc.；
SOIL DISTRIBUTION data D3: Cold and drought Region scientific data center is derived from, data type is Chinese soil data set grid
Lattice data, provider are generation constructed by FAO (Food and Agriculture Organization of the United Nation) (FAO) and Vienna international applications system research institute (IIASA)
Boundary's Soil Database (Harmonized World Soil Database, HWSD), spatial resolution 1km or regional soil
Survey data includes at least: the factors such as soil types, thickness, grit content, humus thickness；
Bioclimate variable data D4: from global climate data WorldClim (http: //
Www.worldclim.org), 19 bioclimate variables including 1950-2000, spatial resolution 1km；
DME D5: data source is freely shared in geographical spatial data cloud platform (http://www.gscloud.cn)
ASTER GDEMV2 data, the data are by Japanese METI and U.S.'s NASA joint research and development and freely distribute towards the public, space point
The dem data that resolution generates after being 30m or the above topographic maps vector quantization of region 1:10000；
Basin classification data D5-1: based on DEM by the hydrological analysis tool in ArcGIS software, obtain basin division result；
Solar radiation quantity D5-2: global climate data WorldClim (http://www.worldclim.org) is derived from,
Or combine solar radiation quantity data of the DEM by relevant Software Create region；
Terrain information D5-3: the gradient, with a varied topography is generated by the terrain analysis function of ArcGIS software based on dem data
Property, terrain roughness, slope position etc. data；
Detailed forest resources inventory and planning data can be found in the following table 1 and table 2 because of sublist.
1 forest resources inventory and planning data of table is because of sublist
2 soil of table, biological variable and environmental factor
Working principle: the present invention uses big data analysis method, main to be extracted, geographical area statistics, established by data
Hierarchical data sets, model training, model accuracy amendment, model calculate and check the several links of overall data, set to Forest Planning
Count bottom class's mean stand height in each year between two investigation phases of investigation, the diameter of a cross-section of a tree trunk 1.3 meters above the ground estimation mainly based on historical data, tie
Close mathematical statistics, machine learning, spatial variability, the methods of space interpolation establish tree species, the age, environmental factors, Operation Measures with
The regression model of Tree Height And The Diameter Breast Height calculates the dominant tree mean stand height and mean DBH increment of target bottom class, in calculated result
Tree species and standing forest to precision lower than 90% carry out random sampling for examining on the spot, further according to on-site inspection result data to mould
Type is modified, until model computational accuracy reaches application requirement.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
A variety of variations, modification, replacement can be carried out to these embodiments without departing from the principles and spirit of the present invention by understanding
And modification, the scope of the present invention is defined by the appended.
Claims (2)
1. a kind of mean height and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method for taking neighborhood and geographical difference into account, it is characterised in that: the calculating side
Method includes the following steps:
1) data are extracted:
A, using forest inventory control platform software, previous year forest reserves bottom class data are replicated, backup bottom class's mean height is put down
The equal diameter of a cross-section of a tree trunk 1.3 meters above the ground, and empty bottom class's mean height, mean DBH increment, bottom class's data are decomposed into the tree species bottom class for having nature growth model and
The tree species bottom class for lacking nature growth model carries out data update using corresponding model, will lack nature growth model bottom class
Data establish level figure layer by tree species, age, county and forest farm range, forest ownership, and the age is added 1 accordingly, establish to be calculated
Target tree species bottom class distributed data collection T0；
B, according to T0Middle tree species, age, county and forest farm range, forest ownership are distinguished from previous year forest reserves bottom class data
Corresponding bottom class's data are extracted, corresponding figure layer data set S is established0；
C, annual cutting area data D is arranged1, and the high, diameter of a cross-section of a tree trunk 1.3 meters above the ground corrects or is backfilled to S according to age, tree species, bottom class number by tree0Phase
The bottom class answered；
2) geographic area counts:
A, using the range statistics function of ArcGIS software to T0With S0Zhong Ge bottom class counts SOIL DISTRIBUTION data by graphics field
D3, bioclimate variable data D4, solar radiation quantity D5-2, terrain information D5-3Each achievement data, in T0With S0In increase newly phase
The field answered will be in the corresponding newly-built field of value filling；
B, data D is fostered in management2Forest ownership is used in the case where lacking Operation Measures data；
C, basin classification data D5-1According to each basin distributed data that hydrological analysis tool obtains, by by west to east in region,
From north orientation southing row Unified number, by number result assignment to T0With S0In the basin field of data set；
3) hierarchical data sets are established:
Using the selection function in ArcGIS software by tree species, age, forest ownership by T0With S0Age-based level layering, is established
T1With S1Data set, while verifying survey data if there is previous year field operation, then this partial data is accordingly also added to S1
In data set；
4) model training:
For S1Each tree species, annual bottom class data Layer, carry out model training and accuracy checking respectively, save and meet essence in data set
It spends desired model and calculates corresponding T with the model1Bottom class's figure layer data in data set carry out step 6) model and calculate, right
The amendment of step 5) model accuracy is carried out in the model for being unsatisfactory for required precision；
5) model accuracy is corrected:
From the history bottom class distributed data for the tree species for being unsatisfactory for modeling accuracy requirement, random or system extracts 40% bottom class's number
According to field operation supplement survey data are generated, tree height, diameter of a cross-section of a tree trunk 1.3 meters above the ground field operation supplement survey are carried out, survey data is back to S1In data set,
Then carry out step 4) model training to model accuracy again and reach requirement.The partial data will also enter next year mould simultaneously
In type training；
6) model calculates:
Applying step 4) model training model result to T1Corresponding tree species in data set, bottom class's data corresponding to the age into
Row calculates, and is checked abnormal data and is corrected；
7) performance data is checked:
Check data processing precision, the data that precision is unsatisfactory for requiring then carry out the amendment of step 5) model accuracy, and precision satisfaction is wanted
It asks, the Update order of the predict function of result application model or SQL statement is updated to T0, then generate final year
Update performance data T3。
2. a kind of mean height and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method for taking neighborhood and geographical difference into account according to claim 1,
Be characterized in that: the hierarchical data sets referred in the step 3) can may be virtual figure layer for solid data figure layer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810829343.7A CN109002621B (en) | 2018-07-25 | 2018-07-25 | Forest stand average height and breast diameter calculation method considering neighborhood and geographic difference |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810829343.7A CN109002621B (en) | 2018-07-25 | 2018-07-25 | Forest stand average height and breast diameter calculation method considering neighborhood and geographic difference |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109002621A true CN109002621A (en) | 2018-12-14 |
| CN109002621B CN109002621B (en) | 2022-11-15 |
Family
ID=64598030
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810829343.7A Active CN109002621B (en) | 2018-07-25 | 2018-07-25 | Forest stand average height and breast diameter calculation method considering neighborhood and geographic difference |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109002621B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110059561A (en) * | 2019-03-18 | 2019-07-26 | 北京木业邦科技有限公司 | A kind of method and device of Forest Growth modeling and prediction based on machine learning |
| CN110470258A (en) * | 2019-09-17 | 2019-11-19 | 中国林业科学研究院资源昆虫研究所 | A kind of Pinus yunnan Stand increment determines method and system |
| CN110853699A (en) * | 2019-10-30 | 2020-02-28 | 北京林业大学 | Method for establishing single-tree growth model under large-area condition |
| CN111143632A (en) * | 2019-12-19 | 2020-05-12 | 北京林业大学 | Same-resource cluster classification system based on plant updating mode |
| CN111191386A (en) * | 2020-01-20 | 2020-05-22 | 浙江省森林资源监测中心 | Multi-scale compatible forest tree annual growth model building method |
| CN111667879A (en) * | 2020-07-17 | 2020-09-15 | 浙江农林大学 | Moso bamboo forest stand structure optimization method based on minimum entropy |
| CN112257214A (en) * | 2019-07-02 | 2021-01-22 | 北京林业大学 | Logarithmic difference growth prediction method for 36 kinds of Chinese arbor species |
| CN112863596A (en) * | 2021-01-26 | 2021-05-28 | 东北农业大学 | Method for constructing and predicting aerobic fermentation humification degree prediction model based on bacterial community data |
| CN113159977A (en) * | 2020-01-07 | 2021-07-23 | 北京林业大学 | Dominant tree species diameter step distribution method based on growth model |
| CN115862304A (en) * | 2023-03-03 | 2023-03-28 | 吉林省林业科学研究院 | Intelligent early warning system and method for ecological restoration of degenerated natural secondary forest |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009015069A1 (en) * | 2007-07-20 | 2009-01-29 | The Trustees Of Columbia University In The City Of New York | Methods and systems of evaluating forest management and harvesting schemes |
| CN101876544A (en) * | 2009-12-15 | 2010-11-03 | 中国林业科学研究院资源信息研究所 | Forestry measurement device |
| CN103162635A (en) * | 2013-02-21 | 2013-06-19 | 北京林业大学 | Method for measuring diameter at breast height and tree height with camera |
| CN104794324A (en) * | 2015-02-05 | 2015-07-22 | 北京林业大学 | Bucking method using 3D angle gauge for forest clear cutting by compartment |
-
2018
- 2018-07-25 CN CN201810829343.7A patent/CN109002621B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009015069A1 (en) * | 2007-07-20 | 2009-01-29 | The Trustees Of Columbia University In The City Of New York | Methods and systems of evaluating forest management and harvesting schemes |
| CN101876544A (en) * | 2009-12-15 | 2010-11-03 | 中国林业科学研究院资源信息研究所 | Forestry measurement device |
| CN103162635A (en) * | 2013-02-21 | 2013-06-19 | 北京林业大学 | Method for measuring diameter at breast height and tree height with camera |
| CN104794324A (en) * | 2015-02-05 | 2015-07-22 | 北京林业大学 | Bucking method using 3D angle gauge for forest clear cutting by compartment |
Non-Patent Citations (1)
| Title |
|---|
| 冷燕等: "思茅松天然林林分胸径多样性的区域差异研究", 《西南林业大学学报(自然科学)》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110059561A (en) * | 2019-03-18 | 2019-07-26 | 北京木业邦科技有限公司 | A kind of method and device of Forest Growth modeling and prediction based on machine learning |
| CN112257214A (en) * | 2019-07-02 | 2021-01-22 | 北京林业大学 | Logarithmic difference growth prediction method for 36 kinds of Chinese arbor species |
| CN110470258A (en) * | 2019-09-17 | 2019-11-19 | 中国林业科学研究院资源昆虫研究所 | A kind of Pinus yunnan Stand increment determines method and system |
| CN110853699A (en) * | 2019-10-30 | 2020-02-28 | 北京林业大学 | Method for establishing single-tree growth model under large-area condition |
| CN111143632A (en) * | 2019-12-19 | 2020-05-12 | 北京林业大学 | Same-resource cluster classification system based on plant updating mode |
| CN111143632B (en) * | 2019-12-19 | 2023-04-18 | 北京林业大学 | Same-resource cluster classification system based on plant updating mode |
| CN113159977A (en) * | 2020-01-07 | 2021-07-23 | 北京林业大学 | Dominant tree species diameter step distribution method based on growth model |
| CN113159977B (en) * | 2020-01-07 | 2023-09-29 | 北京林业大学 | Dominant tree species diameter order distribution method based on growth model |
| CN111191386A (en) * | 2020-01-20 | 2020-05-22 | 浙江省森林资源监测中心 | Multi-scale compatible forest tree annual growth model building method |
| CN111191386B (en) * | 2020-01-20 | 2023-04-14 | 浙江省森林资源监测中心 | Multi-scale compatible forest tree annual growth model building method |
| CN111667879A (en) * | 2020-07-17 | 2020-09-15 | 浙江农林大学 | Moso bamboo forest stand structure optimization method based on minimum entropy |
| CN111667879B (en) * | 2020-07-17 | 2023-04-28 | 浙江农林大学 | Minimum entropy-based phyllostachys pubescens stand structure optimization method |
| CN112863596A (en) * | 2021-01-26 | 2021-05-28 | 东北农业大学 | Method for constructing and predicting aerobic fermentation humification degree prediction model based on bacterial community data |
| CN112863596B (en) * | 2021-01-26 | 2022-04-15 | 东北农业大学 | Method for constructing and predicting aerobic fermentation humification degree prediction model based on bacterial community data |
| CN115862304A (en) * | 2023-03-03 | 2023-03-28 | 吉林省林业科学研究院 | Intelligent early warning system and method for ecological restoration of degenerated natural secondary forest |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109002621B (en) | 2022-11-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109002621A (en) | A kind of mean height and diameter of a cross-section of a tree trunk 1.3 meters above the ground calculation method for taking neighborhood and geographical difference into account | |
| Chave | Study of structural, successional and spatial patterns in tropical rain forests using TROLL, a spatially explicit forest model | |
| KR100750749B1 (en) | Method for Quantifying of Plant Resources using GIS | |
| CN101916337B (en) | Method for dynamically predicting potential productivity of paddy rice based on geographical information system | |
| Almeida et al. | Needs and opportunities for using a process-based productivity model as a practical tool in Eucalyptus plantations | |
| Xenakis et al. | Sensitivity and uncertainty analysis from a coupled 3-PG and soil organic matter decomposition model | |
| Dye | Modelling growth and water use in four Pinus patula stands with the 3-PG model | |
| Mulia et al. | Reconciling root plasticity and architectural ground rules in tree root growth models with voxel automata | |
| Fauset et al. | Individual-based modeling of Amazon forests suggests that climate controls productivity while traits control demography | |
| Parikh | Sustainable Development of Agriculture | |
| De Wergifosse et al. | Simulating tree growth response to climate change in structurally diverse oak and beech forests | |
| Zhang et al. | A spatial-explicit dynamic vegetation model that couples carbon, water, and nitrogen processes for arid and semiarid ecosystems | |
| Tipping et al. | Measured estimates of semi-natural terrestrial NPP in Great Britain: comparison with modelled values, and dependence on atmospheric nitrogen deposition | |
| Fomin et al. | Genetic forest typology as a scientific and methodological basis for environmental studies and forest management | |
| Petit-Cailleux et al. | Tree mortality risks under climate change in Europe: Assessment of silviculture practices and genetic conservation networks | |
| Zhou et al. | Predicting forest growth and yield in northeastern Ontario using the process-based model of TRIPLEX1. 0 | |
| Biali et al. | Application of GIS technique in land evaluation for agricultural uses. | |
| Ali | Modelling of biomass production potential of poplar in short rotation plantations on agricultural lands of Saxony, Germany | |
| Iravani et al. | Assessing the provision of carbon-related ecosystem services across a range of temperate grassland systems in western Canada | |
| Causarano et al. | Improved modeling of soil organic carbon in a semiarid region of Central East Kazakhstan using EPIC | |
| Coughenour | Savanna—landscape and regional ecosystem model: model description | |
| Mansourifar et al. | Optimization crops pattern in variable field ownership | |
| Ooba et al. | Biogeochemical forest model for evaluation of ecosystem services (BGC-ES) and its application in the Ise Bay basin | |
| Mithra et al. | Computer simulation of cassava growth | |
| Hack-ten Broeke et al. | The leaching potential as a land quality of two Dutch soils under current and potential management conditions |
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 |