| With the development of social economy and the enhancement of environmental protection awareness,the importance of natural mixed forest in economic development and ecological protection is increasing.At present,most of the research on stand is pure forest,and the research on the growth and harvest of natural mixed forest is still in its infancy.There is no unified modeling method,so it is an urgent research topic to establish a suitable growth prediction model.The section area of the stand is not only an important part of the growth and harvest prediction model,but also an important index to evaluate the quality of the site.At present,most of the researches on the growth model of cross-sectional area take the stand as the basic modeling unit,without considering the influence of the forest layer on the growth of cross-sectional area.In order to simulate the growth process of multi-layer forest more objectively with the growth model,it is necessary to carry out the research on the growth model of cross-sectional area based on the division of forest layer.In this paper,51 fixed plots of Quercus natural secondary forest were set up in Lutou forest farm of Hunan Province.Taking this as the object,the influence of the forest layer of secondary forest on the growth of the sectional area was studied,and the growth model of the sectional area compatible with the forest layer was constructed,which provided the theoretical basis for the management of Quercus secondary forest in Hunan Province.The specific research contents are as follows:(1)The optimal forest layer division method for oak secondary forest was determined.Based on the fixed plots of 51 natural oak secondary forests in Hunan Province,this paper uses the most commonly used clustering methods after extensive understanding of various forest layer division methods at home and abroad.The three methods of whole tree height clustering,IUFRO standard and TRSRAT divide the forest layer,and test and analyze the results.The three forest layer division methods all meet the division requirements(2)A basal area model based on forest layer division was established.According to the growth law of the area of the forest and its research status,the common theoretical growth equation form is mainly selected in this paper.The stand density index(SDI)represents the stand density,and the average height of dominant trees(HT)represents the site quality to construct the growth model of the area of the forest.The Schumacher model containing age(T),mean height of dominant trees(HT)and stand density index(SDI)was finally determined as the optimal basic model,with the determination coefficient R2 of 0.9277,mean absolute error(MAE)of 2.6186 and total relative error(TRE)of less than±3%.Based on this,the basal area growth model based on forest layer division is simulated.The simulated basal area growth models of the whole forest and different forest layers have high fitting degree.The determination coefficient(R2)of the area growth model of different forest layers after stratification is 0.9521?0.9863,the average absolute error(MAE)is 0.4511?1.2679,and the total relative error(TRE)is not more than±3%.The total stand model after stratification was effectively improved by adding the prediction results of the basal area growth model constructed by the three methods after stratification and comparing with the prediction results of the basal area growth model without stratification.The determination coefficient R2 was increased from 0.9277 to 0.9602?0.9721,and the average absolute error(MAE)was reduced from 2.6186 to 1.3219?1.0962,and the(TRE)was not more than±1%,which proved that there was forest layer differentiation in oak natural forests.Among them,the determination coefficient(R2)of the basal area model established after IUFRO division standard stratification is the highest,which is 0.9721,and the average absolute error(MAE)is the lowest,which is 1.0962.Therefore,IUFRO division standard is the optimal forest division method determined in this paper.The optimal division method of IUFRO forest layer is divided into standard division results,which are brought into the other five basic models for parameter fitting to verify the applicability of establishing a forest layer basal area growth model after dividing the forest layer.The basal area growth models of different forest layers in each model have high fitting degree.The predicted results of the forest layer basal area growth model are added and compared with the predicted results of the non-stratified forest basal area growth model.The fitting degree of the models obtained after stratification is effectively improved,and the determination coefficient(R2)is between 0.89 and 0.97,the average absolute error(MAE)is between 1.0962 and 3.4350,and the total relative error(TRE)is between±3%.Among them,the determination coefficient(R2)of model(M4)has the highest lifting range,and the lifting ranges of six models are M4>M3>M1>M2>M5>M6 from high to low.This shows that the forest layer has a significant influence on the estimation of the basal area,and the forest layer effect must be considered when establishing the basal area growth model.(3)A compatible basal area growth model based on forest layer effect is established.Because there is a logical relationship between the total basal area of forest and the basal area of each forest layer,the compatibility model of forest basal area based on forest layer level can be constructed to ensure that the sum of components is equal to the total logic.at the same time,it can improve the accuracy of the forest cross-sectional area model.This paper takes the results of(IUFRO)forest layer division as an example.The compatible basal area growth model was constructed by the method of nonlinear simultaneous equations,and the determination coefficient(R2)of each component of the compatible forest layer basal area growth model was more than 0.98.Compared with the non-simultaneous independent prediction model after stratification,the determination coefficient of each model is significantly improved,and the error of the simultaneous model is relatively reduced.Compared with the independent prediction model,the determination coefficient(R2)of the whole stand model increased from 0.9721 to 0.9801,the average absolute error(MAE)decreased by 16.25%,and the total relative error(TRE)did not exceed ±3%.The determination coefficient(R2)of the main forest layer model increased from 0.9739 to 0.9861,the average absolute error(MAE)decreased by 26.82%,and the total relative error(TRE)did not exceed ±3%.The determination coefficient(R2)of the secondary forest layer model has no obvious change,the average absolute error has no obvious change,and the total relative error is less than ±3%.(4)A basal area growth model based on dumb variables of forest layer effect and forest types was established.After forest layer division,the results of forest layer division and stand types were taken as dumb variables,and a unified model was established for stands with different scores.This not only reduces the workload of modeling,but also makes the growth models of different forest layers have a unified form.In this paper,taking the stratification results of the IUFRO as an example,the stratification results and stand types were taken as dumb variables,and dumb variables were added to the model parameters and their combinations respectively.According to the evaluation index of the model,the optimal parameter combination of dumb variables of forest layer basal area growth model was determined.After fitting,it is known that it is best to introduce all parameters into dumb variable R2,but the parameters are too complex to be conducive to practical application.From the point of view of simplified model and model error,the basal area growth model based on dumb variables of forest layer effect finally selects the model with dumb variables on b0,b1,b2,b3 as the optimal model,and its determination coefficient(R2)is 0.982,and the average absolute error(MAE)is 0.6219.The total relative error(TRE)is 0.0008;The basal area growth model based on dumb variables of stand type finally selects the model with dumb variables on parameters b0 and b1 as the optimal model,the determination coefficient(R2)is 0.9555,the average absolute error(MAE)is 1.0131,and the total relative error(TRE)is-2.3924,not more than ±3%.The three models of compatibility model and dumb variable model are compared and analyzed,and the optimal model is the compatibility model.(5)The study proved that the forest layer has a significant influence on the prediction of basal area and the rationality of constructing a compatible basal area growth prediction model according to the division of forest layer.the simulation results provide reference and basis for more effective management of Hunan oak natural forest. |
PDF Full Text Request