Research Of Household Scale Grass Yield Monitoring Based On The Neural Network In Xilinguole

Grassland play an important role in the ecological environment.In terms of ecological benefits,grassland resources can help regulate climate,maintain soil moisture,reduce soil erosion and promote ecological balance.Its great environmental significance and economic benefits have aroused great attention.In recent years,with the continuous development of society,people's demand for grassland resources is also increasing.Irrational utilization of grassland resources has become the main reason for grassland ecological environment.Grassland is a necessary condition for herdsmen to survive.If we do not make rational use of it,besides being unable to meet the daily life requirements of herdsmen,the destruction of the ecological environment will also threaten the production and life of the grassland people in the future.The grass yield of grassland as the most basic production index,which reflects the dynamic changes of grassland resources,but also to the grass and livestock,livestock balance the basis.The accurate estimation of grass quantity in grass land has become the most important prerequisite for the rational and sustainable utilization of grassland resources,and it is also the guarantee for the healthy development of grassland ecosystem.The grass volume estimation methods usually include traditional statistical methods such as quadrat statistics and empirical estimation.However,these methods are difficult to monitor grassland dynamically on a regional scale.With the development of remote sensing technology,estimating grass production become the common estimation method using remote sensing data,compared with the traditional research methods of ecology,spatial remote sensing estimation method can obtain detailed vegetation,is time-saving,yield a wide range of large,fast speed.The vegetation index is an important method of remote sensing monitoring grassland production,of which MODIS-NDVI is most widely used,and the temporal and spatial variation characteristics of large scale growth can reflect the vegetation,but the remote sensing method in small scale changes reflect the grass yield and the lack of.Taking Xilinguole grassland as an example,using BP neural network toolbox of Matlab,select the MODIS satellite remote sensing data to calculate the normalized difference vegetation index NDVI,combined with field survey data,the establishment of neural network yield estimation model,and through the pasture ownership data have been obtained that the superposition of animal husbandry households grassland boundaries and remote sensing image analysis,calculation of household scale grassland the amount of grass,household scale grassland utilization evaluation.This paper draws the following conclusions:(1)On the East Ujimqin banner as an example,by using the neural network model for the first time accurately estimated 30 herdsmen pasture grass yield in different grassland types in 2009,comparing with the calculation results of NN estimation method,the current Xilinguole Meng estimation model and two linear yield estimation model on 30 herdsmen pasture grass yield the same estimation results the comparison study found that grassland types belong to temperate steppe type 10 grass belongs to the excessive use of state,8 grassland types belong to the temperate meadow steppe type waste grassland resources situation.(2)It is found that the yield of different scale pasture grassland types under the same household units are different.(3)The determination coefficient of the model is R~2=0.64 and the root mean square error RMSE=2200kg/hm~2.Compared with other estimation model and estimation method of Xilinguole Meng current estimation accuracy and good estimation performance with high precision and high efficiency,achieve high yield estimation to provide new ideas and methods for rational utilization and protection of grassland.(4)Although in Xilinguole grassland as the research area,but the model is universal,work yield estimation can be applied to other areas.(5)The model of high portability,can use this model through the.NET platform,using embedded Matlab engine function library interface technology and programming to the relevant module of big data platform"Xilinguole ecological protection and construction of service platform",to assist in the completion of grass yield calculation,livestock balance assessment and ecological compensation awards and so on work,provide data support,improve work efficiency.