Study On Ecological Evaluation,Monitoring Of Meadow Steppe In Inner Mongolia Based On Remote Sensing

Grassland degradation in arid areas has become a serious ecologicalproblem,and the development of grass-livestock relations and animal husbandry in pastoral areas of Inner Mongolia is facing serious challenges.With the development of big data technology and cloud computing platforms have changed the research model of geology and ecology.The meadow grassland is widely distributed in Inner Mongolia and is an important grassland resource and landscape community in China.In the meadow grassland,by monetizing the ecosystem service value,quantifying the ecological risk level of regional landscape and monitoring the key ecological parameters of grasslands,it provides data reference for the relevant policies of fine grassland conservation and utilization.It is of great significance to China's ecological sustainable development and ecological civilization construction.In this paper,the Inner Mongolia Meadows and Grasslands area is selected as the study object,and the study is carried out in multiple dimensions on the basis of big data cloud platform in a data-driven manner.On the basis of Google Earth Engine integration,the ground data such as ecological parameters,meteorological observation data and climatic observation data from sample sites,combined with remote sensing satellite data such as Landsat 8,Sentinel-2 and MODIS to build a sky-ground integration evaluation method.Geographic grid computing,machine learning,mixed pixel decomposition,empirical model construction and other methods are used to carry out grassland ecological evaluation and monitoring research.The research focuses on the analysis of the spatio-temporal distribution of land cover in the meadow and grassland area,the hierarchical evaluation of ecological risk changes in the landscape,the spatio-temporal adjustment of the value of ecosystem services and monetary value accounting,the inversion of the basic ecological parameters of the meadow and grassland area,the estimation of grazing yield and livestock carrying capacity in the meadow and grassland degradation detection,and other multi-dimensional research results show that:(1)The grassland in Inner Mongolia meadows and grasslands has been restored obviously,and the ecological risks are controllable.The land-use characteristics of the Meadow and Grassland area are similar to those of the whole Inner Mongolia region,as shown by the gradual increase in the area of grassland and the faster recovery of the area of grassland in the Meadow and Grassland area compared to the whole region.But there are no large population gathering areas,no large-scale river and lake wetlands,the future should continue to pay attention to the protection of natural grasslands,resolutely inhibit natural grasslands are reclaimed for agricultural land.Landscape ecological risk levels in the Meadowlands are generally low,with individual areas having moderate or high levels of risk that require warning.(2)The value of ecosystem services is increasing year by year,of which the largest contribution is grassland,in which the climate regulation capacity has produced the greatest value.The most important ecosystem service function in the Meadowlands is the regulating function,and the climate regulating function is a top priority.Its value contribution is followed by ecosystems such as forests and farmland,and the timing is mainly the growing season from April to October.This shows that the protection of meadows and grassland areas in Inner Mongolia should be given top priority in ecological construction.(3)Grassland cover and growth in the meadow grassland area is trending upwards,and the greening monitoring reflects the true situation on the ground.The quantitative inversion and qualitative analysis of the pasture coverage and growth have achieved good results to satisfy the monitoring needs.Compared with the linear regression model,the results of applying the mixed pixel decomposition model to invert the meadow grassland coverage are more consistent with the actual situation.One reason is that the linear regression model is simple and cannot fully exploit the correlation between bands and coverage,and another reason is that the mixed pixel decomposition can fully conserve the vegetation spectral features of the grassland.Analysis of the correlation between coverage and meteorological environmental factors found that rainfall is an important factor affecting pasture coverage.The greening monitoring results have a strong correlation with the ground henology collection data and accurately express the chronological and spatial distribution characteristics of greening.(4)Pasture yields are increasing year by year and precipitation is a decisive factor affecting yield.Using the original bands and related derivative bands of Landsat 8,Landsat 7,Sentinel-2,MODIS images respectively to construct a unitary and multivariate statistical model.Through the model verification accuracy comparison analysis,the multivariate linear model based on Landsat 8 data has the best accuracy.Using this model for spatio-temporal analysis and mapping,it was found that the yield of the meadow grassland area in 2013-2019 is generally increasing,which corresponds to the monitoring results of the coverage pixel dichotomy model,and is also consistent with the results of land utilization research based on MODIS data products.The consistency of grassland parameters obtained different data sources and different algorithms can demonstrate the reliability of the results.The research combined the production capacity for the analysis of animal carrying capacity and thematic mapping,and obtained the thematic map of animal carrying capacity for the warm and cold seasons of Inner Mongolia in 2019.Analysis of the correlation between yield and meteorological environmental factors found that rainfall in the study area is an important factor affecting yield.(5)The application of random forest model inversion grassland desertification has high accuracy,and it can be extended to inversion of grassland desertification in the whole region.Based on the grassland desertification ground sampling data,random forest model and mixed pixel dichotomy model were constructed respectively,and using the model inversion results analyzed the spatial and temporal distribution characteristics of grassland desertification in the meadow grassland.Due to the large error and uncertainty of the end-cell extraction of mixed pixels,resulting in the accuracy of desertification and inversion of mixed pixels is low,while the accuracy based on random forest model can obtain a high accuracy,with an overall accuracy of 0.6383 and a kappa coefficient of 0.4495.Random forest model of grassland desertification can be used for full domain time series spatial mapping.