Distribution Change And Analysis Of Different Grassland Types In Hulunber Grassland

Grassland has important economic and ecological functions as a broad type of land cover.The classification of grassland is not only an important method for people to understand grassland,but also a theoretical basis for rational utilization,effective protection and cultivation and scientific development.Due to the characteristics of grassland,such as relatively small difference of vegetation with different grassland types,relatively large spatial heterogeneity of the same grassland type,relatively complex vegetation composition and relatively many disturbing factors,grassland classification based on remote sensing technology mostly classifies grassland as one type,and it is difficult to classify different grassland types in high-precision and large area.The fine classification of grassland types in Hulunber(1992 and 2015)was obtained by object-based image analysis(OBIA)and machine learning based on the analysis of the influence of separability with different phases and data(remote sensing data and auxiliary data).The impacts of dynamic change of different grassland types were studied by using climatic data,statistical data and remote sensing data which based on the quantitative analysis of the distribution,landscape pattern and change of different grassland types and the process of transformation of cropland,forestland and building land,etc.The main conclusions were as follows:(1)The separability of different grassland and vegetation types varied at different times.The information in the growing season was conducive for distinguishing temperate meadow steppe from temperate steppe,and it was favourable for extracting lowland meadow and swamp information in the non-growing season.The DEM and NDVI also effectively reduced the number of image segmentation objects(15.35%)and improved the segmentation effects.(2)A few main variables played a major role in the classification,the classification accuracy of five main feature variables increased from 44.79%to 87.43%,spectral and textural features were more important than geometric features in this study.Support vector machine and random forest classifier can classify different grassland types appropriately,and the classification accuracy were 94.41%and91.78%respectively.(3)The grassland area in the study area decreased by 4531.73 km~2 in 2015 with a rate of5.96%.The temperate steppe had the largest distribution area,with the area of 3989.15 km~2 and42697.57 km~2 in 1992 and 2015.The area of temperate meadow steppe were 19617.52 km~2 and13982.73 km~2,and the area of lowland meadow were 1404.16 km~2 and 12967.11 km~2 in 1992 and 2015.The distribution area of mountain meadow and swamp were relatively small,with the proportion less than 5%.The area of temperate meadow steppe,lowland meadow,mountain meadow,swamp decreased,and the area of temperate steppe increased.The area of temperate meadow steppe decreased greatest by5634.79 km~2(28.72%),31.21%of the reduced area of temperate meadow steppe was converted to cropland and building land,64.17%of lowland meadow,27.77%of mountain meadow,and only 0.23%of swamp.The area of temperate steppe increased by 2837.90 km~2(7.12%),the reduced area caused by cropland and building land was 315.49 km~2,accounting for 11.12%of the total change area.(4)The impacts of climate change and human activities on the distribution of different grassland types were different,the impacts of human activities were fragmented in space,reversible in time and easy to recover,the impacts of climate change were tendentious and difficult to recover,which has guiding significance for understanding the impacts of climate change and human activities on grassland ecosystems,specifically as follows:1)the climate became dry as a whole,the change direction of potential distribution of grassland types obtained by climatic indicators was consistent with the real change,but it magnified the impact of climate on vegetation,2)the increased cropland area in the study area mainly came from temperate meadow steppe,lowland meadow,mountain meadow and temperate steppe,accounting for 98.97%of the total source,and the overall cropland area fluctuated and rose,3)the increased forestland area in the study area came from temperate meadow steppe,lowland meadow,mountain meadow,accounted for 97.30%of the total source,and the area of forestland showed an increasing trend as a whole,4)87.85%of the increased building land area in the study area originated from temperate meadow steppe,lowland meadow,and temperate steppe,and the population and GDP showed an increasing trend as a whole,5)the impact of grazing on grassland were multifaceted,and the number of livestock increased as a whole.