Monitoring Of Vegetation Phenology Based On MODIS Data And Its Response To Climate Change In Tibetan Plateau,China

As adaptive character to the periodic changes of the external environment,phenology is an important indicator in ecosystem change.Therefore,the inversion of phenology and monitoring of its change are helpful to understand the impact mechanism of global climate and environmental change on ecosystems.The rapid development of remote sensing technology can remedy the limitations of traditional ground surveys,and is widely used in long-term monitoring of vegetation at regional scale.At present,a great deal of research work on vegetation phenological inversion has been carried out.However,the verification of extracted results is often based on the ground observation data,but less attention has been paid to quantitative evaluation on spatial representativeness of the results.Besides,most of the inversions are based on single vegetation index and single inversion algorithm,seldom on different vegetation index and inversion algorithm according to different vegetation types.Therefore,it is necessary to explore appropriate vegetation index and inversion algorithm to carry out the inversion of vegetation phenology information according to the characteristics of vegetation types.In this study,the Tibetan Plateau(TP)is selected as the study area,which is sensitive to global change.Firstly,long time series vegetation indexes are reconstructed using Savitzky-Golay Filter.Then,six inversion methods(N-SA/N-STL/E-SA/ESTL/P-SA/P-STL)are used to extracted the phenological parameters,which are generated by combination of reconstructed vegetation index data(NDVI/EVI/NDPI)and different inversion algorithms(Seasonal amplitude/Seasonal Trend decomposition by LOESS,STL).Spatio-temporal characteristics and changes of start of season(SOS),end of Season(EOS)are then analyzed.Finally,the impacts of climate change on vegetation phenology are further discussed.The main achievements of the research are as follows:(1)Applicability assessment of phenological inversion methods in TPPhenological parameters of TP extracted by six methods based on the combination of different vegetation indices data and inversion algorithms are evaluated and compared,and the applicability of each method in different vegetation types is discussed.The results show that the combination method(N-SA)based on SA algorithm and NDVI data is superior to other methods in in phenological parameter inversion for grassland,shrub and savannas.For deciduous broad-leaved forests and crops,the combination of NDPI data and SA algorithm(P-SA)has better temporal and spatial stability than others,which is more reliable.(2)Spatial-temporal characteristics vegetation phenology in TP during 2001 to 2017.SOS mainly occurs between 80 and 200 Julian days and occurs in delay with the altitude increase.EOS mainly occurs between 250 and 310 Julian days,EOS at the south is late than that at the north.(3)Response of vegetation phenology to climate change.The SOS advanced by 3day/10 yr during 2001 and 2017,and exhibited spatial differences.Driven by increasing precipitation,the area ratio of SOS in advance in Gannan ecological region was higher than other regions.SOS delayed in the western and southern ecological regions due to the decrease of surface temperature and precipitation in winter.The EOS advanced significantly by 1day/10 yr during 2001 and 2017.The increase of surface temperature contributes to the advance of EOS in the south and west ecological region.EOS advanced obviously due to the significant decrease of surface temperature in summer and increasing precipitation in autumn.