Study On The Simulation And Influence Mechanism Of Regional Vegetation Phenology Based On Multi-source Data

Vegetation phenology is one of the key source of information for mapping,managing,and monitoring the terrestrial ecosystem from local to global scales.Changes in vegetation phenology reflect the biosphere's dynamic response to climate change.Furthermore,changes in vegetation phenology can affect the energy exchange,hydrology,and carbon uptake of the ecosystem through changes in vegetation transpiration and photosynthesis,as well as vegetation productivity.Climate change over the last several decades has modified the dates of plant flowering,leaf emergence,growth stages,and senescence,collectively termed phenology,over Earth's terrestrial surface with substantial ecological and environmental consequences.Both terrestrial observations and model simulations have found that air temperature has a positive influence on the onset of plant growth in the Northern Hemisphere?NH?.In contrast to extensive research efforts focused on spring phenology,autumn phenology is more challenging to understand,and has not received sufficient attention,while also serving as an important indicator of changing foliar physiological properties.Further,autumn phenology may be as important as spring in regulating the interannual variability of carbon balance.We calculated the end of growing season?EOS?through midpoint and double logistic algorithm based on GIMMS NDVI3g dataset and MODIS vegetation index product over the NH.We analyzed the spatial and temporal variation of EOS,and its correlation with preseason temperature and precipitation.We also investigated the responses of autumn phenology to daytime(T_day)and nighttime(T_night)temperature from ground sites and remote sensing observations.The Tibetan Plateau?TP?has a typical alpine vegetation ecosystem and is rich of snow resources.With global warming,the snow of the TP has undergone significant changes that will inevitably affect the growth of alpine vegetation,but observed evidence of such interaction is limited.In particular,a comprehensive understanding of the responses of alpine vegetation growth to snow cover variability is still not well characterized on TP region.To investigate this,we calculated three indicators,the start?SOS?and length?LOS?of growing season,and the maximum of normalized difference vegetation index(NDVI_max)as proxies of vegetation growth dynamics from the Moderate Resolution Imaging Spectroradiometer?MODIS?data for 2000-2015.Snow cover duration?SCD?and melt?SCM?dates were also extracted during the same time frame from the combination of MODIS and the Interactive Multi-sensor Snow and Ice Mapping System?IMS?data.The main conclusions are as follows:?1?There are differences in the derived EOS by different vegetation index and different methods.In generally,EOS derived from NDVI is later than EVI,while EOS calculated by double logistic algorithm is earlier than midpoint algorithm.The EOS from MODIS data would also be earlier than that from GIMMS NDVI3g dataset.However,the interannual variability was consistent for the overlapping period of both data sets?R²=0.63?.?2?The preseason temperature and precipitation would affect EOS.Nevertheless,the response of EOS to temperature and precipitation varied in different climate zones and vegetation types.In the boreal cold region,EOS showed significant relationship with preseason temperature.The negative correlation was found before 2000,while the positive correlation was observed from that year.?3?We show that rising preseason T_day and T_night had contrasting effects on autumn LSD in the Northern Hemisphere?NH?.If higher Tday leads to an earlier or later LSD,an increase in Tnight drives LSD to occur in the opposite direction.Contrasting impacts of daytime and nighttime warming on drought stress may be the underlying mechanism for this opposing relationship.?4?A new phenological model considering these opposite effects substantially improved autumn phenology modeling accuracy,and predicted an overall earlier autumn phenology by the end of this century compared with traditional projections.Our results change the paradigm of prolonged growth by higher autumn temperatures,and that leaf senescence in the NH would begin earlier than currently expected,causing a positive feedback on climate change.?5?We found that the snow cover phenology had a strong control on alpine vegetation growth dynamics.Furthermore,the responses of SOS,LOS and NDVI_max to snow cover phenology varied among different biomes,eco-geographical zones,and temperature and precipitation gradients.The alpine steppes showed a much stronger negative correlation between SOS and SCD,and also a more evidently positive relationship between LOS and SCD than other types,indicating a longer SCD would lead to an earlier SOS and longer LOS.Most areas showed positive correlation between SOS and SCM,while a contrary response was also found in the warm but drier areas.Both SCD and SCM showed positive correlations with NDVI_max,but the relationship became weaker with the increase of precipitation.Our findings provided strong evidence between vegetation growth and snow cover phenology,and changes in snow cover should be also considered when analyzing alpine vegetation growth dynamics in future.