Land Surface Phenology Characteristics And Its Response To Extreme Climate In Xiaoxing'an Mountains

Vegetation phenology is considered as an important ecological indicator to understand how ecosystem behavior responds to environmental signals.Vegetation phonological variation is closely related to climate changes,and may have important impacts on the ecological processes of ecosystems,such as water-carbon exchange,energy flows and interactions between different species in the land surface and atmosphere.However,the response mechanism of climate change,especially extreme weather events(such as frost,high temperature,drought and heavy rain events)to the development of vegetation phenology is still poorly understood.Xiaoxing'an Mountains is an important ecological protection forest area in the temperate regions of China,and a typical sensitive area for the impact of climate change on vegetation phenology.Therefore,this study is based on the MODIS surface reflectance data,meteorological observation data,land cover classification data,and extreme climate event data in the region from 2000 to 2019,combined with remote sensing phenology data sets and ground phenology monitoring data,and aims to study the surface phenological characteristics in Xiaoxing'an Mountains in the past 20 years and their response mechanism to extreme climates,to explore the feasibility of daily-scale phenological development simulation,and expect to provide a new perspective for accurately evaluating the response mechanism of vegetation phenology to extreme climate factors on a regional scale,and to provide a scientific basis for deepening the application and practice of vegetation phenology.The conclusions obtained in this study are as follows:(1)Based on the two sets of(8-day/daily)EVI2(two-bands Enhanced Vegetation Index)time series data obtained and reconstructed from MODIS surface reflectance data,the vegetation phenological information is extracted using the double Logistic seasonal dynamic threshold method.The phenological monitoring station data and the climatic growth season data extracted from the four climate observation stations are comparable.the phenological indicators and the data from the only surface phenology monitoring station in Xiaoxing'anling and the climatic growth season data obtained from the other four observation stations are all comparable.Compared with the published phenological products,they show a significant correlation(p<0.01).Moreover,the average deviation and error based on the daily EVI2 time series data are less than the 8 day synthetic time series data.The results indicate that the surface phenology extraction method used in the study can better extract the key vegetation phenology period,and improving the time resolution of time series data can improve the estimation accuracy to a certain extent.(2)Compared with the marginal area in this region,in the central area,the start of season(SOS)and the end of season(EOS)occurred earlier,the length of season(LOS)lasted longer,and the age fluctuations were smaller.The reason for the difference is related to different vegetation types.The cultivated land SOS occurred the latest(mean value is the 161st day),and has the largest floating range(119-180 days).The EOS occurrence time of different vegetations was relatively consistent.The deciduous broad-leaved forest has the longest LOS(average value of 140 days),and cultivated land has the shortest LOS(average value of 116 days).At the most lush growth stage,the greenness of the cultivated land is the highest(the average EVI2 is 0.655),while the evergreen coniferous forest is relatively the lowest(the average is 0.587).The SOS and EOS are both predominantly advanced in Xiaoxing'an Mountains in 20 years.There are both shortened LOS areas(58.84%)and extended areas(41.15%)in the whole area.The results is within 1d/a,indicating that the interannual change trend of surface phenology is not significant.(3)The response of Xiaoxing'an Mountains to extreme temperature is significantly different in time and space.The average value of extreme low temperature from March to May had an early effect on the occurrence of SOS in the northern,southwest and southeast regions(accounting for 43.52%),mainly in the study area of cultivated land and grassland around.The extreme low temperature in September also had an early effect on the occurrence of EOS.Because after September,the temperature dropped sharply,and the low temperature accelerated the discoloration and fall of vegetation leaves.At the same time,the extreme high temperature in February and May delayed the occurrence of SOS in the northern and eastern marginal regions.Because February has not yet entered the growing season,and May has entered the growing season,and the effect of extreme high temperature has gradually weakened.The extreme high temperature in September prolonged the growing season,which mainly delayed the occurrence of EOS.(4)From 2000 to 2019,El Ni?o had no significant effect on SOS in Xiaoxing'an Mountains,but had a certain advance effect on EOS(negative anomaly was 64.41%).The La Ni?a phenomenon has a certain delay effect on SOS(positive anomaly 58.47%),and an early effect on EOS(negative anomaly 63.26%),indicating that the early spring and late autumn climate is too cold to affect the growth and development of vegetation in the study area,resulting in a shortened growing season.The response results of climate extreme indexes(ONI,SOI and PDO)and phenological period further verify this conclusion,but from the point of view of significance level,the impact of periodic changes in global climate on regional phenology is weak.(5)The constructed BHST-LSP(Bayesian Hierarchical Space-Time Model for Land Surface Phenology)model can well capture daily phenology development,and is generally better than autumn in predicting the development of ground surface phenology in spring.Spatially,compared with other regions,the central part of Xiaoxing'an Mountains has lower R~2 and higher NRMSE.This is because most of the vegetation in the northwest is deciduous broad-leaved forest,and the southern edge is mainly farmland,which is similar to the evergreen broad-leaved forest and mixed forest in the central part.In contrast,vegetation seasonality is more significant.Among them,the spring model underestimates deciduous broad-leaved forests,grasslands and mixed forests,and overestimates other vegetation types.The highest underestimation is grassland(-0.028±0.031),and the largest overestimation deviation is cultivated land(0.033±0.072).The autumn model has the smallest prediction deviation for deciduous coniferous forest(0.005±0.016)and deciduous broad-leaved forest(-0.0084±0.018),while the prediction deviation for cultivated land(0.026±0.015)and grassland(-0.015±0.026)is the largest.(6)The rapid increase in temperature during the day,the increase in precipitation,and the high frequence high temperature events can accelerate the rate of vegetation greening in spring,while reducing the rate of vegetation decay in autumn.The rapid decrease in night temperature,the frequent heavy rain events,frost events and drought events will reduce the speed of vegetation turning green in spring,and at the same time accelerating the rate of vegetation withering in autumn.Spatially,the autumn vegetation decay rate in northwest area is more sensitive to daytime temperature anomalies.The impact of frost events on vegetation in the northwest is more significant than that in the center,especially in the south,while the frequent high temperature events and abundant precipitation are more significant in the south and southeast.Deciduous broad-leaved forests and cultivated land are most sensitive to day and night temperature anomalies.The frost events have less impact on wetlands.The grassland and deciduous coniferous forests are the most sensitive to heavy rain events.The evergreen broad-leaved forest has the lowest sensitivity to accumulated precipitation and the strongest resilience to heavy rain and drought events.