Remote Sensing Monitoring Of Post-fire Vegetation Recovery In Great Xing ’an Mountains Based On Google Earth Engine Cloud Platform

Forest fire,as one of the main disturbance factors of forest ecosystem,can change the community composition,age structure,energy flow and nutrient cycle of forest ecosystem,and thus have an important impact on regional carbon cycle,biodiversity and global climate change.Remote sensing technology has the ability of continuous spatio-temporal observation and has been widely used in forest fire monitoring and post-fire vegetation restoration.With the emergence and integrated application of cloud computing technology and time series analysis algorithm,the spatiotemporal scale monitoring of vegetation restoration process has realized denotation and expansion,which plays an important role in understanding the influence of fire disturbance on vegetation dynamics and carbon cycle at regional,national and even global scale.Great Xing’an Mountains forest zone is with high incidence of forest fires and the most serious harm in China.The forest vegetation restoration work has been carried out systematically in this region after the fire in 1987.At the same time,the implementation of national key ecological projects such as natural forest protection has accelerated the process of vegetation restoration in this region.Therefore,Mohe and Tahe county in Great Xing’an Mountains major fire zone were selected as the study area.Supported by GEE cloud platform and long time series remote sensing image data,the OTSU method and decision tree classification model were used to accurately extract the burned land in the study area for each year.Combined with vegetation index and net primary productivity product,a comprehensive analysis of vegetation restoration from1986 to 2020 was made.Based on GEE cloud platform and Land Trendr algorithm,automatic detection and extraction of forest disturbance and recovery pixels were realized,and spatio-temporal distribution pattern characteristics such as occurrence year,duration and intensity of forest disturbance and recovery were analyzed.The main conclusions are as follows:（1）Based on d NBR2 index,the burned area of Great Xing’an Mountains fire in1987 was extracted and the fire intensity was graded.The results show that the total area of the burned area is 11375.08km²,and the proportion of burned area of with different fire intensity was moderate>mild>severe.（2）Based on GEE cloud platform and burning area index,the annual burning area was automatically extracted,and the overall classification accuracy of the extracted results was 90.50%.The northwest and east regions of Tahe county and the southwest regions of Mohe county were the areas with high fire incidence.The total burned area in the study area was 140108.85 hectares and the average annual burned area was5189.22 hectares.From 1986 to 2010,forest fires occurred all the year round in the study area,and the frequency of fires decreased after 2010.（3）The trend analysis of long time series vegetation index showed that the vegetation NDVI in mild fire region returned to the original state after fire one year,the vegetation NDVI in moderate fire region returned to the original state after fire two years,and the vegetation NDVI in severe fire region returned to the original state after fire four years.Analysis of vegetation index changes at different slopes shows that vegetation with slopes between 6°to 25°is more susceptible to fire.It took 1 year for the vegetation NDVI of flat area to return to its original state,2 years for the flat slope area,4 years for the gentle slope area,2 years for the slope area and 3 years for the steep slope area.Analysis of the change trend of vegetation index of different years of fire shows that 30-35 years group has the best restoration effect,followed by 20-30 years and 10-20 years,indicating that the longer the year of fire occurs,the longer the restoration period and the better the restoration effect of vegetation.In addition,the NDVI of affected vegetation returned to the original state within 1-3 years after fire disturbance occurred in the sample plots with different fire years.（4）The overall mean NPP of vegetation was 450.98 g C·m^-2 from 2000 to 2020.In terms of spatial distribution,it is high in the south and low in the north.NPP values of vegetation range from 400 to 500 g C·m^-2in most areas.The low values area were mainly distributed in the northwest and east of the study area.In terms of temporal variation,NPP of vegetation fluctuated and increased by 0.96 g C·m^-2.NPP restoration of vegetation in moderate burned area>unburned area>mild burned area>severe burned area.NPP restoration of vegetation in steep slope area>slope area>gentle slope area>flat slope area>flat area.（5）Based on GEE cloud platform and Land Trendr algorithm,automatic detection and extraction of forest disturbance and recovery pixels were achieved,and the overall accuracy of spatial distribution data was 83.33%.The study found that a total of13206.11km² was lost from 1986 to 2020,with an average annual loss of 377.32 km².The duration of forest disturbance events mainly occurred in the range of 1-5 years,and the area of forest loss decreased with the increase of duration.The proportion of different degree of forest disturbance area was mild>moderate>severe.From 1986 to2020,the forest increased by 2438.65 km²,with an average annual increase of 686.82km².The duration of forest restoration events in the study area mainly occurred in 3-8years and 34 years,indicating that vegetation restoration is a long and continuous process.Similarly,the area of mild forest restoration.The proportion of different degree of forest recovery area was mild>moderate>severe.By comparing and analyzing the spatial and temporal distribution patterns of forest disturbance and restoration in burned and unburned area,the study found that since the Great Xing’an Mountains fire in 1987,forest disturbance and restoration events occurred more frequently in unburned area than in burned area during 1988-2020.