| Forest ecosystem is the largest carbon pool among the terrestrial ecosystems and the mainbody of global natural resources and ecological environment construction.Dynamic monitoring of forest type and forest aboveground biomass is an important foundation for assessing forest succession trends and achieving sustainable forest management purpose.Landsat remote sensing image has relatively high temporal,spectral and spatial resolutions,particularly,its historical archival data for nearly 50 years,enabling it possible to monitor forest change at a landscape scale in a long-term manner based on the Landsat images covering the National Nature Reserve of Yaluoping spanning from 1987 to 2020,a total of 41 Landsat scenes,the forest cover data set for the reserve was developed by using the vegetation change tracker(VCT)model,and the mapping results were validated by using visual interpretations.On this basis,a vegetation index NDVI＿DR considering the phenological characteristics of different forest types was proposed to distinguish coniferous forest from broad-leaved forest,followed by an analysis of the spatial-temporal patterns of forest type changes.Furthermore,multiple modeling factors including remote sensing spectral signatures,vegetation indices,textural measures derived from gray level occurrence matrix and wavelet transform analysis and topographic attributes were developed to model forest aboveground biomass(AGB)in the study area from 1987 to 2020.Ultimately,the driving factors responsible for the observed forest cover and AGB changes were analyzed to provide references or suggestions for scientific and reasonable protection and development of the reserve.The main results were as follows:(1)In the long-term forest cover information extraction,compared with the object-oriented decision tree classification algorithm,VCT model had higher extraction accuracy,with an overall accuracy of over 90%,kappa coefficient above 0.6.This may be attributed to the fact that VCT algorithm is able to adequately make use of the long time series of spectral signatures in the Landsat images stack.The forest area of the National Nature Reserve in Yaoluoping was estimated at 117.08 km~2 in 1987,116.79 km~2 in 1992,109.81 km~2 in 1997,110.68 km~2 in 2002,112.63 km~2 in 2007,113.53 km~2 in 2011,111.64 km~2 in 2017 and 114.58 km~2 in 2020respectively.The forest area of the reserve decreased first and then increased,the forest cover in1997 was the lowest,and after 1997 the overall trend of forest change was upward.Forest fire,landslides and freezing damage and excessive harvesting in the earlier stage were the major reasons for the reduction of forest area in the reserve.In the later period,the policy of Land conservation and Grain for Green,and the establishment of management and operation system of"Integration of district and township,community participation management"contributed to a continuous improvement of forest resources.The institutionalized construction of resource protection and management in the reserve has stepped up a new stage,thus the forest resources in the reserve are fully protected,and forest damage has been greatly restrained.(2)Broad leaved forest was dominant in area in Yaoluoping National Nature Reserve,and coniferous forest was mainly distributed in the relatively high altitude areas in the south of the study area,and a small amount of coniferous forest was also distributed near the low altitude valleys.The area of coniferous forest was estimated at 10.87,8.34,8.28,10.23,7.30,5.34,8.82and 11.04 km~2 respectively for the eight mapping years;the broad-leaved forest area was at106.21,108.45,101.53,100.45,105.33,108.19,102.82 and 103.54 km~2 respectively.The area of coniferous forest decreased in the southwest,northwest and southeast of the reserve during the period 1992 to 2002.The area of coniferous forest increased mainly in the valley of Baojiahe River Basin,during the period 1992-2007.The forest condition in the core area was stable for a long time,and the forest change areas were mainly concentrated in the experimental area.(3)The stochastic gradient boosting algorithm was used to map AGB according to the mapped forest types,separating coniferous forest from broad-leaved forest.The independent validation process produced a R~2 and RMSE at 0.63 and 11.18 t/ha for broadleaved forest type and at 0.61 and 14.267 for coniferous forest types respectively,which indicated that the model had high biomass estimation accuracy.The average biomass of the eight years retrieved by SGB model was estimated at 57.37,61.56,64.38,70.14,73.21,75.52,76.23 and 78.85 t/ha respectively.There was a continuous increasing trend in average AGB in Yaoluoping National Nature Reserve and the fastest growth period was in the period 1992-2002.For different forest types,the total biomass of coniferous forest was much lower than that of broad-leaved forest due to its extremely small distribution areas in the years,but its average AGB was higher than that of broad-leaved forest.Forest harvesting and natural disasters are the main reasons for the reduction of forest area and AGB in the study area.In order to alleviate the contradiction between economic development and resource protection in the nature reserve and to achieve the goal of dual protection of natural resources and community interests,the reserve should implement differentiated ecological compensation policies,differentiated financial support policies and differentiated Township assessment methods. |
