| Fractional vegetation coverage(FVC)can reflect the growth of surface vegetation communities and is an important reference indicator for judging the ecological environment status.The Three-North region constitutes one of the most crucial natural resource reserves in China,and is a key area for forest and grass construction and new carbon sinks in the future.Mastering the spatial-temporal distribution and change trend of FVC is helpful for ecological protection,environmental restoration and policy formulation in the Three-North region.Remote sensing estimation has the advantages of fast,real-time,and widespread coverage,which offers technical support for the monitoring and evaluation of FVC.The mixed pixel decomposition model can obtain more accurate vegetation information and further improve the accuracy of FVC.However,the vegetation growth in the most areas of Three-North region is sparse,and the vegetation spectra obtained from remote sensing images are weak and susceptible to background factors.The spatial heterogeneity of mixed pixels also makes the selection of pure endmembers uncertain.Therefore,it is of great significance to explore the mixed pixel decomposition method applicable to the region,so as to obtain the information of vegetation coverage and its dynamic changes in a large scale and long time series.In this paper,the Three-North region of China was taken as the study area,and the MOD09A1 and MOD13A2 remote sensing image products were obtained.Abundance Sumto-one Constrained Nonnegative Matrix Factorization(ASC-NMF),L1/2 SparsityConstrained Nonnegative Matrix Factorization(L1/2-NMF),Minimum Volume Constrained Nonnegative Matrix Factorization(MVC-NMF)and Improved Constrained Nonnegative Matrix Factorization(IC-NMF)were used to estimate the FVC in the Three-North region.The accuracy of each model was compared,and the spatial distribution of FVC was visualized to explore the vegetation coverage estimation method suitable for this study area.Based on the FVC of the Three-North region from 2000 to 2020,the Theil-Sen Median trend method combined with the Mann-Kendall test was used to analyze the change trend of FVC in the past 21 years,and the Hurst exponent was used to predict the future trend of FVC.Combined with data such as natural factors and human factors,the influence of different driving factors on the spatial differentiation of FVC was quantitatively analyzed by geographical detector.The conclusions are as follows:(1)The estimation accuracy of the Improved Constrained Nonnegative Matrix Factorization model was generally better than that of the other models.The spatial distribution of FVC obtained is basically consistent with the actual distribution,resulting in the best mapping effect.The estimation results of IC-NMF model were closer to the validated values,with the highest coefficient of determination(R2),lowest root mean square error(RMSE)and lowest mean absolute error(MAE)in each year,and its residual autocorrelation was generally lower than the rest of the models.The spatial distribution results obtained by ASC-NMF,L1/2-NMF and MVC-NMF models for FVC estimation in different years varied greatly and had low stability,and there were obvious overestimations and underestimations for the FVC.The IC-NMF model appeared to estimate the high value of FVC as 1.However,compared with the above three models,the spatial distribution of FVC obtained by IC-NMF in the corresponding areas were closer to the actual situation,and the spatial distribution was relatively more reasonable and stable.(2)The FVC in the Three-North region existed with obvious territoriality,and the change trend showed the spatial distribution characteristics of overall improvement and local degradation.The FVC in Northwest China were generally between 0.00 and 0.15,which was at a low level for a long time;the FVC in North China gradually increased from north to south,ranging from 0.00 to 0.70;the FVC in Northeast China were mostly above 0.70,and the vegetation condition was favorable.From 2000 to 2020,the average value of FVC in the Three-North region ranged from 0.34 to 0.42,increasing at an average growth rate of 0.0024 a-1.Improvement and maintaining basic stability dominated the FVC over the past 21 years,in which the areas with improved,basically stable and degraded fractional vegetation coverage accounted for 50.35%,35.05%and 14.60%of the total study area,respectively.The persistent characteristics of future changes of FVC in the Three Norths region were stronger than the anti-sustained characteristics,and mainly remained stable and continued to improve,but the overall trend was weak,and there was a possibility of reversal in FVC without strengthening ecological protection.(3)Average relative humidity and accumulated precipitation were the dominant factors driving the spatial distribution of FVC in the Three-North region,both of which had q values greater than 0.600.The explanatory power of natural factors on the spatial distribution of FVC was stronger than that of anthropogenic factors,but the effect of anthropogenic factors,such as population density and land use type,on FVC gradually increased.The effect mechanism of different factors on FVC was varied.The two-factor interaction enhanced the explanatory power of the factors on the spatial distribution of FVC,and the interaction between the factors was dominated by the Bi-enhance and Nonlinear-enhance effect.Among them,the interaction factor that played the largest leading role was the combination of average relative humidity and altitude and potential evapotranspiration in the corresponding year.This study provides a reference for estimation and spatial-temporal evaluation of fractional vegetation coverage in the Three-North region.The temporal and spatial resolution of the analysis needs to be further improved in the future in order to obtain more accurate and finer spatial distribution of fractional vegetation coverage. |
PDF Full Text Request