Above-ground biomass estimation of Pinus densata based on Landsat time series images and AHTC algorithm
[Objective]To improve the accuracy of above-ground biomass estimation of Pinus Densata,a filter algorithm was developed to reduce the noise image of Landsat time series data.[Method]Based on the data of national forest resources consecutive inventory fixed samples in 1987,1992,1997,2002,2007,2012 and 2017,and the Landsat time series images from 1987 to 2017,using Google Earth Engine(GEE)and Python,the LandTrendr filter,Savitzky-Golay filter,horn convolution and adaptive horn topography convolution(AHTC)algorithm based on spatial convolution theory were used to filter Landsat time series data.Random forest regression(RFR)was used to construct the above-ground biomass estimation model of P.densata in Shangri-la city,and select the optimal estimation model to invert and map the aboveground biomass of P.densata in 1987,1992,1997,2002,2007,2012 and 2017.[Result]1)According to the root mean square error(RMSE)and peak signal-to-noise ratio(PSNR)of the direct evaluation index of the image,the image quality after filtering by AHTC algorithm was the best;2)In the case of using the RFR method,the filtered data showed higher estimation accuracy than the original data;3)The time series data filtered by AHTC algorithm had the best effect on estimating the above-ground biomass of P.densata,with the decision factor R2 of 0.885,root mean square error RMSE of 34.63 t/hm2,prediction accuracy P of 60.75%,and relative root mean square error rRMSE of 43.59%;4)The inversion results using AHTC and RFR methods are:12 million 360 thousand tons(1987),11 million 550 thousand tons(1992),14 million 550 thousand tons(1997),13 million 330 thousand tons(2002),13 million 140 thousand tons(2007),13 million 450 thousand tons(2012),16 million 540 thousand tons(2017).[Conclusion]The use of AHTC filtering method can eliminate a lot of noise and uncertainty carried by the time series image itself to a certain extent,effectively improving the quality of time series images,and also providing a new approach to improve the accuracy of remote sensing estimation of above-ground biomass of P.densata.
filteringLandsat time seriesPinus densatabiomassAHTC
NETL
NSTL
万方数据
