Linear Spectral Mixture Analysis of SPOT-7 for Tea Yield Estimation in Pagilaran Estate, Batang Central Java

2nd International Conference of Indonesian Society for Remote Sensing 2016

地球科学;计算机科学

Fauziana, F.^1 ; Danoedoro, P.^2 ; Heru Murti, S.^3

Remote Sensing, Geography Faculty, Gadjah Mada University, Yogyakarta

55281, Indonesia^1

PUSPICS Geography Faculty, Gadjah Mada University, Yogyakarta

55281, Indonesia^2

Cartography and Remote Sensing, Geography Faculty, Gadjah Mada University, Yogyakarta

55281, Indonesia^3

关键词: High-accuracy;    Linear spectral mixture analysis;    Model results;    Multispectral remote sensing;    Normalized difference vegetation index;    Root mean square errors;    Tree shades;    Yield estimation;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/47/1/012034/pdf
DOI  :  10.1088/1755-1315/47/1/012034

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Remote sensing has been utilized especially for agriculture yield estimation. Tea yield is effected by biology characteristic including crown density. The challenge of tea yield estimation uses multispectral remote sensing data is the presence of object beside tea. This mixed pixel problem can disturb spectrally to recognize tea tree, so it is necessary to use pixel approach. The aims of this research are (1) to determine fraction of tea and non-tea; (2) to estimate crown density percentage based on tea Normalized Difference Vegetation Index (NDVI); (3) to estimate tea yield based on crown density. SPOT-7 was utilized for this application. Linear Spectral Mixture Analysis (LSMA) has applied to determination fraction percentage each pixel. Each pure endmember was read the NDVI value. NDVI of tea tree has sensitivity with crown density. Counting tea NDVI was applied for NDVI mixed pixel. Linear regression analysis has applied for estimating crown density and tea yield. The results of this research are SPOT -7 which can recognize tea, tree shade, impervious and soil each pixel with accuracy 99,84%. Although it produced high accuracy, it has overestimate at certain tea estate because of the attendance of impervious. Regression analysis of crown density and NDVI showed coeffisien determination 52%. This model result 4-100% crown density percentage, where crown density 4-55% were located beside tea tree or pruned-tea block. Regression analysis of crown density and tea yield relation showed coeffisien determination 45%. This model produced 161,34-1296,8 kg/ha. Each this model resulted Root Mean Square Error (RMSE) 14,27% and 551,52 kg/ha.