Estimation of aboveground biomass of arboreal species in the semi-arid region of Brazil using SAR (synthetic aperture radar) images

分类： 地球科学 >> 地理学 提交时间： 2023-06-13 合作期刊: 《干旱区科学》

Janisson B de JESUS Tatiana M KUPLICH Íkaro D de C BARRETO Fernando L HILLEBRAND Cristiano N da ROSA

摘要：The Caatinga biome is an important ecosystem in the semi-arid region of Brazil. It has significantly degraded due to human activities and is currently a region undergoing desertification. Thus, monitoring the variation in the Caatinga biome has become essential for its sustainable development. However, traditional methods for estimating aboveground biomass (AGB) are time-consuming and destructive. Remote sensing, such as optical and radar imaging, can estimate and correlate with vegetation. Nevertheless, radar imaging is still a novelty to be applied in estimating the AGB of this biome, which is an area with little research. Therefore, this study aimed to use Sentinel-1 images to estimate the AGB of the Caatinga biome in Sergipe State (northeastern Brazil) and to verify its influencing factors. Nineteen sample plots (30 m×30 m) were selected, and the stems of individuals with a circumference at breast height (1.3 m above the ground) equal to or greater than 6.0 cm were measured, and the AGB through an allometric equation was estimated. The Sentinel-1 images from 3 different periods (green, intermediate, and dry periods) were used to consider the phenological conditions of the Caatinga biome. All the pre-processing and extraction of attributes (co-polarized VV (vertical transmit and vertical receive), cross-polarized VH (vertical transmit and horizontal receive), and band ratio VH/VV backscatter, radar vegetation index, dual polarization synthetic aperture radar (SAR) vegetation index (DPSVI), entropy (H), and alpha angle (α)) were performed with Sentinel's Application Platform. These attributes were used to estimate the AGB through simple and multiple linear regressions and evaluated by the coefficients of determination (R2), correlation (r), and root mean squared error (RMSE). The results showed that the attributes individually had little ability to estimate the AGB of the Caatinga biome in the three periods. Combined with multiple regression, we found that the intermediate period presented the equation with the best results among the observed and estimated variables (R2=0.73; r=0.85; RMSE=8.33 Mg/hm2), followed by the greenness period (R2=0.72; r=0.85; RMSE=8.40 Mg/hm2). The attributes contributing to these equations were VH/VV, DPSVI, H, α, and co-polarized VV for the green period and cross-polarized VH for the intermediate period. The study showed that the Sentinel-1 images could be used to estimate the AGB of the Caatinga biome in the green and intermediate phenological periods since the SAR attributes highly correlated with the estimated variable (i.e., AGB) through multiple linear equations.
 

 点击量 1788  下载量 243  评论