Estimation of rubber stand age in typhoon and chilling injury afflicted area with Landsat TM data: A case study in Hainan Island, China

Stand age in rubber (Hevea brasiliensis) plantation is an important factor for production management and ecosystem modeling. The feasibility of detecting difference in rubber stand age using Landsat Thematic Mapper (TM) data in Hainan Island, an area frequently disturbed by typhoon (or hurricane, tropical cyclone) and chilling injury, was evaluated. The results indicated that most of the TM bands were negatively correlated with stand age, of which the near-infrared bands had the most stable and strongest coefficients. However, the correlation of infrared bands with stand age was dramatically weakened by disturbance of severe typhoon and chilling injury, which suggested that, to study the stand age of rubber plantation by remote sensing, it is more appropriate to use an image acquired in dry season and far from severe natural disaster. The TM bands, normalized- and ratio-based vegetation indices that computed from red and infrared bands (3–5 and 7), and tasseled cap components were employed to develop models for estimating stand age through multivariate regression analysis techniques. Four statistically significant models with R2 ranging from 0.74 to 0.82 were obtained. When applied to the validation data, all the regression models showed an overestimation of young stands (about less than 25 years) and more severe underestimation of old stands due to the growth characteristics of rubber tree and disturbance of natural disasters. The incorporation of original bands with vegetation indices and tasseled cap components as independent variables could yield the highest estimated correlation index squared (I2) value of 0.58, the RMSE value of 5.96 years, and less model bias than the other test models. This study demonstrated a close relationship between rubber stand age and Landsat TM data, and a promising prospect in application of TM data to investigating stand parameters in Hainan Island.

► The near-infrared bands show most stable and strongest correlation with stand age.
► The correlation was dramatically disturbed by severer typhoon and chilling injury.
► The combination of bands, VIs, TCT components are good predictors of stand age.
► TM data is useful for retrieving rubber stand age despite the estimating bias.