Spectral-temporal modelling of bamboo-dominated forest succession
Greig, C., Robertson, C., & Lacerda, A. 2018. Spectral-temporal modelling of bamboo-dominated forest succession in the Atlantic Forest of Southern Brazil. Ecological Modelling. 384(24):316-332
Utilizing remotely sensed imagery to derive bamboo successional dynamics will support the development of synoptic mapping tools for the Atlantic Forest which will directly support conservation and adaptive management planning.
Accordingly, this paper aims to:
- Identify and characterize the spectral-temporal signature of bamboo-dominated forest succession
- Determine whether the perpetuated bamboo-dominance cycle is a synchronous process occurring on a landscape scale by comparing a 32-year Landsat time-series stack to a hybrid model
- Compare performance of the hybrid model to a time-weighted dynamic time warping model approach to assess the effectiveness of the hybrid model, and more generally spectral-temporal signatures, to determine bamboo lifecycle synchrony and identify regions of bamboo-dominance.
To evaluate whether the hypothesized perpetuated bamboo-dominance cycle was occurring on a landscape-scale, we developed a semi-empirical spatially explicit hybrid spectral-temporal model representative of bamboo-dominated forest succession from the time-series of Landsat images. The 32 modelled values of the hybrid model were converted into a 32-year hybrid model raster time-series stack, which was then compared to the 32-year Landsat time-series of smoothed EVI values to determine whether the hybrid model effectively described bamboo-dominated forest succession (Below). Bias and root-mean square error (RMSE) between each pixel in the study area and the hybrid model were computed and summarized, creating raster outputs of both bias and RMSE values.