COS 95-2 - Monitoring the post-fire vegetation recovery using digital repeated photography data

Friday, August 12, 2016: 8:20 AM
220/221, Ft Lauderdale Convention Center
Bruna Alberton1, Jurandy Almeida2, Ricardo Torres3, Swanni Alvarado4, Leonardo Cancian1, Bruno Borges1, Thiago Silva4 and Patricia Morellato5, (1)Department of Botany, UNESP Universidade Estadual Paulista, Rio Claro, Brazil, (2)Institute of Science and Technology, UNIFESP Universidade Federal de São Paulo, São José dos Campos, Brazil, (3)Institute of Computing, UNICAMP Universidade de Campinas, Campinas, Brazil, (4)Department of Geography, UNESP Universidade Estadual Paulista, Rio Claro, Brazil, (5)Instituto de Biociências, Universidade Estadual Paulista (UNESP), Rio Claro, Brazil
Background/Question/Methods

Fire is a critical component of several ecosystems and can shape vegetation physiognomy and determine the diversity of species. Phenology can be an accurate indicator of plant responses to wildfires, helping evaluate fire impact and serving as an indicator of the resilience of natural ecosystems. Digital cameras have been proven to be a reliable tool for monitoring plant phenology in the tropics. We have monitored daily phenological responses using digital images in different areas immediately after fire, aiming to: (i) assess the reliability of digital cameras as a novel tool for monitoring fire-prone vegetation, and (ii) evaluate the recovery of different vegetation physiognomies after an extensive fire event on 2014, over a rupestrian grassland landscape of the tropical Espinhaço mountain range, southeastern Brazil. We monitored five vegetation sites of a heterogeneous landscape of rupestrian grasslands with time-lapse cameras. We selected Regions of Interest (ROIs) [see 1] on each image and extracted the digital RGB color information over a time-span of 110 days after the fire event. From this data, we calculated the Excess Green index (ExG) to track vegetative phenology. For each ROI, we extracted the phenological metrics of initial date, peak date, and amplitude. 

Results/Conclusions

A total of 7,150 images were processed, and ExG time series were extracted from eight ROIs corresponding to a peat bog (1), a rocky grassland (3), a wet grassland (3) and an altitudinal grassland (1). Our results showed different ExG curves for each vegetation type, corresponding to changes in green color response that represent temporal changes of vegetation greening after fire. Peat bog vegetation had the faster post-fire recovery, reaching a peak of green signal on the 33rd day after fire, followed by wet grasslands (35th, 40th and 51st), while altitudinal grassland leafing peaked in the 49th day, and rocky grasslands peaked last (62nd, 78th and 79th). Our results indicate that the harsh environment experienced by rocky grasslands led to longer recovery times. Wet grasslands and peat bogs had the highest amplitude values, which may indicate a larger biomass gain for these vegetation types. This is the first time that digital cameras have been used for post-fire monitoring. As the reliability of this tool is confirmed, efforts in the development of new indices and analysis to extract further ecological information from vegetation communities can be implemented widely.