Annual South American Forest Loss Estimates (1989-2011) Based on Passive Microwave Remote Sensing
Abstract:Vegetation dynamics, such as forest loss, are an important factor in global climate, but long-term and consistent information on these dynamics on continental scales is lacking. We have quantified large-scale forest loss over the 90s and 00s in the tropical biomes of South America using a passive-microwave satellite-based vegetation product.
Our forest loss estimates are based on remotely sensed vegetation optical depth (VOD), which is an indicator of vegetation water content simultaneously retrieved with soil moisture. The advantage of low-frequency microwave remote sensing is that aerosols and clouds do not affect the observations. Furthermore, the longer wavelengths of passive microwaves penetrate deeper into vegetation than other products derived from optical and thermal sensors. This has the consequence that both woody parts of vegetation and leaves can be observed. The merged VOD product of AMSR-E and SSM/I observations, which covers over 23 years of daily observations, is used.
We used this data stream and an outlier detection algorithm to quantify spatial and temporal variations in forest loss dynamics. Qualitatively, our results compared favorably to the newly developed Global Forest Change (GFC) maps based on Landsat data (r2=0.96), and this allowed us to convert the VOD outlier count to forest loss.
Our results are spatially explicit with a 0.25-degree resolution and annual time step and we will present our estimates on country level. The added benefit of our results compared to GFC is the longer time period. The results indicate a relatively steady increase in forest loss in Brazil from 1989 until 2003, followed by two high forest loss years and a declining trend afterwards. This contrasts with other South American countries such as Bolivia and Peru, where forest losses increased in almost the whole 00s in comparison with the 90s.