Examining the potential of using information on fire detected by MODIS and socio-economic variables to highlight potential coca cultivations in forest areas in Colombia

Fires in forest areas are considered an important threat to the Andean Region and the Amazon rainforest. In Colombia, fire is used to expand the agricultural frontier (including illicit crops) which results in deforestation. Given the importance of avoiding deforestation and to control coca expansion, this paper aims to: 1) understand the relationship between fires and deforestation, coca and deforestation and hence coca and fires; 2) examine the potential of using fire data from remote sensing and socio-economic variables to predict the occurrence of new coca fields in forest areas in Colombia. The analysis was undertaken over a ten year period (2000-2010) at a municipality level in to areas with high coca dynamics (Central Region and Putumayo - Caqueta) using Pearson correlation and three different models: a Linear Probability model, a Logit model and a Probit model. The results show that there is a positive relationship between fire and deforestation. Although in general the correlation between coca and deforestation is positive, it differs at the municipality level depending upon the area of forest cover and the coca plot size. The results of the Logit and Probit models show that fire and expulsion, which is a measure of forced displacement by violence, can be used as indicators to highlight coca expansion in forest areas