A MODIS-Based Algorithm to Detect Forest Degradation: A Case Study in Mexico
MetadataShow full metadata
With global warming becoming a major concern worldwide, forest degradation impacts on environmental services, especially those related to climate regulation through carbon sequestration, have received increasing attention among the scientific community. However, monitoring forest degradation has not been easy to accomplish due to the non-discrete nature of the process, in which changes are subtle and alter vegetation gradually. This research proposes an algorithm to detect forest degradation using Moderate Resolution Imaging Spectroradiometer (MODIS) images collected over Central Mexico (tile h08v06) between 2002 and 2017. The underlying assumption of a constant negative relationship between vegetation greenness and surface temperature, which has guided several studies that aim to identify ecosystem disturbances, was discarded as a foundation on which to build the algorithm. An evaluation of the annual and intraseasonal relationship between Leaf Area Index (LAI) and Land Surface Temperature (LST) demonstrated that the relationship between these two variables in the study area is not constant and its nature (i.e., sign) varies depending on the temporal scale and forest type under analysis. The use of LAI was proposed to facilitate consideration of the structural changes evident from degradation though not necessarily observable through widely used vegetation spectral indices, such as Normalized Difference Vegetation Index (NDVI)and Enhanced Vegetation Index (EVI). Thus, the proposed algorithm focused on vegetation greenness and overcame the challenge of detecting subtle and gradual vegetation changes through a trend analysis of LAI. Overall, the results indicate that 52% of the study area has experienced increasing LAI trends, 37% has remained unchanged, and 11% exhibits some level of forest degradation (i.e., decreasing LAI trends). Particularly, the algorithm estimated that 0.6% (385 km2) is highly degraded, 5.3% (3,406 km2) moderately degraded, and 5.1% (3,245 km2) slightly degraded. The non-degraded (89%) and degraded (11%) areas served as scenarios to investigate the effect of precipitation on LAI in the context of forest degradation conditions. The results showed that the response of LAI to precipitation is predominantly positive and its occurrence is higher in non-degraded pixels (43%) than in degraded pixels (28%). This dissertation contributes to the body of knowledge focused on monitoring forest degradation and comprehending vegetation-climate feedbacks at regional scales.