Quantifying moderate resolution remote sensing phenology of Louisiana coastal marshes

Abstract

Coastal ecosystems are under multiple stresses ranging from global climate change to regional hazardous weather and human interventions. Coastal marshes in Louisiana are inherently vulnerable to these threats because they are microtidal and inhabit a narrow portion of the intertidal zone. Phenological dynamics of the marshes offer valuable information on the stressors’ impacts, yet they have rarely been reported or compared. Here, we study the landscape-level phenologies of the marshes under different climatic conditions, using Landsat-derived Normalized Difference Vegetation Index (NDVI) records (30×30m2 spatial resolution) and a nonlinear mixed model that enables a quantitative analysis of nonlinear and piecewise functions involving repeated measures. In 2007 (a normal year), the Gaussian function was the best phenological model for Louisiana coastal marshes (pseudo R2 0.56–0.85), showing that: (1) NDVI of all marshes peaked within one month from late July to mid-August; (2) freshwater marshes had the highest peak NDVI, followed by intermediate, brackish, and saline marshes; and (3) saline marshes had the longest growth duration, followed by brackish, and then intermediate and freshwater marshes. Phenological shifts were found in years featuring extreme weather events: (1) a two-month delay in the peak NDVI day of saline marshes in 1999 (a drought year) compared to 2007; and (2) a shortening in growth duration of all marshes by approximately half in 2005 (a hurricane year). This work presents a methodgology to analyze and predict Louisiana coastal marshes' phenological dynamics in response to current and future stresses.