Monitoring Hydroperiod and Hydropatterns of coastal wetland systems in Mexico using Landsat time series

The main objective of this study was the evaluation of remote sensing methods that allow for extraction of metrics that link riparian flow regimes to hydro-periods (duration) and -patterns (extent) of wetland systems known to be of critical importance to migratory water fowl and shorebirds along the Pacific Flyway in Mexico. In this study we emphasized the use of freely available and easily accessible optical remote-sensing data and their processing using free and open-source tools. 

Through application of a set of common and well documented water and vegetation indices on the full Landsat 5 and Landsat 7 record spanning two decades, we created a data set that captures episodic, intra-annual and inter-annual variability in inundation for two contrasting wetland systems. For this study we focussed on the Marismas Nacionales wetland system along the Pacific coast and the Alvarado Lagoon system on the Gulf coast. A comparison of indices designed to extract vegetation and water characteristics from Landsat data (NDVI, EVI, NDWI, Tasseled Cap and MNDWI) led us to conclude that the Modified Normalized Difference Water Index (MNDWI) was most effective for identifying inundated areas while the Normalized Difference Vegetation Index (NDVI) worked best for identifying differences in vegetated areas. Our study also established that the high sensitivity to thresholds requires site specific optimization.

For the study we developed metrics to represent the hydro-pattern and hydro-periodicity of waterbodies in the study areas. The first method provides a metric for the intra-annual and inter-annual permanence of water bodies, while the second method quantifies recurrence of seasonal inundation. The Marismas Nacionales revealed a surprisingly strong and direct relationship between inundated area and gauge meassured discharge of the Rio San Pedro Mezquital. Annual and multi annual hydropatterns in this system are very strong and predictable, and primarily driven by large scale inundation of the delta of this river as it enters Marismas Nacionales. The relationship between discharge and inundated area was so string that the inundated area (up to several hundreds of sqare kilometers during peaks) remained correlated throught the full range of the hydrograph. For this system recurrent inundation patterns and their timing metrics were linked to specific ecosystem types and used to inform a bird conservation planning effort.

At the Laguna de Alvarado a very different dynamic was observed, where large scale inundation was less frequent, permanent water bodies were much more persistent in space, and the correlation between inundated area and discharge was much weaker. In this region persistent cloud cover was an issue and SAR based approached may be the only way to monitor inundation dynamics more consistently. Earlier studies by WIckel et al for other systems using PALSAR data for wetland systems in Colombia revealed other technical shortcomings of these kinds of data. A study by Colditz et al for wetland systems in Mexico revealed a strong potential of MODIS derived MNDWI data as well. We propose that future efforts explore the possibilties of aplications of combined (optical and SAR) products.