CES Home / Research / Everglades Salinity

Salinity Impacts on the Greater Everglades

Funding for this project provided by the USGS
 

Changing Salinity as an Indicator of Restoration and Climate Change Impacts on the Greater Everglades

 

 

Florida Bay

Florida Bay. Image Credit: SFWMD

FAU researchers, Dr. Caiyun Zhang, Dr. Zhixiao Xie and Dr. Leonard Berry, have been assessing salinity changes in the Greater Everglades using remote sensing techniques. The ongoing project explores the potential of the Landsat Thematic Mapper (TM) sensor to monitor salinity in Florida Bay, supporting the Comprehensive Everglades Restoration Plan (CERP) efforts. Remote sensing models have been built for Northeast Florida Bay area first, but eventually for the entire Florida Bay. The results of the initial pilot study indicate that remote sensing has the capability to monitor salinity in the selected study areas and can serve as a supplementary tool to reduce the cost of salinity monitoring programs in CERP.

Previous research (including citations from two resulting journal articles): During the 2011 – 2012 pilot study, researchers  investigated  the  applicability  of  Landsat  remote  sensing  data  for  surface salinity monitoring and modeling in Florida Bay. First, the team collected field salinity data surveyed by  USGS  from  the  South  Florida  Information  Access  (SOFIA,  http://sofia.usgs.gov/)  and Landsat TM data at the USGS’s Earth Explorer website. Then, they spatially and  temporally matched  these  two  datasets with  an  attempt  to  find  the  empirical  relationships between surface salinity and TM data. Two study areas were selected: the northeastern Florida Bay, and then the entire Florida Bay. The northeastern bay area is the discharge location of the wide C-111 canal and Taylor Slough carrying a large volume of fresh water into the bay, which makes its water mass different from its surroundings. Florida Bay is a key feature in the Everglades. Salinity monitoring within these two regions is important in CERP. The team  tested  several models  including remote  sensing  classification  methods,  multivariate  regression  analysis  models,  and geographically weighted  regression  (GWR) models. Some conclusions are indicated from this project in terms of the applicability of Landsat TM data to salinity monitoring and modeling over the northeastern Florida Bay:

  • Landsat TM data appear to be effective for salinity assessment in northeast Florida Bay. A highly significant relationship between the TM data and salinity are identified for both dry and wet seasons. Expected salinity patterns are presented on the TM estimated salinity maps. Time-series salinity maps can provide variability of salinity in the bay, which can be used to measure the effects of restoration projects in CERP.
  • The  empirical  approaches  for  quantitative  salinity  estimation  generate  more  acceptable results  than  the  classification  methods  for  qualitative  salinity  assessment.  The empirical algorithms are statistically significant and are preferable for operational purposes in this area.
  • Bands 1, 3, and 4 were suitable for salinity estimation when used together. A combination of these  three  bands  in  the  established models  explained more  than  70%  of  the  variation  in salinity. They afford a reliable surface salinity prediction capability.
  • Salinity  in  the  dry  season  is more  predictable  than  in  the wet  season. Heavy rainfall and runoff in the wet season make the bay environment more complex. This causes the salinity assessment in the wet season more difficult.

Further experiments for the entirety of Florida Bay demonstrate that location-specific correlations are evident between surface salinity and spectral response in TM data. GWR is proven to be suitable and utilized in modeling and predicting such spatially variable relations. To build an effective GWR model, it seems  necessary  to perform  a principal  component analysis  (PCA)  of  the TM data  to  remove  the  collinearity  between  explanatory  variables  while  incorporating  as  much information  from multiple  spectral bands.  In summary, initial experiments suggest  that  remote sensing could potentially serve as a less costly alternative or a supplement to field salinity survey currently undertaken in the coastal areas of the Everglades. The research findings were presented in  two  conferences  (SEDAAG  2011,  INTECOL  2012)  and  two  peer-reviewed  journal  articles were published.

Citations:

Zhang, C., Xie, Z., Roberts, C., Berry, L., & Chen, G. 2012. Salinity Assessment in Northeast Florida Bay Using Landsat TM Data. southeastern geographer, 52(3), 267-281.

Xie, Z., Zhang, C., & Berry, L. 2012. Geographically weighted modelling of surface salinity in Florida Bay using Landsat TM data. Remote Sensing Letters, 4(1), 76-84.



 Last Modified 4/3/23