Research efforts have focused on both lakes; for one because the are in close proximity for another because they are physically connected with water flowing from Lake Azuei (LA)  to Lake Enriquillo (LE). The latter means that they have to be investigated as a linked system with water levels in both lakes controlling the amount that is flowing from LA to LE. Shrink- and Growth patterns have been attributed to a number causes. One suggestion has been that earthquakes along the Enriquillo Garden fault have altered the lake beds allowing additional sub surface flows to add to the volume, others have suggested that is anthropogenic impact (mostly deforestation but also water diversions) that has led to the rise, some speculated that there had to be significant interbasin transfer (again subsurface) made possible because LE is so low in elevation, while others suggest that this is mostly climate change based. 

While initial efforts sought to develop complete (numerical, for example SWAT based) representations of the lakes' watersheds our efforts ultimately focused on analyzing time series data on lake extent and volume as well as climatological data instead. The decision was made for a number of reasons. Firstly, data on land cover, land use, and soil all of which are essential to numerical model was poor to non-existent resulting in substantial gestimating in addition to very coarse (both geospatial and temporal) hydro-climate data sets. Secondly, while the DR maintains a climatological station in Jimani, data collection efforts on the haitian side was/is low and behold non existent, especially as hydro-climate and stream flow data are concerned. Thirdly, best data sources proved to be remote sensing imagery (LandSat, about 34 yrs), and the somewhat longer term climate data collection efforts in Jimani which also provided some precipitation sets in addition to dynamically downscaled gridded hydro climate data created by researchers at the City College of New York. This allowed us to focus on time variant water budget calculations for the lake system, rather than trying to map the watersheds at much higher spatial and temporal scales.



What caused the lakes in 2005 to start rising to the unprecedented levels of 2014? Why then and not before, and why did the levels seemed to arrest in 2014 and now even slightly receded?

To identify the key contributors to the rise and their relative importance. As a result, build a model that would allow some predictive capability and could be driven with forecast data.


Data Collection:

Due to the darth of data collected locally in the region the focus shifted to using data sources with global scope, i.e. remote sensing data augmented by local data collection campaigns.


1) A key effort was the collection of bathymetric data to get a digital representation of the lakes and thus their volume. Two campaigns in 2013 and 2014 yielded sufficient sonar based depth data from which we constructed two Digital Bathymetry Models, DBMs, for the two lakes.

2) Remote sensing data focused on using LandSat imagery (available from 1979) as well as Shuttle Radar Topography Mission, SRTM, data. We used the former to extract lake extent information over time, and the latter to help create so-called "bathtub" models of the lakes which we could fill and drain numerically.


3) We used as much as possible hydro-climate data from the region of the lakes which was mostly supplied by the Oficina Nacional de Meteorologia, ONAMET, of the DR station located in the border town of Jimani. This was augmented by data from other stations in the lake regions as much as possible.


4) Local efforts also included the installation of hydro-climate stations, i.e. several stations along transecs running vertically from lake level to higher elevation to catch the climatological stratification as elevations change. A key focus was the Montane Forrest belt (elevations 500 - 1000 meters) as producer or collector of moisture volumes that would add to the those produced by storm systems moving through the regions.

5) We installed two lake level sensors (one each for LA and LE) to have a denser time series data set with better capability to show short term responses of the lakes to extreme events such hurricanes or tropical storms; something that LandSat imagery could not provide due its 16-day collection intervals.


6) Lastly, we conducted numerous interviews with inhabitants living around the lakes to get a better feel for the impacts, responses, government activities to alleviate the problems, moods and fears and concerns, in addition to trying to unearth historical observations on the lakes reaching further back than the data sets we had. 

Time Series Analysis:

We are using a tong-pronged approach to gain insight into the dynamics that govern the lakes behaviors.


1) the first prong concerns the hydro-climatic processes with a focus on the larger picture, or supra regional connections. These concern the linkages of Hispaniola's climate and weather patterns to the Caribbean region which in turn are coupled the North Atlantic weather systems. Because of these very large scales (as compared to just the lakes region) research has been carried out by climate modelers at the City College of New York (Dr. Jorge Gonzalez) and NOAA (Dr. Daniel Comarazamy) to develop hydro climate data for smaller spatial scales through dynamic downscaling of GCMs data. Details of this work can be found at Comarazamy et al., 2015 in the Publications section.

2) The second prong concerns local data time series information of geospatial nature (extent and volume of the lakes) as well as hydro-climate data (mostly precipitation but also evaporation, evapo-transpiration, air temperature and solar radiation) in addition to extreme event data. The central assumption here is that the lakes' dynamics are controlled by climate and weather patterns and events in addition to human controlled activity. The former is intended to establish links between hydro climate patterns and lake response pattern on larger temporal scales, which are then superimposed with extreme event (short time scales) analysis that could help explain otherwise inexplicable marks/events in the time series. We have deployed time series analyses tools such as Wavelet Spectra, Trend analysis, Change Point identifications, and simple statistics to characterize the lakes in their temporal domain. 

© 2017 Created by Michael Piasecki

  • Facebook Social Icon
  • Twitter Social Icon
  • Google+ Social Icon
Change Point Detection

Pettit Test to identify significant change points in TS