13.07.2015 Views

Estimating the Water Requirements for Plants of Floodplain Wetlands

Estimating the Water Requirements for Plants of Floodplain Wetlands

Estimating the Water Requirements for Plants of Floodplain Wetlands

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

flood regime. These categories were based on visual inspection andexpert interpretation <strong>of</strong> <strong>the</strong> flood regime <strong>for</strong> <strong>the</strong> ‘natural’ scenario fromIQQM. The natural scenario uses <strong>the</strong> historic rainfall series to generateflows in <strong>the</strong> river system with no regulation and no diversions. Onehundred-yearsimulations using IQQM <strong>the</strong>re<strong>for</strong>e allowed time series <strong>of</strong><strong>the</strong> health status <strong>of</strong> each vegetation class to be generated.Category 2: hydrologic/empiricalApproaches in this category again rely on hydrologic modelling (waterbalance) <strong>of</strong> <strong>the</strong> water regime, but use empirical vegetation–hydrologyrelationships derived from analysis <strong>of</strong> relevant data. The empiricalrelationships may have been derived <strong>for</strong> <strong>the</strong> location being investigated,or may have been derived from prior investigations in similar locations.The empirical relationships may be based primarily on analysis <strong>of</strong> spatialdata, or on analysis <strong>of</strong> time series data, or on complete analysis <strong>of</strong> spatiotemporalpatterns. The nature <strong>of</strong> <strong>the</strong> available relationships will dictate<strong>the</strong> nature <strong>of</strong> <strong>the</strong> possible predictions. Relationships may take various<strong>for</strong>ms, including regression models (linear or higher orderrelationships), and probabilistic models based on definition <strong>of</strong> jointprobability distributions; <strong>for</strong> example, <strong>the</strong> probability <strong>of</strong> vegetationmeasure ‘A’ occurring in water regime class ‘B’ (Section 4).Prairie wetlands, ND. An example is <strong>the</strong> modelling work <strong>of</strong> Poiani andJohnson (1993) <strong>for</strong> two semi-permanent prairie wetlands in NorthDakota. The modelling included a simple daily water balance model <strong>for</strong>individual wetlands. The water balance included direct rainfall inputs,local run-<strong>of</strong>f, and evapotranspiration. Groundwater exchanges were notdirectly considered. The water elevations determined by <strong>the</strong> waterbalance model were used toge<strong>the</strong>r with ground elevation data tocalculate water depths <strong>for</strong> each cell in a grid representation <strong>of</strong> <strong>the</strong>wetland. Six vegetation types (<strong>of</strong> species with similar life-histories) andan open water category were used in <strong>the</strong> vegetation response model.The vegetation response model is a series <strong>of</strong> rules that describes <strong>the</strong>seasonal water regime conditions necessary to effect a change from onevegetation category to ano<strong>the</strong>r. These rules were based on data,observations and analysis from a significant amount <strong>of</strong> prior fieldresearch by various investigators in <strong>the</strong>se prairie wetlands. Both modelswere developed and calibrated using data from one wetland, and <strong>the</strong>nrun and evaluated <strong>for</strong> a second wetland.Riparian wetlands, Ontario. A second example is <strong>the</strong> logisticregression vegetation–hydrology model <strong>of</strong> Toner and Keddy (1997) <strong>for</strong>riparian wetlands in Ontario, Canada. In this investigation, models weredeveloped to simply predict <strong>the</strong> presence or absence <strong>of</strong> woody cover, asdetermined by locating <strong>the</strong> woody–herbaceous boundary within awetland. A set <strong>of</strong> seven hydrologic variables was selected to reflect <strong>the</strong>depth, duration and time <strong>of</strong> flooding. Relationships between vegetation(woody or herbaceous) and <strong>the</strong> hydrologic variables were establishedby logistic regressions. Relationships were developed using first, eachhydrology variable and four o<strong>the</strong>r site variables, and second, all possiblecombinations <strong>of</strong> <strong>the</strong> seven hydrologic variables. Statistical criteria wereused to select <strong>the</strong> best model generated. The model was notindependently validated by <strong>the</strong> authors, nor was it linked to a hydrologymodel <strong>for</strong> use in scenario simulations.Section 7: Predicting Vegetation Responses 87

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!