Plugin was written by Andrew Freebairn, it contains and Harmful Algela Bloom probability model (HAB), a Black Water model and functions that are used in specific projects for traffic light style reporting. The HAB and Black water model have been developed by Senior Research Scientist Klaus Joehnk of CSIRO [email protected].
|License||As-is, use at your own risk|
There will be a report released soon which describes the model in more depth. This page will be updated once the report is published.
The blackwater model included in Source under the project time constraints of implementation is not capable of describing all processes in the floodplains and wetlands. For larger wetlands it is recommended to apply a more complex blackwater generation tool like the Blackwater Risk Assessment Tool (BRAT, Whitworth and Baldwin 2013) to generate DOC and DO at the outlet point. For Chowilla such an extension was recently implemented by adding a modification for anabranch wetlands (Rob Daly personal comm., Wallace et al. 2015). A consistent application for such a tool at all connected wetlands complemented with detailed knowledge on vegetation diversity and litter loads on a spatial scale would significantly improve the Source blackwater model. Here we approximate production of DOC in a wetland or floodplain by assuming instantaneous release of DOC in the inundated area and constant values for leave litter loading. A constant monitoring of litter loading and/or a litter accumulation tool would be necessary to give a better risk assessment. In the main river reaches DOC and DO are simulated using DOC and DO inflows from wetlands and floodplains, decomposition, and reaeration as described by (Howitt et al. 2007, Whitworth et al. 2013, Whitworth and Baldwin 2013). The inundated area of the floodplains are estimated using their relation to flow rates as described by the River Murray Floodplain Inundation Model (RiM-FIM, Overton et al. 2006). For each reach defined in Source the DOC and DO development has to be simulated given internally simulated variables like flow rate, inundated area, and water temperature (the calculation of the latter is described in the section for HAB). Additional inflow of DO and DOC is coming from confluences, e.g. from wetlands. Furthermore it is assumed that a given flow will immediately affect the whole reach to give a time series of DO values.
Large amounts of particulate organic carbon exist on floodplains as litter (>500 gC m–2) and coarse woody debris (~6 kgC m–2). Bacteria will rapidly decompose the organic carbon in wetlands and floods may release about 50 gDOC m–2 from leaf litter. In Chowilla floodplain River Red Gums are contributing an order of magnitude more to BOD demands which is exerted within the first 3-days following inundation (Brookes et al. 2007). Lets denote the amount of litter loading on a floodplain in terms of dry weight per unit area as, B [gDW m-2]. Litter load will increase seasonally by litter fall, which is described by a litter fall rate dB [gDW m-2 d-1]. Values for these are given for example in the BRAT model (Whitworth and Baldwin 2013) yielding in that example case a total litter fall over one year of about B~140 gDW m-2 y-1. According to Robertson et al. (1999) on floodplains, primary production by river red gum (Eucalyptus camaldulensis) forests is about 600 gC m–2 y–1. DOC will be leached from this organic material with a leaching rate b [1/d], where the maximal content of carbon available for leaching is given by m [mgC gDW-1]. The amount of DOC released from organic matter over time is formulated as exponential decay for the inundated part of the floodplain, A [m2]. The total available carbon content in the inundated area is then m(T)BA.
Prerequisites and limitations of this blackwater model are
- Knowledge of litter loading in the different regions along the river channel is necessary.
- Usage of a more complex model like BRAT in wetlands like Chowilla using the DOC and DO output as input into the Source blackwater model is advised.
- Litter loading is not calculated independently for different inundation areas and timings, a parameter is used to “re-distribute” litter each time step. This can lead to an overestimation of DOC loading and thus underestimation of oxygen minima. Thus, this tool serves as a risk estimator.
Here we take an approach to determine the risk of cyanobacteria blooms instead of simulating their cell numbers per se. While the latter would be preferable to know, it is not feasible in the terms of this project as it would need a much higher amount of data to verify the model and minimize uncertainties. A risk index only looks at the environmental conditions and if they are favourable or not for cyanobacteria growth. The main factors driving cyanobacteria are known, they are flow, water temperature, vertical mixing, light availability, and nutrient availability. Some of these fields are not regular monitored or cannot be derived from other data.
Harmful Algal Bloom Model
A conceptual model for the HABs (and blackwater) is set up according to the following scheme. Based on variables simulated by Source, water temperature calculated from air temperature, and a set of HAB specific parameters, risk probabilities in the range [0,1] for the occurrence of HABs are calculated for different reaches, or storages and then passed on to the next node for seeding to derive a combined risk index. Based on the calculated risk index a flag will be raised.
Conceptual model of harmful algal blooms used in the Source modelling
Model of risk indices, P, along a river section
For each reach or storage a risk index for HABs is calculated. This is then passed along to seed the next reach. The combination of risk indices as they flow through the system is calculated using flow weighting at confluences, i.e. mixing, or simply passed along. In a reach the risk index is calculated on basis of the local water temperature, flow, or water level, describing temperature and nutrient dependent growth and mixing behaviour. Seeding and local risk are then added together for a final risk value. This is done for each time step at each node, where a risk index calculation is set up.
Using the plugin
Load using the Plugin manager. Custom functions have been written for these models and so are accessed via the Function manager. Each function has been described and the arguments listed so that the functions are used correctly.
Example of the above custom function used in a normal model function
Source code available at https://bitbucket.org/ewater/sourceplugin.csiro.rivermurraydss/