You are on page 1of 5

MIKE 11 HD and RR Practical Exercises

A series of practical exercises, based on exploring and revising the original PUB model covering the whole Marina Barrage catchment.

Exercise 1: Adding & georeferencing a background image in your Nwk file


a) Add a background image to the network file as a useful reference b) Georeference the added image c) Can also add the sub-catchment polygon shapefile The image (MarinaCatchment.jpg) and a text file containing the georeference coordinates (Image georeference.txt) are located in: \\PUB_Training_Material\2_ReferenceImage\ Catchment shp (NAM_Catchments.shp): \\PUB_Training_Material\2_RRCalibration\Catchments\ Adding layers: Layers Add/Remove Editing layers: Layers Properties Note that the map projection used for the network file in general, and that specified for the background image(s) used should be the same. Other image files (bmp, gif, png, tif), shapefiles (shp) and grid files (dfs2, dt2) can also be added as background layers. The properties (colour, opacity, thickness, etc) of added layers can be edited as necessary.

Exercise 2: Setting up a Rainfall-Runoff (RR) file using NAM methodology


Since the Marina catchment is a relatively complex system, a simplified NAM calibration (looking at the Singapore River catchment only) will be performed, as an intro to using NAM for RR modelling. Suppose you have measured discharge data (m3/s) for a point along Singapore River, and rainfall & evaporation data over the same period. Based on the known catchment area draining to that location, (e.g. from a shapefile), set up and calibrate a simple NAM model using the provided data. a) Open a new Rainfall-Runoff file (.rr11), change to Basin View (View Basin View), and import the NAM_SingaporeRiverCatchment shapefile, (Basin Work Area Import Basin Definitions). You can test the integrity of the polygons using the Test Fill Catchments tool ( ), and then convert them to actual catchment extents using the Create Polygon Catchments tool ( ), remembering to choose NAM catchments. Rename the main gauged catchment, (in the west), to SR, (the others can be left as they are). b) Next specify your rain gauges, (Create New Stations tool ), based on the table below. Note that youll first have to extend your working area to fit in all the rain gauges and catchment polygons, (Basin Work Area Resize Work Area). Gauge X Y S31 27,571 28,308 S71 22,345 30,545 S72 29,659 28,503 S77 25,638 30,686 S92 23,002 29,724 S111 28,311 32,527 S118 29,421 31,343 S120 26,293 32,347

c) Calculate Thiessen polygons based on available rain gauges, (Basin Work Area Thiessen Options), & check resulting polygons using the Thiessens polygons tool ( ), clicking on each catchment in turn to see which gauges are contributing rainfall to that catchment (rain gauge & polygon edges shown in red) and which are not (shown in green). d) Calculate weighted rainfall for each catchment, (Basin Work Area Calculate Mean Precipitation), and then to save time for future simulations uncheck the Weighted Timeseries column in the RR Timeseries table tab, (only needs redone if inputs change). e) Specify on Catchments tab that you want a Calibration Plot for catchment SR and then on NAM Autocalibration tab, check the Include autocalibration. Since were looking at a small urbanised catchment, we might confidently adjust the lower and upper bounds for some parameters, (e.g. Umax: 0.1 to 10, Lmax: 0.1 to 100, CQOF: 0.5 to 1, CK12: 1 to 50). f) Run this .RR11 file and evaluate the RR Calibration results, (comparison of simulated and observed discharge time-series and also accumulated discharge volumes).

g) Impact of varying autocalibrations objective functions? (e.g. peak/low flow RMSE) h) Impact of varying NAM parameters? (e.g. CQOF similar to runoff coefficient / SCS CN)

Exercise 3: Run Original (unmodified) model and evaluate results


Starting with original model, run it once, evaluate the results & discuss how it might be improved. Note: the only change made to the model so far is to reduce the value of delta from 1 to 0.85, (0.85 is the maximum recommended for models with tidal influence see excerpt from MIKE 11 Manual).

\\PUB_Training_Material\0_OriginalModel\ For example, consider the Kallang River longitudinal profile, with both water level and discharge displayed, (on different y-axes), to evaluate these results.

Explore the results further: Longitudinal profiles along other river/drain branches Hydrographs at various points in network Water level plots Discharge through structures Compute and display 1D flood results Instabilities (in water level/discharge)?

Exercise 4: Improve model schematisation


For every cross-section inserted, a h-point is generated, (with an additional Q-point in between that h-point and the h-points upstream & downstream). Using more cross-sections to describe your channel geometry will thus result in more computation points & slower run times, and may also cause instabilities in your model, (especially if there are major changes in invert level or conveyance between the adjacent cross-sections). When inserting/editing cross-sections, its important to remember: Objective is not to represent the physical structure of the drain as realistically as possible Objective is to schematise the drain such that h and Q may be solved stably and accurately Only include necessary cross-sections: e.g. if you have a reach of drain with multiple crosssections of identical (or very similar) conveyance and invert level (or slope between them), it is not necessary to include them all. Perhaps youll only need one at the start and end of that section, (or one in the middle, for a very short section). (Note that additional Q- and hpoints will be automatically inserted during the simulation, based on the Max dx specified in the NWK file, such that h is solved for all along the drain, not only at the cross-sections.) Use Processed Data option in cross-section file, check variation in Conveyance between different cross-sections Consider an example from upstream of Singapore River:

Identical cross-sections, on a constant slope doesnt require so many cross-sections model can be made more efficient by simply removing unnecessary crosssections

Cascade can be better represented in other ways, (although slope here not too high)

For the first section of drain, a number of unnecessary cross-sections are present, which could be removed to improve the models run times. The reach also shows an example of a cascade, where a sudden drop in the drains invert level is represented using two crosssections very close together. While this is physically accurate, it is not well schematise the Q-point in between the two h-points at the cross-sections is solving for a very high slope and is sure to become unstable at some point. A potential solution for cascades is to model them using a Weir structure, (which solves for Q using the Energy Equation), which would more accurately account for the energy dissipation involved. Consider other locations throughout the model network that might be improved similarly?

Exercise 5: Representation of structures


Culverts and road crossings: sometimes represented using cross-sections very close together, with different cope levels. As discussed, cross-sections very close together can cause instability and will definitely slow the model down. These may be better represented using the Culvert structure option in the NWK file.

Cascades: as mentioned briefly, if it is important that particular cascades be included in the model, they might be represented using weirs to model the drop in invert levels (and associated energy dissipation). Barrage gates: Try to include the barrage gates in the model using the Control Structures option

You might also like