Message/Author |
|
|
Posted from Conversations with Greg Hancock & Ken Evans, regarding using CAESAR for overland flow/soil erosion prediction. About tuning CAESAR. First some context, It’s a catchment model concieved for green temperate UK catchments, where slope processes are somewhat more benign than in arid/semi-arid environments, so it has more focus on the channel processes - with the slope processes being creep and landslips. If there is no creep or slips then the only way material will travel from slope to channel is if there is enough overland flow for it to get there. So, I think there are two ways you can increase the input from the slopes (which I think is probably why CAESAR may be under-predicting compared to some other models e.g. RUSLE). Firstly, the creep parameter could be tweaked up to factor in some overland flow terms. Creep is simply Slope*Timestep*Constant In the code this is slope * time(in years) * constant(the constant inputted) / DX(grid cell size). I suspect this could be tweaked to represent a different slope term quite easily (e.g. add an exponential exponent etc..). Secondly, Increase the overland flow. In CAESAR there is one input paramter (in the 'numerical' tab) that acts as a threshold, above which the model calculates a flow depth, thus overland flow, thus fluvial erosion. Its in the 'Min Q for depth calc' box. Above this discharge a grid cell will calculate a flow depth. Now, it’s a bit of a rule of thumb job here as what to use. But I take the view that if you have a 10m wide cell - and have 1 cumec flowing over it at 1ms-1, then you have 10cm of water depth - enough to move things. However, 0.01 cumucs, 0.1ms-1 and you have 1cm.. A bit more of a tougher call. In practice, you can run the model and when a storm/flood comes along, if Min Q is set lower you can see that more of the catchment wets up. This will mean that (probably) more material from the slopes will be swept into the channels - but the downside is that the model will run more slowly... Because more wet cells = more calcualtions = slower CAESAR. |
|
|
Some comments from a email conversation with Katharine, T Hi Tom, Thanks for the reply. I have a few more questions/confirmations...! 1. Is root just set to 7.07 as a default value in the variables? The actual value of 'root' is calculated from the cell size in the header file when the ascii loads up? Yup - just a default value. 2. Does dividing by root instead of cell size throughout means that the diagonal cells get less water/less creep taking place than the up,down,left,right cells? Yes - because the distance is longer as they are diagonal. Its so the slopes are properly calcualted. 3. Within slide_gs, are the temp/diff/amount variables used to switch sediment/soil/water from one cell to another? Yes - slide_gs ensures that grainsizes from one cell are transferred to another when there is a mass movement (soil creep, soil erosion or landslide). The name of the function represents this - slide and gs for grainsize. 4. From what you were saying about area[x,y] , does this mean that this is the contributing area above a cell ie. area[x,y] are the cells contributing to moving sediment and water into any one cell? Not exactly - they are the number of cells that drain into that cell. the contributing area. 5. The last part of the soil erosion rate equation, you said was slope length - does this just mean the cells contributing directly 'in a straight line' above the cell in question ie. [x,y+1], [x,y+2], [x,y+3], [x,y-1] etc.? rather than an area (such as area[x,y]) whereby the cells such as [x-1,y] and [x+1,y] etc. can contribute to slope length? Its very hard to calculate the exact slope length - as this varies very rapidly from one part of the catchment to another - how do you determine which top of a slope you measure to! So the easy way around this is to use the square root of the drainage area, hence the Math.Pow(area[x,y],0.5) line... which returns the square root of area[x,y]. Sorry for all the questions, just trying to get the code clear in my own mind! Thanks Katharine Subject: RE: Soil erosion rate Hi Katharine, Time is the time in years... area[x,y] is the drainage area (in cells) above a point. The last part of the equation here returns slope length (simmilar to square root of the drainage area) Tom Sent: 08 February 2008 15:25 To: Thomas J Coulthard Subject: Soil erosion rate Hi Tom, I've got 2 questions regarding the soil erosion rate parameter within CAESAR (from the line below) temp = ((elev[x, y] - elev[x, y - 1]) / DX) * SOIL_RATE * time / DX * Math.Pow(area[x,y]*DX*DX,0.5); 1) What is 'time' and why is it set to 0.028 in particular? 2) What is area[x,y]? Our thoughts are that it might be if certain cells are wet/dry OR is there water amount attached to each cell for this part? Is that right? |
|
Back to top |
|