| Message/Author |
|
|
|
|
Message from Shulong
Questions are as follows:
1. A new and unique method for radius of curvature. In your paper , you say 'Passing of a 3*3 filter
over edge cells ,counting number of dry cells and wet cells in filter.' but it will have no way to pass of a
3*3 filter for boundary cells as shown cell_1Acell_2 Acell_3 Acell_4Acell_5, so I can not calculate their
radius of curvature in your method. What is your way about them?
2. About wet cells' radius of curvature, you say ' This is carried out by applying an averaging filter
across all wet cells to smooth the values across the channel '. I can not think out your operation for wet cells.
How can I calculate them ?
3. About the operation of the equation ' ', you say 'where n and n-1
respectively denote the donor cell and the receiving cellf, but how can I know who is donor cell and who is
receiving cell?
With kind regards, |
|
|
|
|
OK,
Some answers:
1. At the edges, all the values are set to 0 - so the diagram in the paper is a little misleading (its supposed to be a central section of a dem). This simply stops lateral erosion at the edges of the DEM
2 and 3. Here we use the smoothed edge term to determine a cross stream gradient. For example, on the outside bank you may have a value of +1 on the inside -1 and three wet cells in between (the channel - wet cells being those with water in them.) If you then run a smoothing filter over those wet cells inthe centre, it will determine a distance weighted average value between the +1 and -1, so after a few passes of the filter you would have
inside bend bank cell -1
inside wet cell -0.5
centre wet cell 0
outside wet cell -0.5
outside bend bank cell 1
You can then use these values or this gradient between outside and inside bend to drive or route sediment across the channel - thus allowing a point bar to develop in the inside edge.
Tom |
|
|
 Lui Shulong (Shulong) posted on Wednesday, September 03, 2008 - 03:25 am
|
|
|
Dear Mr.Coulthard,
Thanks for your wondeful and enlightening reply. You are so careful and patient for a inquisitive student that I can see what is scientific spirit. Maybe for you it is only a reply , your reply will have an impact on my life. You are a excellent example to us. Thank you.
However, there are still some I do not understand, and I would like to make a recommendation to you.
Questions:
1. You say ¡®on the outside bank you may have a value of +1 on the inside -1¡¯, the operation works on the premise that you had known who is opposite side (in your reply, you mean : cell_+1 is opposite to cell_-1, but why ? What is this definition based on?).
As shown below Figure 1 , what way can I use to make a judgement who is opposite cell_1 (cell_2 ? cell_3 ? cell_4 ? )? What is the criterion ? Similarly, who is opposite cell_5(cell_6 ? cell_7 ? cell_8 ? )?
****(Figure_1 has been emailed to you)
2. You say ¡®it will determine a distance weighted average value between the +1 and -1, so after a few passes of the filter¡¯. Since you ¡®average value¡¯,why ¡®a few passes¡¯?
Recommendation & Questions:
Why do you not to use small-scale cell ? Small-scale may give better results.
With kind regards,
Liu Shulong |
|
|
|
|
Hi Lui - sorry is it better to say Liu or Shulong?
OK.
1. If you use a 3 by 3 filter and take an average of all the values in that area (assuming you cannot change the values of the bank cells) - then pass that filter over the whole domain several times, then the code need not know which is the opposite side, or the nearest bank, it will simply interpolate the bank values across the channel.
2. The number of passes is required to ensure that the values *above* have been correctly interpolated. In the code it does this by checking how much values change by between the passes of the filter and when they stop changing by much (0.0001 or somethinglike that) then it stops
Tom |
|
|
| Lui Shulong (Shulong) posted on Thursday, September 04, 2008 - 09:25 am
|
|
|
Dear Mr.Coulthard,
You really a British gentleman. named me Liu, it is ok.
(As some formula can not be displayed in your 'CAESAR discussion board', so I e-mail to you )
I use 'fortran 77' to write my code, so please forgive me if I asked you some low-level questions. Now I'm learning c#, so as to understand your code and discuss better.
According to my understanding of the CAESA'S principles, there are some places seems to be abrupt, e.g. the Equation of a lateral sediment flux , Is it a physical meaning? Please forgive me and allow me to trouble your CAESAR.
Can I continue to ask some temporary low-level questions?
1. I still do not quite understand the specific operations about wet cells' radius of curvature,but, In order not to occupy your valuable time frequently I will delve into your code.
you say 'However, the positive(outside) and negative(inside) values are used to determine a cross-stream gradient of curvature. This is carried out by applying an averaging filter across all wet cells to smooth the values across the channel. Results are shown in Figure 5b. This cross-stream gradient of curvature is then used to calculate a lateral sediment flux .'. How dose determine a cross-stream gradient of curvature? In your Equation ,you say ‘where n and n-1 respectively denote the donor cell and the receiving cell.’ But n have eight cell around it, Who will serve as the n-1, What is the criterion?
2. Why do you not to use small-scale cell ? Small-scale may give better results.
Thanks again, you are so patience and tolerance for me, for a student.
With kind regards,
Liu |
|
|
|
|
Hi Liu,
OK, 1. n and n-1 refer to n being the centre cell and n-1 any of the neigbours. So it examines the gradient between the cell in question and all of those surrounding it.
2. Maybe - maybe not! There is a genuine question as to whether or not there is an ideal cell size for cellular models. There is a natural controlling length in natural channels (due to turbulent structures - burst, sweeps etc..) that means many facets of river morphology can be scaled to the width for example. e.g. pool riffle or meander spacing. This is hard to do within a cellular framework - so too small a grid cell and things get too complex and how does the river know wheterh it is a small cell in a small stream or a large cell in a large stream etc.. |
|
|
| Lui Shulong (Shulong) posted on Friday, September 12, 2008 - 07:37 am
|
|
|
Dear MR.Coulthard,
Just read your paper 'A CELLULAR MODEL OF HOLOCENE UPLAND RIVER BASIN AND ALLUVIAL FAN EVOLUTION'. A number of problem have arisen.
QUESTION:
In this paper you mentioned the issue of 'the described routing algorithm'.
1.Is 'the described routing algorithm' suitable for all river pattern?
2.What physical mechanism are you based on?
There are still some problems, but first and foremost is the two macro question. In our university we are banned from surfing foreign web sites, so I cherish your help especially. As maybe you are busy recently, I have not received your CAESAR's manual.
With kind regards,
Liu |
|
|
| Lui Shulong (Shulong) posted on Friday, September 12, 2008 - 08:28 am
|
|
|
Dear MR.Coulthard,
I still have an intrusive question:
Can you tell me that which subroutine is the curvature solution operation? and which subroutine is the fluent direction's determination operation?
Sincerely thank you and hope you can continue to help me.
Liu |
|
|
| Back to top |
|
