The area average problem

I just use the monthly u-wind data from NCEP/NCAR, it has four dimensions, namely lon,lat,level and time. I want to calculate the area average of (100E-120E , 0-20N). But the result is not very good. So I found that the process of read may have some problems. But it has no scale and add factors and I just used the common command. So is there someone have used the same dataset, please feel free to contact me! Thank you! It is some kind of emergency!

Answer 1

[Chad Greene]

Your question is unclear, so I will make some assumptions about what you are asking.

First, I think u is three dimensional, not four dimensional. Its dimensions are probably lat x lon x time, or lon x lat x time. Are lat and lon 1D or gridded 2D? If they're 1D, I often find it easiest to grid lat and lon using meshgrid. Then, find the row and column indices of lat and lon in your range of interest:

[row,col] = find(lat>=100 & lat<= 120 & lon>=0 & lon<=20);

It is not clear whether you want to bin all values in this lat/lon range to create a single average value, and it is not clear what you want to do about time. mean(u,3) will give you a 2D grid of averages across all time for each grid cell.

ubox = u(row,col,1);

Will give you a 2D grid of just the values in your lat/lon box for time 1. Then you can average all those values with

mean(ubox(:))

But again, you will have to be very clear about what values you are seeking.