netcdf_matlab_examples/maximenko_drifter/README.txt


==============================
Email from Nikolai Maximenko :
==============================


http://iprc.soest.hawaii.edu/~nikolai/ForJohnMarshall_1.dat.gz

It's written as
 
      write(2,'(f7.3,f8.3,2f8.2,i5,2f9.2)') x,y,u,v,num,uu,vv
      
The only difference from the previous file is two added columns
containing <U'U'> and <V'V'> in (cm/s)^2, so that EKE = (<U'U'> +
<V'V'>)/2 .

Please notice that EKE requires more data than <V>, so that not all
locations are good for the estimate. (For instance, if num = 0,
<U'U'>=<V'V'>=0 .) Simple criterion that I would suggest to select bins
with sufficient statistics
is:

- estimate the time that drifter typically spends within the bin:

T = 0.25 * 111 km / |<V>|
(|<V>|=sqrt(u^2+v^2) and u and v are in columns 3 and 4)
(Other estimate can be sqrt(u^2+v^2+uu+vv). )

- estimate the number of independent measurements available for this bin

N = num * 6 hours / T
(num is in column 5)

- select bins with N > N_0. I would expect that N_0 = 3 - 5 would  give
a pretty nice figure of the EKE. 

More accurate estimate of the "number of independent measurements" would
require information on time scales of main processes at given locations
(they are 
obviously different between the areas where ENSO or small eddies
dominate)

Comparison with the geostrophic velocities from satellite altimetry may
require some extra caution, like:

1. To avoid biases due to nonuniform drifter ensemble, altimetry
velocities 
should be averaged over the drifter times and locations.

2. Ekman velocities should be subtracted from drifter data.

When doing qualitative, illustrative, comparison between drifters and
altimetry these two issues can be considered as a potential sources of
discrepancies.

Good luck and Merry Christmas,

Nikolai
1	edhill	1.1
2			==============================
3			Email from Nikolai Maximenko :
4			==============================
5
6
7			http://iprc.soest.hawaii.edu/~nikolai/ForJohnMarshall_1.dat.gz
8
9			It's written as
10
11			write(2,'(f7.3,f8.3,2f8.2,i5,2f9.2)') x,y,u,v,num,uu,vv
12
13			The only difference from the previous file is two added columns
14			containing <U'U'> and <V'V'> in (cm/s)^2, so that EKE = (<U'U'> +
15			<V'V'>)/2 .
16
17			Please notice that EKE requires more data than <V>, so that not all
18			locations are good for the estimate. (For instance, if num = 0,
19			<U'U'>=<V'V'>=0 .) Simple criterion that I would suggest to select bins
20			with sufficient statistics
21			is:
22
23			- estimate the time that drifter typically spends within the bin:
24
25			T = 0.25 * 111 km / \|<V>\|
26			(\|<V>\|=sqrt(u^2+v^2) and u and v are in columns 3 and 4)
27			(Other estimate can be sqrt(u^2+v^2+uu+vv). )
28
29			- estimate the number of independent measurements available for this bin
30
31			N = num * 6 hours / T
32			(num is in column 5)
33
34			- select bins with N > N_0. I would expect that N_0 = 3 - 5 would give
35			a pretty nice figure of the EKE.
36
37			More accurate estimate of the "number of independent measurements" would
38			require information on time scales of main processes at given locations
39			(they are
40			obviously different between the areas where ENSO or small eddies
41			dominate)
42
43			Comparison with the geostrophic velocities from satellite altimetry may
44			require some extra caution, like:
45
46			1. To avoid biases due to nonuniform drifter ensemble, altimetry
47			velocities
48			should be averaged over the drifter times and locations.
49
50			2. Ekman velocities should be subtracted from drifter data.
51
52			When doing qualitative, illustrative, comparison between drifters and
53			altimetry these two issues can be considered as a potential sources of
54			discrepancies.
55
56			Good luck and Merry Christmas,
57
58			Nikolai