Can we use Mann Kendall trend Test for 3D matrix??

Helo

I have temperature data of dimension (61 x 61 x 45). where (lat x lon) is of (61 x 61) and temperature is on 3rd dimension (45).

Can I calculate MK trend test for each grid?

Any help !!

[David Goodmanson]

Hi Jitesh,

Certainly you can do a separate Mann Kendall test for, say, each and every point on the 61x61 grid as you run through the length 45 dimension (time?). Or a take the mean temperature of each grid, and then run through the length 45 dimension. That gives a single test result. Or divide the grid into subblocks, take the mean temperatue of each one, do a Mann Kendall test on each subblock. Are any of these situations what you have in mind?

I assumed a test along a particular dimension but this would apply to a grid of any two axes while running along the third.

[Jitesh Dadich]

Hi David, Thanks for your concern. I will try to do as whatever you said, but as my approach every grid in the box 60x60 will have 45 values of temperature. Now i have to calculate the mk trend for each grid. Since the function MK trend test which we have in our matlab library cant calculate this as i see it. And also do we use the mean of temperature while finding mk trend..? Could you help me with any script like it? Thanks

[David Goodmanson]

HI Jitesh,

Are you using an MK trend test from the Mathworks file exchange? I don't see it in the statistics toolbox. I imagine you have MK code that works on a single vector of data.

It's not totally clear but it sounds like you have a 61x61 grid with 60x60 temperature values. If you are going to do the test for each grid square (3600 in all), then you can set up a 'for' loop, run the test 3600 times and save the result in a 60x60 matrix. A bunch of detail there, with perhaps a lot of variation in the result. Another more global approach is to take the mean of each grid, creating just 45 temperature values, and run the MK test one time on that. Then the whole works is boiled down to one number. Not being in the field, I don't know what the most sensible approach is.

[Jitesh Dadich]

I have solved it .

thanks David Goodmanson !!

[Abel Diaz]

Hello Jitesh, can i ask how do you solved it? thanks in advanced.