根据日期间隔汇总值

Question

I have a database containing monthly precipitation values at some measurements locations. The structure of my table is:

describe   pp_lunare;
Field      Type       Null
ID         int(11)    NO
DATA_OBS   date       NO
PLUTON     float      NO
LEGHIN     float      NO
DUMBRAVA   float      NO

A sample of my data:

ID    DATA_OBS     PLUTON    LEGHIN   DUMBRAVA
1    1977-01-01    14.4      33.3     25.1
2    1977-02-01    18.7      12.9     13.2
3    1977-03-01    32.8      26.7     18.3
4    1977-04-01    109.6     123.8    140.6
5    1977-05-01    98.5      104.7    59.9
6    1977-06-01    192.9     172.8    66.6
7    1977-07-01    101.4     85.8     79.4
8    1977-08-01    116.4     103.3    105.7
9    1977-09-01    54.5      47.4     51.8
10   1977-10-01    23.6      15.6     11
11   1977-11-01    59.7      44.3     29.7
12   1977-12-01    28.7      13.1     10

In my case I need to get the sum of the precipitation for every column at every 3 months something like this:

ID   DATA_OBS      PLUTON    LEGHIN    DUMBRAVA
1    1977-03-01    65.9      72.9      56.6
2    1977-06-01    401       401.3     267.1
3    1977-09-01    272.3     236.5     247.9

and so on...

Thanks.

Answer 1

You can get the month from your date with month(data_obs) which returns a month from 1 to 12. Convert this to a value for the quarter by doing something like floor((month(data_obs) - 1)/3) as quarter to get quarters from 0 to 3.

For example, select data_obs, concat(year(data_obs), floor((month(data_obs) - 1)/3)) as quarter from pp_lunare; should show you both the original date and the derived quarter side by side.

Then group by this new value and sum the rest:

select concat(year(data_obs), floor((month(data_obs) - 1)/3)) as quarter, sum(pluton), sum(leghin), sum(dumbrava)
from pp_lunare
group by quarter
order by quarter;

If you want the exact date format you had in the question, you'll have to do a little bit of string manipulation (add 1 to the quarter above, multiply that by 3, pad with 0s, then tack on a "-01" for the date).

Answer 2

You can get the month data at first, and convert it into [quarter], which is from 0 to 2. Then you can create three temporary tables given [quarter]=0,1,2, and the easy and final part will be to add up the i_th row in these tables, the result will be just like below. And you can also have the data_obs data as well.

ID DATA_OBS PLUTON LEGHIN DUMBRAVA 1 1977-03-01 65.9 72.9 56.6 2 1977-06-01 401 401.3 267.1 3 1977-09-01 272.3 236.5 247.9