Calculete monthly anomalies time series temperature

Hi!

I have this time serie (temperature_93_18.txt) and this monthly mean values (MonthlyAverage.txt) (Attached)

I want to calculete the anomalies for each value of the time serie, this mean to substract for each value of Junary of every year the same value from MonthlyAverage.txt data.

Any fast solution?!

[Image Analyst]

There are a variety of outlier functions, like isoutlier(), iqr(), mad(), etc. Have you tried any of them? What defines an anomalie for you?

[Pepe Grillo]

Hi, the anomalie is the difference betwen the value and the mean (x-x`)

What I want is from the time serie make the difference between values of each day per month and the mean value in MonthlyAverage.txt and for all years,

I didnt try or know that functions I wolud have a look in the meantime

thanks for your reply

[Pepe Grillo]

I attached now the date ant temp in one file just in case

Thanks!

Answer 1

[Peter Perkins]

You are not doing yourself any favors by using raw numeric vectors with no timestamps. This is almost trivial using datetime and a timetable. Here's one way:

>> Temp = readmatrix("temperature_93_18.txt");
>> AvgTemp = readmatrix("MonthlyAverage.txt");
>> year0 = 1993;
>> temps = timetable(Temp,'RowTimes',datetime(year0,1,1:height(Temp)));
>> head(temps)
ans =
  8×1 timetable
       Time        Temp
    ___________    ____
    01-Jan-1993    6.57
    02-Jan-1993    6.53
    03-Jan-1993    6.14
    04-Jan-1993    6.02
    05-Jan-1993    6.02
    06-Jan-1993    5.92
    07-Jan-1993    5.88
    08-Jan-1993    6.58
>> Month = ["Jan";"Feb";"Mar";"Apr";"May";"Jun";"Jul";"Aug";"Sep";"Oct";"Nov";"Dec"];
>> avgTemps = table(Month,AvgTemp)
avgTemps =
  12×2 table
    Month    AvgTemp
    _____    _______
    "Jan"    5.0984 
    "Feb"    5.2968 
    "Mar"    4.9642 
    "Apr"    4.1152 
    "May"    3.2809 
    "Jun"    3.0204 
    "Jul"    2.8472 
    "Aug"    2.7302 
    "Sep"    2.4709 
    "Oct"    2.8043 
    "Nov"    3.1844 
    "Dec"    4.2404 
>> t = temps;
>> t.Month = month(temps.Time,"shortname");
>> head(t)
ans =
  8×2 timetable
       Time        Temp     Month 
    ___________    ____    _______
    01-Jan-1993    6.57    {'Jan'}
    02-Jan-1993    6.53    {'Jan'}
    03-Jan-1993    6.14    {'Jan'}
    04-Jan-1993    6.02    {'Jan'}
    05-Jan-1993    6.02    {'Jan'}
    06-Jan-1993    5.92    {'Jan'}
    07-Jan-1993    5.88    {'Jan'}
    08-Jan-1993    6.58    {'Jan'}
>> t = join(t,avgTemps,"Key","Month");
>> head(t)
ans =
  8×3 timetable
       Time        Temp     Month     AvgTemp
    ___________    ____    _______    _______
    01-Jan-1993    6.57    {'Jan'}    5.0984 
    02-Jan-1993    6.53    {'Jan'}    5.0984 
    03-Jan-1993    6.14    {'Jan'}    5.0984 
    04-Jan-1993    6.02    {'Jan'}    5.0984 
    05-Jan-1993    6.02    {'Jan'}    5.0984 
    06-Jan-1993    5.92    {'Jan'}    5.0984 
    07-Jan-1993    5.88    {'Jan'}    5.0984 
    08-Jan-1993    6.58    {'Jan'}    5.0984 
>> temps.Anomaly = t.Temp - t.AvgTemp;
>> head(temps)
ans =
  8×2 timetable
       Time        Temp    Anomaly
    ___________    ____    _______
    01-Jan-1993    6.57     1.4716
    02-Jan-1993    6.53     1.4316
    03-Jan-1993    6.14     1.0416
    04-Jan-1993    6.02    0.92161
    05-Jan-1993    6.02    0.92161
    06-Jan-1993    5.92    0.82161
    07-Jan-1993    5.88    0.78161
    08-Jan-1993    6.58     1.4816

Answer 2

[Image Analyst]

isoutlier(), rmoutlier(), or mad() should work. Write back if they don't.

[Pepe Grillo]

I see this functions are more for outliers to remove them,

Maybe I didnt explain so well, let me try again

so lets say we have this list of daily temperaute

date temp

1993 01 01 5.2

etc

1993 12 31 8.1

1994 01 01 5.1

etc

and so on untill 2018 12 31

So for each day of Junary want to substract this value lets say 3.4

for each day of each year of March substract 2.1 and like that for all the serie,

Thanks for your replys!

[Pepe Grillo]

So this functions didnt work for what I want, maybe other idea to make monthly remove of the mean to the serie?

Thks!

[Image Analyst]

Make it easy for people to help you. Give us a start by giving your code to read in the various text files and plot them.