Hello.
I understand how to detrend annual climate data, e.g. temperature, rainfall, and radiation.
However, for long-time series of daily climate data, what should I do? Maybe some software does that?
Hope to hear your opinions.
For temperature and solar radiation, it could be the same with that for annual values. Due to dependence of temperature (solar radiation) trends on seasonality, you have to do it on a basis of each of 365 days for the whole time period. For precipitation, it should be a little difficult, because precipitation is not a continuous variable, but you could simply proportionately allocate the annual adjustment to each of the days with precipitation based on the amount of daily precipitation.
You can use "cdo" detrend function, "NCL", "Python" can also do that.
Hello,
you can use the detrend command of the matlab software to detrend your climate data. There are two options (1 for detrending linear trends and 2 for quadratic trends).
Simple linear detrending works for an "interval" variable like temperature (and is what you get e.g., from CDO detrend) - even for long climate data, though in that case you should detrend each day of the year separately, i.e., detrend all 1st Januarys, then all 2nd Januarys, etc. Otherwise, try to remove the seasonal cycle first.
However, a "ratio" variable like precipitation should probably be de-trended differently than an "interval" variable like temperature. If you try simple linear detrending with precipitation, you will most likely get some unphysical negative rainfall, and turn some dry days into wet ones. So a "multiplicative" or geometric de-trending function is needed instead. See the attached presentation for the ideas and the formula:
Pdetrended = Poriginal * [1 + (Pmean – Plinear-estimate)/Pmean ]
This avoids negative precipitation and avoids turning dry days into wet ones, but it does not preserve the original mean (Pmean). If that is important, compute the mean of the new Pdetrended series, then scale all Pdetrended values by the factor (Pmean/Pdetrended_mean). This produces a final de-trended precipitation series that preserves the original mean.
That method for detrending precip. (avoiding negative precip., conserving dry days as dry days, and conserving the mean) was recently published, as part of the TRANSLATE project in: https://www.frontiersin.org/articles/10.3389/fclim.2023.1166828/full . Please cite that source if you use this method. It's a good, simple method!
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence