Hello, I am doing my MSc Thesis and using Mann-Kendall trend analysis for discharge flows.
To implement the Mann-Kendall test, is it obligatory to use uncorrelated time series? I checked for autocorrelation in my time series (1 lag autocorrelation and significance level 5%) and found that my time series is autocorrelated.
After implementing the original Mann-Kendall and Mann-Kendall Test with Trend-Free Prewhitening, I realised that the autocorrelation of the original Mann-Kendall had overestimated the trends.
However, should I include in the results for both the original Mann-Kendall and commend the difference with the Mann-Kendall Test with Trend-Free Prewhitening? Or is it wrong to use the results of the Original Mann Kendall because the time series is autocorrelated?
Thank you in advance for any help!
It is established that the original MK test fail to measure representative trends. Therefore, you should either remove autocorrelation or use a modified MK test ( g.g. a Modified MK test by Hamid and Rao, etc. ). It is seen that using a modified test gives better results than whitening.
Sheikh Hefzul Bari Thank you for your reply. However, if we have annual mean runoff values, it is sensible to have an autocorrelation in the time series for 1-lag (which means one year) since with this way the time series expresses the "seasonality" (every year we have almost the almost the same annual mean runoff values). As a result, why to pre-whitening (or modified and get rid of the autocorrelation)? Thank you for any help :)
Akrivi Alexandraki , Even annual data might have a correlation in my sense. For example, if your data has a clear trend over time it may show a correlation. Autocorrelation is the relation between lagged values. So, seasonality is not mandatory for autocorrelation.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing