I have calculated the monthly mean from daily ESA CCI Soil moisture data, but even then there are many pixels with missing values for different months. I am concern about the data gaps in the ESA CCI Soil moisture data. When I apply the Mann-Kendall test for the trends of soil moisture for each pixel, only 0.5% of the pixels have showed some trend. I believe that this is because most of the pixels have nan values even for the monthly average so my Mann-Kendall test is unable to find trend for those pixels.
My question is, is it feasible to find the trend by Mann-Kendall ignoring NaN values, or is it recommended to fill the gaps first and then Mann-Kendall test?
Regards
Dear Alvin
Thanks for your valuable information. Actually, the missing values are random and most of the time > 5%.
In your case, if the soil texture and land use of your area is roughly homogeneous, you may estimate the gaps by using a geostatistic method such as kriging and then apply Mann-Kendall.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series