I have collect several data sets for a small region in Germany called Hessisches Ried (about 90 km²) from different sources for a time period from 2015 to 2019. Daily rainfall amount & daily evapotranspiration and for irregular time steps soil moisture data from radar satellite imagery.
My goal is to apply an unsupervised classification of at least the soil moisture data (better with all these parameters) to derive a soil property map. The ability to store moisture over time depends mainly on the amount of sand, silt, clay in the soil. So the idea is that different kinds of soil shown different pattern in soil moisture behavior. But here I am stuck. I can't find any module in python that deals with unsupervised multivariate time series classification or clustering.
The data currently stored in a xarray gridded dataset with dimension (latitude, longtitude, time)
What kind of method can you recommend for such a task? Somebody recommends the SOFM algorithms - Self organized Feature Mapping. But I am only able to deal with python modules. I don't have the ability to implement such techniques by myself.
I find this module called tslearn. I will try it out and let you know if I am able to do some classification or clustering with it.