Implement LULC classification uncertainty cut-off

we use a user input classification uncertainty cut-off

• everything above is assumed to be “real”
• everything below (unknown) will be calculation-filled with the assumption that it contains the same emission distribution (test the distribution beforehand, test for normal distribution)
• the emission distribution will be reported with error bars (e.g. 1sd, 2xsd …)

Remarks from DFNS presentation June 2024:

• accuracy of LULC classification is just for one time step, so what is the propagated error for the LULC changes?
• provide uncertainty measure for emission estimates