Experiment Design

[kdorheim]

TODO need to add table describing the different calibration protocols to try out, instead of opening an issue per experiment will probably add the PR number to the table

Experiment	Description	Original PR	PR with expanded bounds
1	Recreate manuscript protocol	#15	x
2	Pre process temp data with 3 year running mean	#16	x
3	Pre process temp data with 5 year running mean	#16	x
4	Pre process temp data with 10 year running mean	#16	x
5	MSE weighting	#17	#19
6-8	NMSE and smoothing	#20	#20
9	NMSE modified to account for confidence interval	#21	#21
10	Optimizing ECS, NMSE accounting for confidence interval	#25	#25
11	Optimizing ECS & Alpha using NMSE accounting for CI	#27	#27 (also tried even larger bounds)
1-11	Rerunning experiments 1-11 after fixing temperature bug	#32	#32
12	Optimizing ECS & Alpha using NMSE for T, CO2, OHC (and accounting for CI)	#33	x
13	Optimizing ECS & Alpha using MVSSE for T, CO2, OHC	#34	x
14	Optimizing Alpha using NMSE for T, CO2, OHC (and accounting for CI)	#35	#35
15	Optimizing ECS & Alpha using MAE for T, CO2, OHC	#37	x
16	Optimizing ECS & Alpha using NMAE for T, CO2, OHC	#37	x