Table 1.

Metrics used to assess the ability of our ensemble estimate of historical temperatures and precipitations to replicate observed conditions.

Region	Temperature					Precipitation
	ρpb	M	RMSEw	rSD	md	ρpb	M	%-bias	rSD	md
Global	0.99	91.3	1.98	1.03	0.95	0.89	64.5	10.9	0.93	0.76
High-North	0.97	83.3	1.99	1.08	0.89	0.75	51.5	21.0	0.98	0.61
Mid-North	0.97	84.3	2.17	0.99	0.89	0.91	63.8	15.5	0.99	0.78
Mid-South	0.96	78.2	1.70	0.97	0.85	0.81	52.8	20.9	0.90	0.65
High-Tropics	0.81	61.1	1.83	0.86	0.72	0.83	57.4	0.10	0.92	0.73
High Latitudes*	0.96	78.9	2.28	1.12	0.87	0.72	66.1	13.6	0.94	0.61
Mediterranean and Sahara*	0.96	81.2	1.53	0.87	0.85	0.97	77.1	−2.1	0.97	0.86
North America (East)*	0.99	89.0	0.99	0.94	0.92	0.89	62.5	6.4	0.79	0.72
Southern Africa and West Indian Ocean*	0.79	60.4	1.72	0.89	0.75	0.90	48.1	31.5	0.76	0.55
Australia and New Zealand*	0.99	78.9	1.42	0.94	0.83	0.93	66.1	13.2	0.79	0.71
Neotropical#	0.95	79.1	2.10	0.91	0.88	0.67	40.5	−2.1	0.77	0.59
Oriental#	0.86	76.3	2.34	1.05	0.81	0.84	61.9	−3.4	0.95	0.75
Palearctic#	0.98	86.5	2.13	1.05	0.91	0.88	57.5	22.8	0.89	0.67

ρ_pb = percentage bend correlation⁷⁵, where higher values indicate more agreement between observed and simulated conditions; M = m statistic⁷⁶ (×100), where higher values indicate more agreement between observed and simulated conditions; RMSE_w = Root-Mean-Square-Error weighted by latitude, lower values indicate better agreement between simulated and observed conditions; rSD = ratio of standard deviations, values closer to 1 indicate better agreement between simulated and observed conditions; md = modified index of agreement⁷⁷, values closer to 1 indicate better agreement between simulated and observed conditions; %-bias = percentage bias, the tendency of the simulated values to be larger or smaller than observed. *IPCC AR5 regions from van Oldenborgh, et al.²⁴. ^#Biogeographic realms following Holt, et al.²⁵.