Table 1 Spearman rank correlations among forcing variables and water fluxes and number of observations based on different observational or observation-driven datasets and the Budyko equation

Flux	Forcing	Source	Nr	Wet–warm (15%)		Wet–cold (23%)		Dry–cold (28%)		Dry–warm (34%)
				ρ_s	Count	ρ_s	Count	ρ_s	Count	ρ_s	Count
E_a	P	Budyko*²⁰	1	0.84	m.e.	0.83	m.e.	0.98	m.e.	1.00	m.e.
E_a	P	FLUXCOM⁴³	2	0.57	m.e.	0.76	m.e.	0.71	m.e.	0.88	m.e.
E_a	N	Budyko*²⁰	1	0.95	m.e.	0.99	m.e.	0.59	m.e.	0.79	m.e.
E_a	N	FLUXCOM⁴³	2	0.93	m.e.	0.94	m.e.	0.79	m.e.	0.91	m.e.
R	P	MacDonald⁴⁰	3	(0.0)	4	-	0	-	0	0.84	130
R	P	Moeck³⁹	4	-0.06	234	0.66	83	0.29	100	0.74	4790
Q	P	Budyko*²⁰	1	0.94	m.e.	0.87	m.e.	0.90	m.e.	0.99	m.e.
Q	P	GSIM^47,48	5	0.62	1259	0.71	1211	0.32	517	0.80	900
Q	P	GRUN⁴⁹	6	0.86	m.e.	0.74	m.e.	0.27	m.e.	0.94	m.e.
Q	N	Budyko*²⁰	1	0.45	m.e.	0.42	m.e.	0.11	m.e.	0.69	m.e.

The percentage of grid cells per climate region is given in brackets. The Budyko equation was forced per grid cell with the same forcing as the models (indicated by *) and thus covers approximately the same extent (except for cells with negative net radiation). The gridded datasets (FLUXCOM, GRUN) are available at the same resolution as the models and thus also cover approximately the same extent (except for non-vegetated areas in the case of FLUXCOM). This is indicated by m.e. for model extent. For datasets without matching precipitation data, we used GSWP3 reanalysis data. Nr corresponds to the numbers used in Fig. 4. The MacDonald rank correlation for the wet–warm region is shown in brackets because of the very small sample size; it is not shown in Fig. 4. Dashes (-) indicate that correlations could not be calculated because no observations were available. ρ_s denotes Spearman rank correlations.

Quick links

Search