CMIP water budget

Model diagnostics

Figure 1. The global water budget as represented in CMIP coupled climate models. Water reservoirs are shown in boxes and arrows represent the fluxes between them. Bold text indicates the name of the corresponding (monthly timescale) model diagnostic/s (see definitions in Appendix below).

Figure 1 is an attempt to map the global water mass budget using CMIP6 diagnostics. As it stands, the suite of CMIP6 diagnostics is not sufficient to characterize the full mass budget (shown on the left side of vertical dashed line). One of the reservoirs doesn't map to any diagnostic (mass of icebergs) and a number of other budget terms weren't listed as high priority diagnostics and thus many/most modeling groups didn't archive them (e.g. mrtws).

It is possible, however, to fully characterize and assess closure of the mass budget for the atmosphere and ocean (shown on the right side of vertical dashed line):

• A cumulative atmospheric mass flux (evspsbl - pr) should be reflected in a corresponding change in storage (clwvi + prw; or just prw since globally integrated clwvi is negligible in comparison)
• A cumulative ocean mass flux (wfo) should be reflected in a corresponding change in ocean mass (masso)

The residual between the moisture flux out of the atmosphere (pr - evspsbl) and into the ocean (wfo) on decadal and longer timescales represents some combination of a change to terrestrial water storage, the volume of global ice (in the form of ice sheets, sea ice and icebergs) and/or mass leakage within/between the land and/or ice components of the model.

Budget closure

Figure 2. Annual-mean, globally integrated energy and mass budget terms for the ACCESS-CM2 pre-industrial control experiment. The dedrifting is performed by fitting (and then subtracting) a cubic polynomial to (from) the corresponding timeseries in the panel above. A 10-year running mean has been applied to the timeseries in panel (f). If a model is mass/water conserving, then the timeseries in panels (b), (c), (e) and (f) should lie on top of one another.

Most models behave similar to the one shown in Figure 2. Key points are as follows:

• Neither the atmosphere or ocean mass budgets close before de-drifting.
• In most models, the cumulative change in water flux into the ocean/atmosphere is much larger in magnitude than the drift/change in the mass of the ocean/atmosphere.
• After dedrifting there is typically good closure of the ocean mass budget but not the atmosphere.

To put the atmospheric mass drifts in context, they can be compared to recent trends in atmospheric water vapor. For the 2000-2014 period, the largest trend estimate from the various observational and reanalysis products is +1.84 % per decade (Chen and Liu, 2016). Assuming that the mass of water vapor in the atmosphere is approximately 1.2 x 10^16 kg, this corresponds to a trend of 2.2 x 10^14 kg yr-1. The model drift in atmospheric water vapor represents a negligible fraction of these observed trends (e.g. ACCESS-CM2 is 3.2 x 10^11 kg yr-1), but the magnitude of the linear trend in the cumulative water flux into the atmosphere (E-P) approaches or is even larger than observed trends for some models (e.g. ACCESS-CM2 is -1.8 x 10^14 kg yr-1).

Looking at the E and P timeseries separately, in general they will trend/drift in the same direction as the atmospheric water content (e.g. in Figure 3 all timeseries show an increase over time), but the slight offset between E and P accumulates over time, hence the budget non-closure.

Figure 3. Individual atmospheric mass budget terms for the ACCESS-CM2 pre-industrial control experiment. All timeseries represent the annual-mean global integral.

All the models show relatively little drift in atmospheric water content, which is something that was noted by Liepert and Previdi (2012) for the CMIP3 ensemble. They note that the extra "ghost" moisture that must be added to the atmosphere to compensate for the E-P leakage (the majority of models have larger global P than E, just like ACCESS-CM2) is typically associated with a latent energy release (of condensation) of about 0.1 W m-2. Some models approach or exceed this value in CMIP6, but most are much smaller. In other words, for most CMIP6 models the latent heat release associated with condensation of this "ghost" moisture does not explain an appreciable fraction of the energy leakage between the top of the atmosphere and ocean surface.

Appendix

Variable name	Units	Long name	Standard name	Model realm	Comments
clwvi	kg m-2	Condensed Water Path	atmosphere mass content of cloud condensed water	atmos	Mass of condensed (liquid + ice) water in the column divided by the area of the column (not just the area of the cloudy portion of the column). Includes precipitating hydrometeors ONLY if the precipitating hydrometeor affects the calculation of radiative transfer in model.
evs	kg m-2 s-1	Water Evaporation Flux Where Ice Free Ocean over Sea	water evapotranspiration flux	ocean	Computed as the total mass of water vapor evaporating from the ice-free portion of the ocean divided by the area of the ocean portion of the grid cell.
evspsbl	kg m-2 s-1	Evaporation Including Sublimation and Transpiration	water evapotranspiration flux	atmos	At surface; flux of water into the atmosphere due to conversion of both liquid and solid phases to vapor (from underlying surface and vegetation)
ficeberg	kg m-2 s-1	Water Flux into Sea Water from Icebergs	water flux into sea water from icebergs	ocean	Computed as the iceberg melt water flux into the ocean divided by the area of the ocean portion of the grid cell.
flandice	kg m-2 s-1	Water Flux into Sea Water from Land Ice	water flux into sea water from land ice	ocean	Computed as the water flux into the ocean due to land ice (runoff water from surface and base of land ice or melt from base of ice shelf or vertical ice front) into the ocean divided by the area ocean portion of the grid cell.
friver	kg m-2 s-1	Water Flux into Sea Water from Rivers	water flux into sea water from rivers	ocean	Computed as the river flux of water into the ocean divided by the area of the ocean portion of the grid cell.
fsitherm	kg m-2 s-1	Water Flux into Sea Water Due to Sea Ice Thermodynamics	water flux into sea water due to sea ice thermodynamics	ocean seaIce	Computed as the sea ice thermodynamic water flux into the ocean divided by the area of the ocean portion of the grid cell.
masso	kg	Sea Water Mass	sea water mass	ocean	Total mass of liquid sea water. For Boussinesq models, report this diagnostic as Boussinesq reference density times total volume.
mrtws	kg m-2	Terrestrial Water Storage	land water amount	land	Mass of water in all phases and in all components including soil, canopy, vegetation, ice sheets, rivers and ground water.
pr	kg m-2 s-1	Precipitation	precipitation flux	atmos	At surface; includes both liquid and solid phases from all types of clouds (both large-scale and convective)
prra	kg m-2 s-1	Rainfall Flux	rainfall flux	atmos	Computed as the total mass of liquid water falling as liquid rain into the ice-free portion of the ocean divided by the area of the ocean portion of the grid cell.
prsn	kg m-2 s-1	Snowfall Flux	snowfall flux	atmos	Computed as the total mass per unit time of solid-phase precipitation falling into the ice-free portion of the ocean divided by the area of the ocean portion of the grid cell. (Snowfall flux includes all types of solid-phase precipitation.)
prw	kg m-2	Water Vapor Path	atmosphere mass content of water vapor	atmos	Vertically integrated through the atmospheric column.
sidmassevapsubl	kg m-2 s-1	Sea-Ice Mass Change Through Evaporation and Sublimation	water evapotranspiration flux	seaIce	The rate of change of sea-ice mass change through evaporation and sublimation divided by grid-cell area.
siflfwbot	kg m-2 s-1	Freshwater Flux from Sea Ice	water flux into sea water due to sea ice thermodynamics	seaIce	Total flux of fresh water from water into sea ice divided by grid-cell area; This flux is negative during ice growth (liquid water mass decreases, hence upward flux of freshwater), positive during ice melt (liquid water mass increases, hence downward flux of freshwater).
siflfwdrain	kg m-2 s-1	Freshwater Flux from Sea-Ice Surface	water flux into sea water due to surface drainage	seaIce	Total flux of fresh water from sea-ice surface into underlying ocean. This combines both surface melt water that drains directly into the ocean and the drainage of surface melt pond. By definition, this flux is always positive.
simass	kg m-2	Sea-Ice Mass per Area	sea ice amount	seaIce	Total mass of sea ice divided by grid-cell area.
sipr	kg m-2 s-1	Rainfall Rate over Sea Ice	rainfall flux	seaIce	Mass of liquid precipitation falling onto sea ice divided by grid-cell area.
sisnmass	kg m-2	Snow Mass per Area	liquid water content of surface snow	seaIce	Total mass of snow on sea ice divided by sea-ice area.
sndmasssnf	kg m-2 s-1	Snow Mass Change Through Snow Fall	snowfall flux	seaIce	Mass of solid precipitation falling onto sea ice divided by sea-ice area.
wfo	kg m-2 s-1	Water Flux into Sea Water	water flux into sea water	ocean	Computed as the water flux into the ocean divided by the area of the ocean portion of the grid cell. This is the sum wfonocorr and wfcorr.