Global surface energy and water cycle variability 2001-2011 from satellite data

We derive internally consistent, monthly to interannual, energy and water budgets, with uncertainties, for all the main continents and ocean basins over 2001-2011 based principally on satellite data. An inverse model is used following the Thomas et al (2019) climatology study and the NASA energy and water cycle study (NEWS), L’Ecuyer et al. (2015), Rodell et al. (2015).
Input data include CERES and Cloud_CCI AATSR (radiation), FluxCOM (land turbulent heat fluxes), JOFURO3 (ocean turbulent heat fluxes), GPCP2.3 (Precipitation), GRACE (total water storage), ERA5 (atmospheric water storage), GRUNv1 (land runoff), and we compare these with alternative products to assess component uncertainties. The different components are then brought together and adjusted within respective uncertainties to achieve balanced energy and water budgets.
Preliminary results focus on seasonal and interannual variability over land. Seasonal modifications to the water budget over Eurasia and N America include a delay in spring runoff (and reduced evapotranspiration over Eurasia) as GRACE data indicates retention of water mass over land. Evapotranspiration adjustments to FluxCOM are strongly seasonal and also result in bringing the land seasonal energy budget closer to the DEEPC Liu et al (2015) results demonstrating the value of coupling the energy and water cycles.
Strong correlated interannual variability in African precipitation, runoff and GRACE derived water storage is found, and we assess the relative consistency of different data products, particularly for precipitation, where multiple datasets are available and uncertainties are large. Consistent African precipitation variability is found in the TAMSAT data, which further supports the water cycle change scheme around year 2006 over Africa. Clear ENSO signals are seen, particularly over South America in 2010 and Australia in 2010-11, with correlated variability in rainfall, runoff and water storage distributions. 
Optimisation is sensitive to the uncertainty of each energy and water budget component expressed in their spatial and temporal error covariances.  We introduce spatial error covariance for turbulent heat fluxes between major ocean basins as well as temporal error covariances for all components expressing the expectation of time mean bias adjustments. The results show improved net surface energy flux pattern with larger heat loss over North Atlantic and Arctic Ocean and more heat uptake for other basins and an intensified water cycle, with increased precipitation, evapotranspiration and runoff and stronger ocean-land water transports.