The earth's climate is fundamentally determined by the amount and transport of water in the atmosphere. Anthropogenic climate change has a strong impact on the water cycle, resulting in complex climate system feedbacks, which are a source of uncertainty in climate change projections. It is therefore important to better understand the hydrological cycle and, in particular, the atmospheric water budget. Reanalysis datasets are suitable tools to study the atmospheric water budget, because of their spatially complete and physically consistent coverage of the entire atmosphere. They are generated with numerical models run in a data assimilation framework, which are a source of uncertainty and, importantly, are not guaranteed to conserve physical properties such as mass and energy, which can therefore affect their water budget. In this work the quality of two widely used datasets - "ERA-Interim" (ERAI) and the more recent and higher resolution "ERA-5" (ERA5) - is assessed first globally and then regionally on the Tibetan plateau, over the period 1979-2015. The focus is held on the components of the atmospheric water budget and its “residual”, which should be zero in a perfectly consistent reanalysis dataset.
The global average shows that the residual of ERA5 is smaller than the residual of ERAI, indicating better performance of ERA5. However, at first glance, ERAI seems to perform better on the Tibetan Plateau: the water budget of ERA5 at the regional scale is noisy, which is attributed to its higher spatial resolution. However, the residual in ERA5 is more constant in time, which hints towards temporal inhomogeneities in ERAI. It was not possible to identify a specific process or parameter that could explain the temporal evolution of the residual, which appears to be random.
In a second step, the relative proportions of the individual budget terms to precipitation was analysed over a domain covering the Tibetan Plateau. On average, the ratio of local evaporation (moisture recycling) to moisture transport from outside the domain is 60:40, whereby this result is more stable in ERA5 than ERAI. Evaporation is the biggest contributor to the overall precipitation, however the correlation coefficients reveal that in terms of precipitation variability, precipitation is more likely modulated by moisture transport to the plateau. Since water on the plateau is limited, the additional water transported by the atmosphere in a specific year will probably increase the evaporation this year as well, which will increase its relative share of the total precipitation and thus ensure a positive precipitation anomaly. To test the robustness of the analyses, the plateau boundaries were shifted back and forth by one pixel and it was tested whether this has an impact on the results. Because of the complex topography the strong gradients of the water budget components in mountain areas, the closure of the water budget is affected by this arbitrary choice. Choosing a domain border where the gradients are weakest is the best in terms of minimizing the residual.
When comparing ERA5 with ERAI, improvements in their moisture budgets are visible, however further work is needed to understand the processes which influence the residual to achieve better closure.