Assessment of continental water mass balance components 2003--2020 over EURO-CORDEX in the DETECT LAMBDA framework

Individual components of the continental water cycle still show considerable variation inbetween available data products. At the same time, the research community is trying to identify processes and their changes in increasingly more detail across several domains of the hydrosphere.

Within the Collaborative Research Cluster (CRC) 1502 DETECT, we have developed the LAMBDA framework in order to assess and analyse available water flux products. Here, we evaluate essential climate variables such as precipitation (P), evaporation (E), terrestrial water storage (TWS) change (dS/dt), and river discharge (Q) by means of kernel-integrated monthly water mass fluxes in the terrestrial water budget equation over multiple time scales. We focus on the water balance equation in its form dS/dt+Q=P-E, which states that net flux P-E between atmosphere and surface must be balanced by the combination of TWS change and river discharge.

(1) In our central assessment 2003-01/2020-12 over the EURO-CORDEX region, we assessed budget flux components for 35 major river catchments, including a wide selection of observational and reanalysis products. In the master run (P: GPCC, E: GLEAMv4.2a, Q: GRDC, S: COST-G) we find that a total of 26 (74%) catchments show drying P-E behaviour, and 22 (63%) in terms of dS/dt+Q. Out of the 23 basins with a maximum P-E net flux of +25 mm/month, 91% show drying trends in P-E, and 87% show negative trends in the combination of TWS and river discharge; a finding that supports the "dry-gets-drier — wet-gets-wetter" hypothesis to some extent.  

(2) While dP/dt, across Europe, is heterogeneously distributed (-1.5±3.0 mm/month/10a), E more consistently increased by 1.2±1.4 mm/month/10a, which leads to an averaged combined P-E trend of -2.7±3.4 mm/month/10a.  On average, TWS losses increased (-0.4±1.0 mm/month/10a), and discharge declined by 0.6±2.3 mm/month/10a, i.e. combined -0.9±2.6 mm/month/10a. Which means that — in contrast to absolute fluxes — dS/dt+Q change appears, on average, ~equally caused by a decline in discharge and TWS; the trends, however, exhibit large uncertainties.

In total, a flux budget misclosure of -2.0±5.8 mm/month remains. Trend-wise and on average, these residuals become more negative (-1.8±2.7 mm/month per decade). The stated ±1σ ranges are a measure of variability across the assessed domains. 

(3) In terms of inter-component variation, we find that even at targets that are comparably well covered with P observations (e.g. Rhine), monthly precipitation values from a selection of sources vary by 10 mm (>10% of mean) on average. Multi-annual averages range from comparably ‘dry’ 70 mm/month (CRU) to as much 89 mm/month (ERA5, IMERG). At the same time, we assessed averaged evaporation to range from as low as 44 mm/month (GLEAM v4.2a) up to 58 mm/month (GLEAM v4.2b), with a mean of 51.3±7.6 mm/month (15%). Across Europe, we find that especially the observation-heavier E products drive winter-time STDs up to 50% of the monthly cross-product mean (10% during summer).

While this illustrates clearly how researchers risk achieving spurious budget closure through implicit or explicit “cherry‑picking” of terms-components, it appears that long-term trends across different data products are comparably stable for interpretation.