As opposed to drought, floods are readily identifiable and relatively short-lived events, caused by “an excess of water mostly but not exclusively from rivers” and described as “the temporary covering by water of land not normally covered by water” (European Union, Directive 2007/60/EC).
The relatively short-lived nature of flood events calls for a higher than monthly to weekly temporal resolution of the GRACE-based gravity field solutions in order to adequately capture the flood water dynamics at hand.
A daily time resolution, however, reduces the number of GRACE observations per time step to the extent that gravity field solutions based on GRACE observations alone are dominated by noise. The daily gravity solutions provided by TU Graz (ITSG-Grace2014) resolve this issue by employing temporal correlation patterns of the gravity field, generated by geophysical models (http://portal.tugraz.at/portal/page/portal/TU_Graz/Einrichtungen/Institute/Homepages/i5210/research/ITSG-Grace2014). An atmospheric, an ocean circulation and a hydrological model provide stochastic prior information to arrive at a prediction of the gravity field, which is subsequently updated with daily GRACE observations (if present) in a Kalman smoother approach. Additional processing converts the resulting gravity field solutions, expressed in terms of spherical harmonics coefficients, into global 1x1 degree gridded maps (approx. 111 km at the equator) of water storage anomaly in equivalent water height (cm).
A 1 degree grid, however, does not equate to a 1 degree spatial resolution. The daily solutions are solved up to the spherical harmonic degree 40, which translates into an approximate 450 km spatial resolution. In general, as a rule of thumb, 150.000 km2 is considered as a threshold for the size of the river basin where the hydrological signal from GRACE exceeds the noise.
Let’s now turn to the Danube river basin as a case in point. The second largest river basin in Europe (801.463 km2 area) has suffered from a number of major flood events in the years overlapping the GRACE mission’s life time (e.g., 2006, 2010, 2013, 2014). The flood in the spring of 2006 was a basin-wide event with the central and lower parts of the river basin most severely affected (https://www.icpdr.org/main/issues/floods). Figure 1 (left panel) shows the anomalies of total water storage from GRACE and the observed river discharge at the Danube basin outlet. The observations are of a different order of magnitude, but highly correlated. Correlation is lower in the Upper Danube sub-basin (right panel, 76.653 km2 area), but several individual discharge peaks can still be identified to correspond with signals in the daily GRACE record.
This seems to indicate a river basin of approx. 75.000 km2 in size, i.e. twice as small as the generally assumed lower threshold, may still yield a detectable flood signal in the GRACE solutions. The size of the Upper Danube sub-basin relative to the entire basin is illustrated in Figure 2, which also shows the maximum basin-average total water storage anomaly for the 2006 flood event on April 17.
Discharge station data is kindly provided by the Global Runoff Data Centre, 56068 Koblenz, Germany.
Figure 1 Daily variations of the gravity field and river flow in the Danube (left panel) and Upper Danube (sub-) river basin (right panel).
Figure 2 Maximum basin-average total water storage anomaly (cm) (ITSG-Grace2014) for the 2006 flood event in the Danube basin (outline) on April 17. The Upper Danube sub-basin is indicated in the western part of the Danube basin.
