Multiple‐Wavelet Coherence of World's Large Rivers With Meteorological Factors and Ocean Signals

Streamflow is controlled by multiple factors concurrently. However, the multivariate relationship between global streamflow and meteorological factors/ocean signals is rarely explored at different temporal scales. Determining a suite of factors that explain most of the variations in global streamflow at multiple scales will be of great significance for water‐resource management and prediction. Temporally dependent multivariate relationships between streamflow and meteorological factors/ocean signals in 16 of the world's large rivers were identified using wavelet transform coherence and multiple‐wavelet coherence. Prior to that, the continuous wavelet transform was used to detect temporal patterns in streamflow. The continuous wavelet transform results showed that significant annual oscillations occurred in all streamflow series over the study period, either with continuous annual periodicity or with intermittent breaks. Oscillations with periodicities of around 4 to 6 months were also found in many rivers. A comparison of the results from the wavelet transform coherence and multiple‐wavelet coherence analyses indicated that streamflow variation could be best explained by one, two, or three meteorological factors. The combination of factors that best explained streamflow variations differed among the rivers, although total precipitation (PRE) or the number of rainy days (WET) either alone or in combination was a dominant factor for all rivers. The most common best predictor was PRE or/and WET combined with potential evapotranspiration. The differences in best predictor were due to differences in latitude, radiation forcing, terrain, vegetation coverage, hydrological processes, and so forth.