Projection and possible causes of summer precipitation in eastern China using self-organizing map

Abstract

Self-organizing map (SOM) is used to simulate summer daily precipitation over the Yangtze–Huaihe river basin in Eastern China, including future projections. SOM shows good behaviors in terms of probability distribution of daily rainfall and spatial distribution of rainfall indices, as well as consistency of multi-model simulations. Under RCP4.5 Scenario, daily rainfall at most sites (63%) is projected to shift towards larger values. For the early 21st century (2016–2035), precipitation in the central basin increases, yet decreases occur over the middle reaches of the Yangtze River as well as a part of its southeast area. For the late 21st century (2081–2100), the mean precipitation and extreme indices experience an overall increase except for a few southeast stations. The total precipitation in the lower reaches of the Yangtze River and in its south area is projected to increase from 7% at 1.5 °C global warming to 11% at 2 °C, while the intensity enhancement is more significant in southern and western sites of the domain. A clustering allows to regroup all SOM nodes into four distinct regimes. Such regional synoptic regimes show remarkable stability for future climate. The overall intensification of precipitation in future climate is linked to the occurrence-frequency rise of a wet regime which brings longitudinally closer the South Asia High (eastward extended) and the Western Pacific Subtropical High (westward extended), as well as the reduction of a dry pattern which makes the two atmospheric centers of action move away from each other.

Appendix: Example of SOM nodes and associated rainfall cumulative probability distributions for Wuhan station

This appendix shows an example (for Wuhan) illustrating how SOM is used as a climate downscaling tool. The 20 SOM synoptic patterns are shown in Fig. 13 and the corresponding precipitation cumulative probability distribution functions (CDF) are shown in Fig. 14. It can be seen that the frequency of each pattern is between 2 and 9%, and the quantization errors are all controlled below 11 with a uniform distribution on different patterns, indicating that each pattern possesses expected and almost homogeneous capability to reflect information of its corresponding samples. The upper left patterns in Fig. 13 are dominated by lower humidity and higher SLP with northerly dry wind anomaly, while the lower right by higher humidity and lower SLP with southwest wet wind. Patterns in the middle are “transition patterns”. Moreover, circulation states of adjacent patterns are similar to each other, while far-away patterns are greatly different. This is an indication that the 20 patterns together have the expected capability in reflecting overall information and synoptic evolution (distributions and variations of circulation elements) at this station. Based on the above SOM synoptic patterns, the CDF of rainfall corresponding to each pattern is obtained (Fig. 14). Similar behaviors can be observed as for the SOM synoptic patterns: CDFs of adjacent SOM synoptic patterns are similar to each other, and show obvious differences for far-away patterns. A joint inspection for what shown in Figs. 13 and 14 reveals expected results, that is, a lack of rainfall corresponds to a combination of higher SLP, lower humidity and northerly dry wind and abundant rainfall takes place in the opposite situation of lower SLP, higher humidity and southwest wet wind. This provides evidence that the relationship between SOM synoptic patterns and observed rainfall is reasonable, and different SOM synoptic patterns can reflect different distribution characteristics of rainfall.

Fig. 13

The 5 × 4 nodes of the Self-organizing map (SOM) obtained for Wuhan station and during the training process (1961–2002) of the neural network. Anomalous fields, as a deviation from their mean climatology, are plotted or superimposed for each node. The green contours show sea-level pressure (SLP) anomaly (green solid lines: positive; green dashed lines: negative and zero) (unit: hPa); Arrows represent 850-hPa wind (unit: m s⁻¹, only plotted when amplitude larger than 0.5 m s⁻¹); 850-hPa relative humidity anomalies (unit:  %) are shown in shading (negative in red and positive in blue). The purple frame depicts the position of our interested area, the Yangtze–Huaihe river basin. The letter ‘h’ above each panel represents occurrence fraction (%) of the pattern, the letter ‘qe’ is quantization error calculated as the sum of the absolute differences of all states forming the cluster from the cluster mean

Fig. 14

Cumulative probability distribution function (CDF) of precipitation at Wuhan station, corresponding to the SOM nodes shown in Fig. 13 for the training period (1961–2002)