Estimating the regional climate responses over river basins to changes in tropical sea surface temperature patterns

Abstract

We investigate how to identify and assess teleconnection signals between anomalous patterns of sea surface temperature (SST) changes and climate variables related to hydrologic impacts over different river basins. The regional climate sensitivity to tropical SST anomaly patterns is examined through a linear relationship given by the global teleconnection operator (GTO, also generally called a sensitivity matrix or an empirical Green’s function). We assume that the GTO defines a multilinear relation between SST forcing and regional climate response of a target area. The sensitivities are computed based on data from a large ensemble of simulations using the NCAR Community Atmospheric Model version 3.1 (CAM 3.1). The linear approximation is evaluated by comparing the linearly reconstructed response with both the results from the full non-linear atmospheric model and observational data. The results show that the linear approximation can capture regional climate variability that the CAM 3.1 AMIP-style simulations produce at seasonal scales for multiple river basins. The linear method can be used potentially for estimating drought conditions, river flow forecasting, and agricultural water management problems.

Appendix: Sensitivity maps

Here we include GTO sensitivity maps (Fig. 8) for other river basins that are not shown in the main text. We briefly identify distinguishing characteristics of the sensitivity patterns that differ from these shown in the main text. These include the Parana, Niger, Lake Chad, Congo, Limpopo, Colorado, Indus, Ganges and Brahmaputra River basins. For Limpopo River basin in southern Africa, the sensitivity pattern shows different features compared to other river basins over Africa. The positive sensitivities for precipitation reach their maximum in DJF and SON over tropical West Pacific Ocean as well as the tropical Atlantic Ocean. Evident negative sensitivities are situated over the Indian Ocean throughout the year except for JJA which shows much weaker negative signals. As for Indus, Ganges and Brahmaputra River basins, they also have similar sensitivity patterns for precipitation but with different magnitude across the seasons. The Indus River basin has strongest dipoles over Indian Ocean in JJA and SON; however, Ganges and Brahmaputra River basins have it in MAM and the dipole disappears in DJF and it is replaced by positive sensitivity information. As a final comment on the sensitivity maps, we note two important features. First, the information is scaled and independent of strength of the SST forcing. This implies that the actual magnitudes of SST variability are required to assess the net changes of the regional climate. Second, the information depends on the background climatological SST applied from different seasons. This allows us to observe seasonal changes of the responses for different locations in the tropical ocean. This also helps us understand the importance of seasonal time scale of regional climate sensitivity over river basins.

Fig. 8

Regional sensitivity maps of temperature at 850 hPa (left) and precipitation (right) for selected river basins. The maps are shown for a Parana River basin, b Niger River basin, c Lake Chad River basin, d Congo River basin, e Limpopo River basin, f Colorado River basin, g Indus River basin and h Ganges and Brahmaputra River basins. The X denotes the location of river basin. Shaded regions indicate where the sensitivity signal is significant by using a two-tailed $t$ test at $10\,\%$ significance level (Units: K per K $\hbox {km}^{2}\times 10^{9}$ for T850 and mm/day per $\hbox {K}\, \hbox {km}^{2}\times 10^{9}$ for precipitation)