Fig. 3: Physical interpretation of the DL model. | npj Climate and Atmospheric Science

Fig. 3: Physical interpretation of the DL model.

From: Exploring dominant processes for multi-month predictability of western Pacific precipitation using deep learning

Fig. 3

Observed 20–100 day filtered sea surface temperature (SST) (a, K), zonal wind at 850 hPa (b, m s–1), convective instability (c, K), and vertical velocity (d, cm s–1) anomalies regressed on the MJO-related precipitation index over western Pacific (120°–150°N, 10°S–10 N°) with a 65-days lag. The 20 yr (1997–2016) data was used. The dotted area represents significance at the 95% confidence level. The heat maps of each DL model. The DL models use SST (e), zonal wind at 850 hPa (f), convective instability (g), and vertical velocity (h), respectively. The validation period data (2012–2016) are used for heat map analysis. Only the values with over 95% confidence level are shown.

Back to article page