Connections between atmospheric perturbations, e.g., thunderstorm activity, and major earthquakes are investigated along with the lithosphere–atmosphere coupling mechanism, concerning the earthquake prediction models. The present research attempts to recognize a possible link between atmospheric processes (rainfall, storms) and subsequent earthquakes (M > 6) across a wide area around Japan. Earthquake data and upper-atmosphere sounding data related to the Severe Weather Threat (SWEAT) index and Skew-T plots were obtained from two Japanese radiosonde stations, Hachijojima and Kagoshima. Using the cross-correlation function (CCF) method, it is shown that SWEAT conditions existed within 30 days before six major earthquakes in 2017 in the Japan region. The Seismo-Climatic Index (SCI) reached a mean of 2.00, 7–8, and 13–14 days before these earthquakes, indicating thunderstorms and extreme weather conditions, further supported by Skew-T plots. Low-pressure systems, deviating from the mean by as much as −50 to −250 m, and hot spots of increased precipitation ranging from ~80 to ~140 mm rainfall within 24 hrs were observed to be geographically associated with these earthquake events. The anomalous atmospheric conditions can be understood based on increased air ionization at the ground-to-air interface due to the influx of positive-hole charge carriers that are stress-activated deep in the lithosphere and spread through the rock column. When the positive electronic charge carriers are accumulated at the lithosphere, preferentially at topographic highs, some steep electric fields are observed capable of field-ionizing the air. The airborne ions then act as condensation nuclei for atmospheric moisture, thermal updrafts, cloud formation, and a statistically significant precipitation increase. This research was conducted based on some experimental indicators in a very important seismological region to examine the successfulness of the proposed mechanism and the given indicators as the possible proxies of pre-earthquake precursors. Hence, the main practical implication of the research can highlight a sustainable way for improving the managerial tools in the field of earthquake prediction.