Predictability of weather and climate

Abstract:

Soil moisture drought, as one of the most important drought categories, is determined by both water supply (e.g., precipitation) and demand (e.g., potential evapotranspiration). To shed light on the underlying mechanisms driving soil moisture drought, the statistical multiple linear regression, machine learning, and modeling experiments methods have been pervasively used in early studies. However, these methods neglect the collinearity and interactions of climate variables, and thus cannot reflect the direct and indirect interaction of factors leading to soil moisture drought. To reveal the synergistic effects of water supply and demand on soil moisture drought, this study quantified the contributions of key drivers to the change of soil moisture drought by a path analysis method to exhibit the relationships between atmospheric movement state and soil moisture drought. Prior to applying the systematic path analysis model, we identified the spatial patterns of soil moisture droughts at different depths by using a state-of-art three-dimensional drought recognition method in China. Our results showed that precipitation deficits dominated the interannual variation of soil moisture drought while increasing potential evapotranspiration only had marginal intensification in drought. The response of soil moisture drought to potential evapotranspiration was magnified by drought deterioration, especially in basically severe drought conditions. The total column water vapor and the horizontal divergence of the vapor flux, as well as temperature, directly affected precipitation and potential evapotranspiration and led to soil moisture drought through various direct and indirect processes. This study highlighted that the interactions among precipitation, potential evapotranspiration, and atmospheric vapor movement state in space and time were important for understanding the drought development mechanisms.

Abstract:

The Northern Hemisphere winter is the main rainy season for the Arabian Peninsula (AP), Central Southwest Asia (CSWA), and Southern Africa (SF), where precipitation predictability is limited or understudied. This dissertation research focuses on improving our understanding of these regions\u27 wet-season precipitation characteristics and predictability. First, I have identified the AP\u27s key moisture sources through a Lagrangian back-trajectory algorithm. Mid-latitude land and water bodies, such as the Mediterranean and Caspian Seas, are the primary moisture sources in the northern region. Areas further south rely on moisture transport from the Western Indian Ocean and the African continent. A significant drying trend in parts of the Peninsula is partly attributed to anomalies in moisture advection from the Congo Basin and South Atlantic Ocean. Next, I have identified key tropical and extratropical forcings that explain about three-quarters of winter precipitation variability in CSWA. Tropical forcing comes from an indirect ENSO forcing pathway, the dominant mode of precipitation variability in the Indian Ocean referred to as Indian Ocean Precipitation Dipole (IOPD). Extratropical forcing arises from a large-scale mode due to internal atmospheric variability. Seasonal forecasting systems effectively depict the characteristics of tropical forcing and its teleconnection with CSWA. Extratropical forcing spatial structure has also been skillfully represented. However, a lack of skill is noted in depicting its interannual seasonal variability and teleconnection with CSWA, which is the main driver of limited prediction skills in models. Lastly, I developed an empirical model using ENSO, Indian Ocean Dipole (IOD), and IOPD as precursors to investigate SF monsoon variability and predictability. I note that a reasonable skill in predicting SF monsoon precipitation can be achieved by preconditioning these modes as early as five months before the monsoon season. Seasonal forecasting systems that represent the interplay of these modes can achieve reasonable prediction skills over SF with a one to three months lead. However, ENSO forcing is overly strong in these models, making their predictions less skillful than the empirical model. These findings offer invaluable insight into the mechanisms of global teleconnections within the investigated regions, which should enhance the ability to predict wet season precipitation more accuratel