Repository logo

Theses and Dissertations

Permanent URI for this collection

https://hdl.handle.net/10217/100335

Browse

Browsing Theses and Dissertations by Author "Arabi, Mazdak, committee member"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    The changing nature of convection over Earth's tropical oceans from a water budget perspective
    (Colorado State University. Libraries, 2023) Leitmann-Niimi, Nicolas, author; Kummerow, Christian, advisor; Maloney, Eric, committee member; Arabi, Mazdak, committee member
    Consistent spatiotemporal hydrologic measurements over Earth's oceans are only feasible with satellite remote sensing. The water budget components of an atmospheric column are precipitation (P), evaporation (E) and horizontal water vapor divergence (divQ). Physically, the sum of these components leaves a residual term: the amount of water vapor stored inside the atmospheric column. When time series of the water budget components are made using independent data products, this residual term is unphysical, which must be a result of measurement error in one or more of the water variables. This study finds that variations in lack of closure are not random, and seeks to reveal underlying sources for long-term, high amplitude trends so that errors in observations may be better understood as the climate system evolves and assumptions built into the algorithms today may bias results into the future. Trends in the residual are particularly significant over the Tropical West Pacific (TWP), Southern Tropical East Indian (STEI) and Tropical Central Pacific (TCP), where there are multi-year residual trends that maintain a consistent magnitude of 1 mm/day. While there are still residual discrepancies over the Tropical Western Indian (TWI), Tropical Eastern Pacific (TEP) and Tropical Atlantic Ocean (TAO), closure is overall better, as residual trends are more annual in variation and less unphysical in magnitude. This study hypothesizes that the first-order explanation for potential long-term biases lies in shifting convective organization. Convective organization changes are quantified using the amount of rain explained by three different regimes of convection (shallow, deep isolated and deep organized), which are dubbed convective rain states (CRS). A second-order explanation lies in relative ice amount. Relative ice amount is represented by ice-rain ratio (IRR), the amount of ice per amount of rain present in the atmospheric volume as determined from spaceborne radars. Changes in CRS can cause biases because rainfall spatial correlations related to well-known errors (e.g. beam-filling, convective/stratiform microphysics) are likely responsible for over-and underestimation of precipitation, while changes in the relative ice amount in individual convective rain states can cause the precipitation to be under or over-estimated due to scattering effects. Over the TCP these changes are purely dictated by the El-Nino Southern Oscillation (ENSO), with organization becoming a clear function of SST. Over the STEI there is a circulation that stems from the Indian Ocean Dipole (IOD), leading to a CRS and IRR dependence on vertical wind shear. Finally, over the TWP, CRS is neither a simple function of SST nor shear, but rather seems to arise from a deeper ocean change of state: a coupling/decoupling of west Pacific SST with central and east Pacific SSTs that coincides with the global warming hiatus period. Two different mechanisms may be at play when it comes to shifts in convective organization during this period. Outside of 2001-2007, the regular mechanism sees low-level shear directing convective organization. During 2001-2007, TWP SSTs are much warmer than surrounding SSTs, leading to an anomalous mechanism that sees water vapor convergence and atmospheric instability favoring isolated convection. The purpose of this study is not to locate specific algorithm retrieval or model deficiencies, but rather to connect long-term measurement errors with fundamental changes in dynamic characteristics of ocean environments. While these proposed mechanisms would explain much of the observed biases, this study cannot address quantitative biases, as these depend on algorithm details and vary from one algorithm to another – although the qualitative trends are consistent among them.