Nowcasting–now and Then

Abstract

Most traditional nowcast products identify and exploit relationships between different subsets of satellite or other observations and variables or phenomena of user interest. As more such, potentially useful relationships are discovered, new products of opportunity such as cloud-drift winds, cloud-top pressure and cooling, or cloud optical and microphysical properties are continually introduced. Due to their quality and timeliness, many of these products became indispensable in operational nowcasting activities, and some found additional use even in Numerical Weather Prediction (NWP) Data Assimilation (DA). Due to their use of limited subsets of observations and the targeting of different user phenomena, nowcast product development efforts carried out by a large group of scientists and institutes, however are unavoidably stovepiped. And because of the use of different and limited sets of observations, the resulting products are also suboptimal and potentially inconsistent. In a possible future alternative arrangement, all satellite and in-situ observations are synthesized into a digital representation of nature that NOAA’s NextGen Weather Program refers to as a Single Authoritative Source (SAS). This 4D datacube would contain all prognostic coupled Earth system model variables that at any time describe the prevailing conditions of the atmosphere including microphysical, aerosol, and chemistry related variables, the land and ocean surfaces, and the cryosphere on a fine 3D grid. All additional variables and phenomena of user interest can then be systematically derived from this comprehensive 4D SAS ensuring a consistent and theoretically highest quality product stream. The well structured synthesis - derivation approach allows more productive and collaborative community engagement to enhance the tools for the observation synthesis phase and expand the toolset for the derivation of user products. While NWP DA is a theoretically sound approach to the synthesis of observations into 4D analysis fields, the fidelity of such analyses to nature and their latency prevent their widespread use in nowcasting today. Hence there is continued need for traditional nowcast tools. A gradual shift of development resources from the traditional to the new paradigm, however is necessary to realize the potential in a more systematic, consistent, and comprehensive new approach. Major challenges on this path include (a) access to observations sufficient to simulate nature; (b) forward models that appropriately link the observations to analysis and model variables; the development of DA techniques that can (c) simulate nature with adequate fidelity (e.g., variational vs sequential solvers with adjoint vs ensemble approaches; analysis of microphysical variables) and (d) run sufficiently fast on future computer platforms (e.g., efficient and accurate minimization on multiple scales) for timely nowcast products generation; (e) numerical models that can offer DA background fields with adequate realism; (f) a gradual realignment of the satellite and related communities from the past pattern of pipelined product generation to contributions to the shared focus areas of observation synthesis and product derivation; and (g) a reconfiguration of a wealth of existing satellite generation algorithms and software into the context of the new paradigm. Through some legacy and more recently developed tools, the presentation will highlight some of the challenges and benefits ahead of us.