Towards objective identification and tracking of convective outflow boundaries in next-generation geostationary satellite imagery

Abstract

Abstract. Sudden wind direction and speed shifts from outflow boundaries (OFBs) associated with deep convection significantly affect weather in the lower troposphere. Specific OFB impacts include rapid variation in wildfire spread rate and direction, the formation of convection, aviation hazards, and degradation of visibility and air quality due to mineral dust aerosol lofting. Despite their recognized importance to operational weather forecasters, OFB characterization (location, timing, intensity, etc.) in numerical models remains challenging. Thus, there remains a need for objective OFB identification algorithms to assist decision support services. With two operational next-generation geostationary satellites now providing coverage over North America, high-temporal- and high-spatial-resolution satellite imagery provides a unique resource for OFB identification. A system is conceptualized here designed around the new capabilities to objectively derive dense mesoscale motion flow fields in the Geostationary Operational Environmental Satellite 16 (GOES-16) imagery via optical flow. OFBs are identified here by isolating linear features in satellite imagery and backtracking them using optical flow to determine if they originated from a deep convection source. This “objective OFB identification” is tested with a case study of an OFB-triggered dust storm over southern Arizona. The results highlight the importance of motion discontinuity preservation, revealing that standard optical flow algorithms used with previous studies underestimate wind speeds when background pixels are included in the computation with cloud targets. The primary source of false alarms is the incorrect identification of line-like features in the initial satellite imagery. Future improvements to this process are described to ultimately provide a fully automated OFB identification algorithm.

3D wind field profiles from hyperspectral sounders: revisiting optic-flow from a meteorological perspective

“…Approximating the vertical integration of the last term of (A3) using the trapezoidal rule, and inverting the derivatives and integrals (assuming smoothness conditions) in the remaining terms of (A3), we obtain by the vertical integration of (A3) (one of the three components of) the partial differential equation We note that the trapezoidal rule used for vertical integration has been chosen for consistency between the model variables: the vertical integrated dynamics depend on vertical winds at the layers interfaces, as for the mass conservation constraint (2.6). 9 A good approximation for synoptic-scale motions is to let ω ; − wρg. Thus, using this approximation, vertical motion can be expressed as: where, in relation to pressure levels p k and p k+1 , we have define the isobaric surfaces s k (x, y) and s k+1 (x, y) in the altimetric coordinate system (x, y, z).…”

“…Energy minimization methods, known in the computer vision literature as optic flow algorithms, have shown promise as approaches in atmospheric science because of their good adaptation to the inherent physical nature of images, and because they can handle low contrast and missing observations, see [8] for a review on optic flow. Among other meteorological studies, these techniques have been applied to the identification of flow boundaries [9], or adapted to the estimation of layered AMV fields at different altitudes in the troposphere by satellite measurements of cloud top temperature [10]. Using the latter approach, the methodology has then been extended to the three-dimensional layered estimation of AMVs [11].…”

Fire Danger Observed from Space

“…But even though some fire danger models, such as FWI, NFDRS and FDRS, use wind speed as one of the models' input parameters, the coarse resolution of the information based on RS (1-30 km) limits considerably the usefulness of wind information for realistic fire danger assessment. Still, the improvement in the spatial and temporal resolution of the new generation of geostationary sensors is starting to provide wind vector information that could be useful to estimate shifts in wildfire direction and pyroconvection (Apke et al 2020).…”

3D wind field profiles from hyperspectral sounders: revisiting optic-flow from a meteorological perspective

“…Energy minimization methods, known in the computer vision literature as optic flow algorithms, have shown promise as approaches in atmospheric science because of their good adaptation to the inherent physical nature of images, and because they can handle low contrast and missing observations, see [15] for a review on optic flow. Among other meteorological studies, these techniques have been applied to the identification of flow boundaries [3], or adapted to the estimation of layered AMV fields at different altitudes in the troposphere by satellite measurements of cloud top temperature [23]. Using the latter approach, the methodology has then been extended to the three-dimensional layered estimation of AMVs [21].…”