801 Remote Sensing Products for Nowcasting at the National Meteorological Service of Argentina: Research to Operations Using Open Source Tools

Tuesday, 14 January 2020
Hall B (Boston Convention and Exhibition Center)
Martin Rugna, National Meteorological Service, Buenos Aires, Argentina; and M. Zeitune, P. Lohigorry, H. Ciminari, L. Vidal, J. J. Ruiz, and A. Arruti

Short-term forecasts issued in Argentina for high impact weather such as damaging winds, large hail, heavy rainfall among others are the main responsibility of the Nowcasting Office at the National Meteorological Service (SMN) of Argentina. In order to accomplish this task, the data from weather radars is used along with data from geostationary satellites and lightning activity sensors to detect and track storms that may develop high socioeconomic impact to the people and the property. The Research and Development Department of the SMN and the Nowcasting Office cooperate on a daily basis to develop and implement new products based on the available sensors and therefore improve the weather monitoring.

The weather radar information was the first in this set that was included in the research to operations scheme and that used Python as the main programming language driven by the large growth of the Argentine radar network. It went over the last 4 years from 4 to 14 radars but the tools formerly used to analyze the data couldn’t be used with these new radars plus it couldn’t display dual-polarization variables. We developed a completely new visualization jointly with the Nowcasting Office using PyART, Cartopy and Matplotlib integrating the information from several types and brands of radars within one data structure. This let us make basic PPI displays with one program saving debugging time but also composites with merged radars among other elaborated products. Also an additional layer is applied with Cartopy to include the severe weather warning and watches on the public SMN website. From a research perspective, the structure made with PyART is then used in a quality control set of algorithms that prepares the data for a regional data assimilation scheme and also a novel nowcasting tool.

The geostationary satellite data was added to the research to operations plan after the success of the radar products implementation on a development environment. It started with the launch of the next generation GOES-R/16 that brought a new and standard way to access and analyze the information from its imager, the ABI, and its new lightning sensor, the GLM. Given the previous generation (GVAR) was not capable of processing the data, we started to integrate the information from these new sensors into several visualizations, including RGB combinations, using netCDF4 and Cartopy. This is also included in the public SMN website. In a research phase is the analysis of different tools to maximize the information that the GLM can provide using the point data but also the areal extent with the glmtools package.

All of these packages are controlled with an internal git server and with documentation for replicating the environments and installing. This method was used for transferring from research to operations so that the programs are watched by the IT department all year.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner