The National Weather Service (NWS) is currently looking into adopting a more efficient translation workflow, according to a recently published request for information (RFI). 

In the RFI, the agency outlined its plans to adopt machine learning technology that can be used to help disseminate information about major weather events, such as hurricane warnings and other forecasts, to a Spanish-speaking audience. Currently the NWS produces such translations manually, however the organization believes this to be an unsustainable practice due to the heavy workload and the amount of time it takes to complete.

“Within the NWS to date, there are numerous locally developed (translation) capabilities, many of which are labor intensive and not sustainable,” the RFI reads. “As there is no nationally supported system to enable the generation of Spanish-language products or services, a near term and long-term enterprise solution needs to be developed.”

The NWS noted that in order to fulfill its mission of protecting people and property in the event of natural disasters and other harsh weather events, a faster, more efficient process is necessary — that’s where machine learning technology comes in.

According to the RFI, the NWS Weather Forecast Office in San Juan is currently responsible for producing manual translations of the National Hurricane Center’s Atlantic basin products into Spanish. However, the NWS believes this to be an unsustainable approach to language access, especially given the fact that the San Juan office has a limited number of staff who are proficient in Spanish.

The NWS Office of Central Processing put out the RFI on Dec. 6, requesting information on machine learning and other types of technology that can be used to streamline the process of translating weather-related information, from forecasts to hurricane warnings, into Spanish. Of course, artificial intelligence wouldn’t completely replace human translation efforts — the RFI notes that it would produce “first guess” translations into Spanish, that would then be edited by human staff to make sure that the weather jargon is translated appropriately.

The RFI outlines several goals, including training machine learning and artificial intelligence models so that they can produce quick translations that are accurate to weather-related terminology. The NWS also plans on developing a GIS map to identify areas of the country with particularly large proportions of Spanish-speaking individuals with limited English proficiency.  

The current RFI only focuses on improving the translation workflow for Spanish, which is the second most widely spoken language in the United States. However, the NWS noted that improving the workflow for Spanish will lay down the groundwork for producing better, more efficient translations into other languages spoken throughout the country and its territories, specifically Samoan and French.