Traditionally, the forecast process, both in education and operations, has been referred to as a funnel. The funnel represents starting at the large scale and cascading to the local scale. Before the recent advances in numerical weather prediction (NWP), this process was essential as the forecaster had to first assess and predict all the large scale forcing and then apply these on the local scale. The funnel approach was necessary to move from the large scale to the end product, be it a city forecast or an aerodrome forecast. However, forecaster ability to supersede NWP is decreasing rapidly to the extent that skill atrophy is occurring. It is necessary to critically examine the forecast process and, correspondingly, education and training in applied meteorology. The current operational forecast system focuses primarily on the next few days. Thus, forecasters tend to start with the synoptic scale, often resulting in a lack of focus on detailed analysis and diagnosis and, instead, a focus on NWP fields. This can be tantamount to working backward, starting with NWP.

Given that NWP continues to improve to the point where forecasters cannot often intervene at the synoptic scale, modern NWP could be used as a best first guess. The forecaster could start focusing on the analysis and on smaller scale processes such as upright convection, cloud height, aviation icing, gap winds, coastal winds, gust fronts, orographic precipitation, ventilation, katabatic winds and mountain waves. Most of these topics can be taught without the need for quasi-geostrophic theory or even the coriolis term in the equations of motion. With modern tools including NWP, the operational forecaster is likely to add value in shorter and shorter timeframes. By inverting the forecast funnel and starting with the small scales, effort will increase on analysis and on the meso and micro scales. Such a cultural change will have to begin in the formative stages, in education and training of meteorologists.