There are many reasons why a forecast can go wrong. The old joke is that meteorologists get paid to produce wrong forecasts. In actuality, professional forecasts are pretty good. It is the few times they go wrong that tends to be remembered the most. Also, perception and acceptable error are factors. If the forecast is for 4 inches of snow and 6 inches falls, this can be interpreted as an okay forecast or a bad forecast depending on who you talk to. The snow did verify and several inches fell but a couple more inches fell that expected. Is this a good or bad forecast? It is a judgment call. Next, several reasons will be given for why a forecast can go wrong.

One reason is that forecasting error increases through time. It is forecasts beyond 3 days out that are more likely to be incorrect. If a forecaster is judged too much by long term forecasts they will be perceived as having more incorrect forecasts. Forecasters will tend to stress uncertainty with a forecast beyond 3 days by using percentages, statements indicating forecast uncertainty and larger ranges of expected values.

Another reason is small scale impacts. The forecast models are best at picking up on larger scale processes. Smaller scale processes such as soil moisture distribution, surface cover, abrupt elevation changes and vegetation type will influence the forecast. The forecast models are getting better with picking up on these smaller processes but for now they can contribute significantly to variations in the local forecast.

Another reason is that slight changes in the weather variables can result in dramatic changes to the forecast. Here are some examples:

1) a slightly stronger cap which inhibits convection
2) a slight change in wind direction that produces more upslope flow than expected and thus more precipitation and cooler temperatures
3) more snow cover on the ground than expected leading to cooler temperatures
4) more moisture than expected which leads to stronger storms and heavier precipitation

A final reason is the forecaster not taking into consideration all the data and resources or overforecasting certain aspects of the forecast. Wishcasting can also take place which is making a forecast based on what a forecaster wants to happen. Model accuracy is getting better. If a forecaster disagrees with this guidance it is important to have a good and compelling reason to disagree and change the forecast. Consistently making the forecast different from model guidance for the sake of making it different without a strong reason can lead to a poorer forecast performance.