Why is the weather forecast inaccurate?

Weather forecast is a forecast made by various forecasting methods based on the climate background and weather evolution characteristics of a specific region, based on synoptic theory and corresponding forecasting models. An important feature of weather forecast is accuracy. Weather forecast can be roughly divided into ultra-long-term forecast, long-term forecast, medium-term forecast, short-term forecast, short-term forecast and ultra-short-term forecast. The change of accuracy is inversely proportional to the length of the prediction time limit, that is, the longer the time limit, the lower the accuracy, but this trend change is not absolute. It is understood that the overall level of domestic weather forecast is slightly lower than the most advanced level in the world. The following reasons are not accurate separately. As a forecasting discipline, weather forecasting has the limitations of all disciplines dealing with nature (including forecasting or non-forecasting, such as earthquake, astronomy, hydrology, biology, geology, physics and so on). ), that is, there are uncertain factors. This is because all kinds of changes in nature are nonlinear processes, that is to say, while it develops according to a certain evolutionary law (not absolutely stable), there are also countless uncertain factors, which interfere with the orderly evolution to a great extent, successively and randomly. Take atmospheric science as an example. Like liquid, the atmosphere is also a fluid, and there are many phenomena similar to liquid, such as vortex, fluctuation, oscillation, turbulence and so on. For example, various weather phenomena in the atmosphere are like eddies and bubbles in a pot of water. Faced with such a pot of water, we need to study its boiling process, judge its boiling time and predict its vortex. Analyze the factors that affect its maintenance, merger, rupture or disappearance, and the various derivative effects brought about by its changes ... This seemingly ordinary change actually contains very complicated physical and chemical processes, which is the difficulty of atmospheric science as a natural prediction discipline. But we are not helpless. The main solutions and their respective limitations are as follows: On the one hand, the theories of these natural disciplines will express the changes in all aspects of nature in a mathematical way by establishing various complex mathematical equations (mostly nonlinear equations), that is, numerical models. The process of establishing numerical model (modeling) needs to consider those uncertain factors, but they will be selected and ignored, so that the model can achieve the balance of accuracy and efficiency as much as possible, because if all factors are considered, even the fastest supercomputer in the world is not competent for the calculation task of the model (in fact, one of the fastest computer applications in the world is meteorology). Weather forecast is also applied to various numerical model products (called' numerical forecast'), and will increase the weight of numerical forecast results in the final conclusion in the future development, but at the same time, we also see that numerical forecast can never be completely accurate, because the nonlinear process will become chaotic and unpredictable in theory to a certain extent. Friends who have studied algebra will understand the concept of infinitesimal. Some curves can be infinitely close to the coordinate axis, but they will never overlap or intersect with the coordinate axis. Digital products are like this curve. With the continuous improvement and perfection of the theory, it can be infinitely close to the' accurate' axis, but it will never overlap or intersect with it. There is an infinitesimal distance between them, that is, there will always be errors. We don't have to admit this. On the other hand, natural science will also make predictions through statistical analysis (that is,' statistical prediction'), and infer the possibility of a certain phenomenon in the future under similar environmental premise by counting the probability of a certain phenomenon in history. Different from the numerical model, which pays more attention to the analysis of causes and the application of mathematical theory, statistical prediction pays more attention to the correlation between various influencing factors and phenomena. Through the method of probability and statistics, find out some factors with the greatest correlation as the focus of prediction. In other words, it doesn't care much about why, but tries to find out who is the most useful. Of course, since the probability method is used, there will be credibility problems and errors. The statistical forecast of atmospheric science also has this problem. The third aspect is the application of people's subjective judgment, because theory is dead and people are alive. In many cases, if the conclusion produced by imperfect theory is too different from the reality, it needs to be corrected manually, but this method depends on the accumulation of workers' own experience, so it has the characteristics of unevenness. Generally speaking, the development trend of future weather forecast is mainly objective forecast (using objective tools such as numerical forecast and statistical forecast) and subjective forecast (personal experience and judgment of forecasters as a powerful supplement). However, it must be admitted that weather forecasting, as a natural forecasting discipline, always has errors. The weather forecast is not absolutely accurate. At the same time, we have to admit that due to the characteristics of thinking, people always firmly remember the process that leads to their unpleasant experiences, while ignoring the process that brings joy or gain. This is the selectivity of memory. When it comes to weather forecast, people will always remember its failure case very clearly and ignore its success case. When the weather station predicted rain, everyone took rain gear, and finally it rained. When the weather station predicted rain, it was sunny the next day, and everyone was really sunny the next day. Please ask yourselves, did you say' The weather station was right again'? I think, many people's answers are no, and everyone will take it for granted that it is' normal' and then forget it soon. However, when the weather station predicted that there would be no rain and it rained cats and dogs all over the street, absolutely everyone would scold:' The weather station should be laid off' and' the white-collar taxpayer of the weather station lost money', and everyone would remember this experience for a long time and enrich it into arguments to prove that the weather forecast was never accurate, so this view would be infinitely magnified. But have you ever thought that as a forecasting discipline, accuracy is' normal' and mistakes are not' normal'? We have no intention to excuse ourselves with this, but in a responsible attitude, while we must admit the existence of mistakes, it is also necessary to remind everyone to go to the next conclusion objectively. The weather forecast is not just finished, and then it needs to be graded strictly according to the actual situation. The grading rules are detailed and strict, and there are clear requirements for reaching the standard. Evaluate the sunny rain, rainstorm (including 24 hours and 48 hours), low temperature, frost and tropical cyclone (including path, landing time and landing point) respectively, and report the statistical results of accuracy and various weather phenomena to the National Meteorological Administration every month and year.