5 Discussion

Ideally, this study would have comprised different evaluations of aviation targeted forecast products such as the TAF, but even though an extensive literature search was made, no such studies were found. As a result , quantifying the reliability of the forecast products could not be done directly. The included literature instead mostly concerns the models behind the products and their handling of convection. To complement this and to help bridge the gap to the essence of this study, input from experienced aviation forecasters was added. Nevertheless , my own conclusions were needed to assess the limitations of the forecast products and to answer the initial question. The answer is carefully formulated and rather sceptical, not putting too much faith in the conclusions and also reflecting the general idea of avoiding lulling pilots into a false sense of security.

The different studies of model evaluations referred to are all limited and quite small and thus too much trust should not be put in any specific study. Only the general results were therefore presented in the summation. The comments from Fyrby and Jakobsson should also be considered with caution as these do not originate from published scientific material, but rather from personal experience. This can also explain the slight difference in the statements.

Concerning the conclusions made about convective Plural Latin nounsindices ,[/annotax] a lot of uncertainties lie behind them. This is partly because of the fact that many of the indices used by SMHI were not evaluated at all because of a lack of studies. Regarding the graupel investigation by SMHI, considering the highly subjective grading method, the investigation could not be used to draw any convincing conclusions. Moreover , the studies made by Haklander & van Delden (2003) and Kunz (2007) over the Netherlands and Germany should be considered with caution since indices are known to perform quite differently over different areas (Groenmejjer et al. , n.d ) and the scores obtained might have proven to be significantly different over Sweden. Comparing these two studies, the difference in the scores could be a result of the different threshold values used, but that in turn is probably in itself a result of the large difference in terrain.

Furthermore , it should be noted that the products and models looked at in this study are not the only ones available for a pilot flying in Sweden. Therefore , complementary products that cover the weaknesses of the ones discussed here as well as models that are able to handle convection more accurately are very likely to exist. For example, www.vaderprognosen.se produces quite detailed forecasts well developed for the use of pilots. Moreover , it is based on the Weather Research and Forecasting model (WRF ) that is not discussed in this study.

Looking beyond the geographical boundaries of Sweden, there are lot of forecast products giving information exclusively about the risk for thunderstorms and severe convective weather. Examples are the products from the European Storm Forecast Experiment (ESTOFEX) (European Storm Forecast Experiment, n.d.) and NOAA’s Storm Prediction (National Weather Service, 2015). These show maps of different probabilistic fields that are easy to interpret. Although ESTOFEX covers the entire continent of Europe, a complementary such forecast might be useful over Sweden.

From the studies looked at it could be concluded that making more precise forecasts from the outputs of the models in use today would bring a lot of uncertainties. Considering this, the shortage of details in the products can be understandable. This way, the pilot might expect the bad weather for a longer time and for a larger area than necessary, but will not put faith in a thorough forecast that proved to be wrong. Hence , the safest forecasts might well be the ones that cover all possible outcomes, rather than the detailed ones.

However , one can still discuss the possibility of making forecast products with a little more information about various aspects of the possible outcomes. For example, in the case that “isolated embedded CB” is printed in a LLF, it should be possible to include an estimate of an approximate number of such cells in that area. It should also be possible to give some kind of information about how vigorous they might grow to be.

Concerning the Pluralmodels’ handling of convection,[/annotax] although there is no doubt more to improve in the physical model configurations such as parameterization of turbulence and moisture fluxes, the big flaw really seems to be the lack of coverage of observations, as mentioned by Fyrby (2015). Well-known from the theory of modelling, a sparse set of input data results in a lot of approximations in the initialization stage. This is bound to cause errors from the very beginning of the simulation, which will propagate and expand the longer the forecast stretches. This propagation is even quicker for chaotic processes such as convection. Thereforetherefore , it cannot be suggested that forecast products should include more details, but rather that the observational network should be expanded so as to improve model simulations.

As the techniques of modelling are constantly improving, a pilot will probably be able to trust more detailed forecasts in the future. For example, ECMWF plans to run a 2.5 km resolution global model by the year 2025 (Urquhart, n.d. ). This will surely surpass the limited-area high-resolution models in use today as a result of not having to make any boundary assumptions. However , as discussed above, it is also required to significantly improve the observational network. It is difficult to say how good future forecasts may become even though there are a lot of possible improvements. The fact that the mathematical equations describing convection are non-linear and chaotic and thus cannot be solved exactly remains.

Looking at last at the effects for the global community, putting a lot of resources into improving weather forecasts would not only help science move onwards, but also result in less / fewerfewer air traffic delays and reduce the number of weather related accidents.[/annotax] Travelling by air will become safer and more people might consider it as a way of transportation. This result might cause some opposition from an environmental point of view, but in favour of air traffic or not, unless something very drastic happens to our planet, it is probably safe to say that people will not stop flying in the foreseeable future. Today’s community is too adapted to a world where transport by air is a possibility for that to happen. Therefore , it should be okay to suggest that putting a lot of resources into improving weather prediction is indeed defensible. Not only does it help science move onwards, but it can also save lives.