Japan has a dynamic landscape and corresponding weather patterns. While it does have four seasons, its climate ranges from subarctic in the north to subtropical in the south. The landscape and weather makes it prone to many types of natural disasters, such as typhoons. An annual average of 10 typhoons come within 300 kilometers of the coast of Japan, with many of them directly hitting the Japanese mainland causing injuries, loss of life and destruction.

Since many severe storms like typhoons occur annually, the necessity for accurate weather prediction is critical to lessen impacts to people and property. A joint program between two Japanese agencies was formed to better understand and improve weather prediction models. The two agencies involved are the Institute of Statistical Mathematics (ISM), Japan's national research institute for statistical science, and the Japan Meteorological Research Institute (MRI), which conducts research for the Japan Meteorological Agency.

Professor Genta Ueno at ISM is leading this program and approached the NV5 Geospatial K.K. team to understand how their scientific programming language, IDL®, could help with their weather predictions. A key foundation of scientific discovery is complex numerical data, and IDL (Interactive Data Language) is trusted across disciplines to create meaningful visualizations out of this data. IDL lets users transform numbers into dynamic visual representations, so they can interpret their data, expedite discoveries and deliver powerful applications to market.

Visual Analysis of Numerical Weather Data

For this project, NV5 Geospatial K.K. conducted research and developed an IDL application for numerical weather prediction model analysis. The IDL application visualizes ISM and MRI barometric pressure and temperature data. To enhance the analysis further, the application also visualizes how the data corresponds with latitude, longitude, altitude and time of day. And it also makes the different parameters adjustable to help create multiple forecasts (Figure 1).

The IDL application for the numerical weather prediction allows users to display changes over time for physical quantities, such as atmospheric pressure and temperature, in three dimensions (Figure 2) – latitude, longitude and altitude. The IDL tool that was built for this visualization is unique for meteorological analysis, as it provides a GUI-based interactive capability for drawing physical quantities, which open-source offerings do not have.

Figure 1: The IDL interface for the two-dimensional drawing application displays physical quantities such as barometric pressure and temperature based on user input for latitude and longitude. The altitude and time of day can be displayed and changed interactively.

Figure 2: The 3D rendering in the IDL application displays physical quantities, such as air pressure and temperature, in terms of latitude, longitude and altitude.

Checking the Variability of the Results

The IDL application also supports the histogram display of ensemble data. Ensemble forecasting is a common method used in weather predictions. Instead of making a single forecast of the most likely weather, a set (or ensemble) of forecasts is produced. This forecasting group gives an indication of the range of possible future states of the atmosphere.

In research fields such as meteorology, objects being analyzed often behave chaotically and the numerical forecast results contain errors that increase with time. In order to identify and mitigate this error, NV5 Geospatial K.K. designed this application to use a method called Ensemble Forecasting.

In the Ensemble Forecasting method, multiple forecasts are made based on different conditions and statistical information, such as the average and the degree of variability, to provide a probabilistic view of the occurrence of specific weather phenomena. Another benefit of using this IDL application is that the user can also display a histogram of the ensemble data. The user can point and click and easily check the distribution of the results, based on multiple forecasts (Figure 3).

The IDL application for numerical weather prediction model analysis is expected to have a significant impact on the accuracy of Japanese weather models. Professor Genta Ueno has highly praised the research and work and says, “I am highly satisfied with the operability and functionality of the application.”

Figure 3: A histogram of the visualized ensemble data, showing it at the selected point in latitude and longitude. This allows the user to check the variability of the forecast data results.