Wolfram Computation Meets Knowledge

Wolfram Summer School


Mohammad Soltanieh-ha

Summer School

Class of 2014


Mohammad is finishing up his PhD in condensed matter physics at Northeastern University, Boston, Massachusetts. His research focuses on strongly interacting electronic systems in low dimensions; organic superconductors, nano tubes, and quantum wires are some examples of these types of materials. His work is theoretical and he takes advantage of computational tools in order to solve the problems that are not normally solvable by hand. He is also interested in network theory, social sciences, and data science. Other interests are camping, mountain biking, kayaking, skiing, and playing pool.

Project: Short-Term Precipitation Forecast

Base reflectivity radar images contain information about the structure of the clouds, how dense they are, and whether they carry water or hail. The unit of reflectivity is in dBZ (decibels), which is a logarithmic scale, and the values normally vary from 5 dBZ to 75 dBZ. Typically, light rain occurs when this reaches 20 dBZ. The purpose of this project is to track the clouds and see when they are reaching a given location on the map and whether it’s going to rain or not for the next hour or so. Also, a rough image for the location and size of the clouds in the near future can be predicted. The cloud images are taken from Weather Underground API every 6 minutes for the state of Massachusetts, and the information about the clouds is derived by image processing.

We are able to predict the precipitation by looking at the circled area shown in the image to see if there is rainfall happening. If we observe some precipitation, most likely it’s going to happen in the targeted location, which in our example is Boston. In order to check the precipitation, one can use the radar data and check whether the value for reflectivity is indicating rain or not, or make a query from either the Wolfram Language or the Weather Underground API current conditions. The average of these values might result in a better prediction since each data may have a different source. For this particular example, the circle is drawn one hour apart based on the speed of the clouds calculated by the program.

One can use this code to download radar images every 6 minutes from Weather Underground API: UWeather.nb. This UWeather.nb code needs to run for at least 60 minutes in order to have enough data for starting the actual prediction application, which is finalProject_realTime_Version2.0.nb

And this program will do the rest for real time: finalProject_realTime_Version2.0.nb

If one needs no visualization for the program, then this will take less than 30 seconds to run with a normal laptop. Here is the code without any visualization: finalProject_realTime_Version3.0.nb

If one needs to run a test for the code in a given time in the past, here is an example predicting Boston precipitation on July 15, 2014, for 5pm to 6pm and comparing it with the actual radar images taken for the same time frame: finalProject_presentation_Version2.0.nb.


How to use and interpret Doppler weather radar by Jeff Duda

Favorite Outer Totalistic r=1, k=2 2D Cellular Automaton

Rule 95406