Wx
A weather prediction system that keeps you current...
Weatheronix is my concept design of distributed weather prediction system.
The history of this project rolls time back to 2012. The year I was gifted with my first tablet computer. After the power up, the first thing I saw was the weather-clock widget, which however displayed wrong weather status. It was chilling clear weather and status displayed was drizzle. This made me explore more on how weather is predicted. It is sophisticated instrumentation that capture weather information of large area. But having it deployed everywhere is currently not possible as it demands a lot of infrastructure. So this made me come up with a small device that can be put at home and is highly affordable. This is Weatheronix < Wx >
Today 03 - Oct - 2018, the predicted weather was sunny and clear because of that I neglected carrying my umbrella. While returning from office, I checked weather status which was sunny. But the real weather was cloudy and humid. I verified various weather information sources but all reported the same. On the way back it started raining heavily.
So this concludes that the tech still needs to be evolved.
Getting back to Wx. It basically consists of few weather monitoring sensors like anemometer, wind direction sensor, UV sensor, rain gauge, thermometer, hygrometer, seismograph, and a sensor that captures clouds boundary, this one needs machine learning to capture cloud data; and a compute engine capable to handle small ML algorithm to predict local weather and estimated clouds and air movement.
However the weather prediction of global weather happens in central high performance computers (HPC). When every home/office/school/etc are equipped with this device, the data generated will be crowd sourced and is in large quantities to predict the global weather. With the help of historical data, a machine learning model can be trained and based on the data generated the weather predication can become much more accurate and close to natural weather, also we can implement a self learning model or active learning model that keeps training the model based on the current and cross validating it. With the help of such architecture prediction of natural disasters can be made easy and counter measures can be taken accordingly.
You can imagine this approach as the calculus problem that was taught in higher secondary school. This diagram represents the approximation of area of the arc using the simpler geometrical structures rectangle, whose area can be computed easily. Now imagine the breadth of rectangles are large and there are only 2 rectangles. The accuracy of the area computed by this method will be very bad and if there are large number of rectangles with very small finite width then accuracy of area computed will be good.
Now if the density of such devices in any area is increased then the predicted weather will be close to natural weather and necessary actions can be taken. But as it is known more accuracy means more effort is needed, here effort will be the expense of the device and setting it up.
Do comment and share you views or email me.
Comments
Post a Comment