# How can we predict if it's raining, drizzling or there is snow fall by using precipitation intensity (inches/hour)?

I was analyzing a dataset in which precipitation intensity(inches/hr) and the type of precipitation was given. For example:

{precipIntensity:0.0029,pricpType:"Rain"}


I'm just curious how it can be predicted that it's raining in a particular area when the precipitation intensity is given ? Is there any formula or threshold by which we can predict the type of precipitation ?

• I'm not sure I understand your question. Certainly the precipitation intensity alone does not uniquely determine the type of precipitation, but perhaps at the extreme high end of the distribution it is more likely to be rainfall. Why don't you plot some histograms for each precip. type to find out for yourself? – gerrit Jan 19 '16 at 14:07
• To add to Gerrit's comment, using temperature with rainfall intensity can also give an indication of precipitation type, particularly for low intensity falls. Snow is more likely to fall than drizzle when it's very cold. You may even need to consider the rate of temperature change, as hail can fall during a storm after a very hot weather. – Fred Jan 19 '16 at 23:04
• Thanks gerrit and Fred . Point noted. I'll plot a histogram and check how this varies. The dataset also has temperature change. I'll do some more research on it to get at any conclusion. – Rahul Jan 20 '16 at 5:25
• @gerrit, at the VERY highest end, it might be frozen again :-p – JeopardyTempest Sep 2 '17 at 20:25
• @JeopardyTempest That looks pretty violent but I have no clue how it compares to a proper tropical rainfall downpour when it comes to quantity. – gerrit Sep 2 '17 at 21:37

TL;DR: When reporting precipitation types, we still have to manually look out the window.

TL;SR: You do not mention how the information is gathered, so I will go over the different ways.

The first and most obvious one is the air temperature. If the air temperature is above freezing then you have rain, below you have snow. Until recently reporting stations were all manned, and so someone would go out and look and determine what kind of precip is falling. Reports such as if it was raining or snowing, cloudy or sunny, are based on the observation at that moment, not over the course of time. With technology, a simple camera and light source can determine precip type. Snow reflects light back to the camera whereas rain lets the light pass through the drop.

Radar is a little more complicated. We now use what is known as Duel Polarization (Duel-Pol) radars. Using signals polarized in different directions, we can measure the shape of the precipitation.

Wind resistance flattens out rain drops, making them wider than tall. Hail tends to be round. Snow may present itself on it's face appearing as hail, or side, appearing as rain. The ratio of height to width gives the Spectral Differential Phase. Snow has a lower value and rain has a higher value with some overlap. Another important calculation is the Correlation Coefficient, which measures how similarly the two polarizations interact with the precip. In the image below is a map of different precip types and their CC. Note that snow has a much wider range of values than snow. Snow type is determined by temperature, wet snow closer to freezing, drier snow as you get colder.

These and other calculations give on the fly estimations for precipitation type.

However, what you see on the radar on TV still uses the old way of mapping station reported temperatures onto the radar screen. On this rain/snow radar loop You can make out the rain in green and yellow, the mix precip in pink, and snow in blue. Note however the mix line (where the air is at 32°F/0°C, moves five times over the 10 frames, and it is not smooth. This line is calculated by the measured temperatures from weather stations scattered over the area, drawing a line that marks the freezing point, much like drawing terrain contours. While the radar updates every 10 minutes, weather stations are only sampled hourly or quarterly.

So without knowing the exact way your data was gathered, I would say someone stuck their head out the window and marked on the computer that it was rain.