In EURAC-Ecohydro/SnowSeasonAnalysis: Analysis of snow and precipitation during winter season

Title: "Snow_detection_TS_PAR.R"

Introduction

The script Snow_detection_TS_PAR.R in folder inst analyzes Soil Temperature (TS) and Photosynthetically active radiation (PAR) to extract snow presence near the station. The algoritm improve results obtained only with one of these elements. The algorithm check if soil temperature (@ 0 cm) has high or low value and how it is the daily amplitude. At the same time it check the daily ratio between PAR @ 2 m and @ 0 cm (soil level) and the daily maximum at soil level. Combining these infomarmations we can determinate the presence/absence of snow without ultrasonic snow sensor.

Description of script

• Section 1: in this section you select input file to process. Input files are in folder data/Input data. After the selection the script import data as a zoo series

• Section 2: here you set inputs used to run models used. You should assign to each variable the corresponding column names of zoo_data:

  • soil_temperature (required)
  • phar_up (required)
  • phar_down (required)
  • snow_height(optional)

  You can also tune these moldel parameters:

  • daily_mean_soil_tempeature_threshold: Default is 3.5 deg C. Threshold on daily mean soil temperature. Under this value the model TS suggest snow presence
  • daily_amplitude_soil_tempeature_threshold: Default is 3 deg C. Threshold on daily amplitude of soil temperature. Under this value the model TS suggest snow presence
  • daily_max_ratio_parup_pardown: Default is 0.1. Threshold on daily ratio between maximum PAR at soil level (phar_down) and maximum PAR on weather station (pahr_up) Under this value the model PAR suggest snow presence
  • daily_max_pardown: Default is 75 w/m2. Threshold on daily maximum PAR at soil level (phar_down). Under this value the model PAR suggest snow presence

  Running row by row this section the algorithm plots the data selected for an help and a fast visual inpsection.

• Section 3: this section run different models and combining results. Two different models are combined in a third. Here we explain how single models is built and how they are combined.

  • TS: Snow presence Soil Temp. This model analyse only soil temperaute at 0 cm and indicate snow presence (flag = 1) every day that have at the same time a small daily mean and a small daily amplitude

  • PAR: Snow presence PAR. This model analyse only phar sensors (PHotosynthetically Active Radiation sensors) at soil level (phar_down) and on weather station (phar_up). It indicate snow presence (flag = 1) every day that have at the same time small radiation incoming that hit the soil passing through grass and a small ratio (that suggest that a lot of incoming radiation is blocking by snowpack)

  • TS+PAR: Snow presence PAR + Soil Temp. This model combine the previous 2 in a smart way. We observe that the model TS are not able to find properly the early snowfall, probably due thermal inerthia of soil. Instead PAR perform better, it is observed an important decreasing on radiation at soil level is sufficient a few centimeter of snow. During spring, the two models perform at the opposite, the PAR is affected by high radiation that penetrate on melting snow. Instead the snow maintains the soil temperature constant until the complete snow melting (melting snow and water has a temperature near 0 deg C). To develop this model we take into account these facts and elaborate an algorithm that consider snow cover the events that staring with model PAR and ending with ST. For a stable results we assume that PAR have to detect snow for at least 2 days consecutively

• Section 4: Save output as .RData for a visualizzation tool and as .csv to create a table

Datasets

INPUT:

1. git_folder: source of package SnowSeasonAnalysis
2. file: name of .csv file to process available in folder /data/Input data
3. soil_temperature: Required. The column name of file corresponding with soil temperature. In the Example: "ST_CS_00"
4. phar_up: Required. The column name of file corresponding with phar up. In the Example: "PAR_Up"
5. phar_down: Required. The column name of file corresponding with phar down. In the Example: "PAR_Soil_LS"
6. snow_height: Optional. The column name of file corresponding with snow height. In the Example: "Snow_Height"
7. daily_mean_soil_tempeature_threshold: Default is 3.5 deg C. Threshold of daily mean of soil temperature that suggest snow presence.
8. daily_amplitude_soil_tempeature_threshold: Default 3 deg C. Threshold of daily amplitude of soil temperature that suggest snow presence
9. daily_max_ratio_parup_pardown: Default 0.1 (10%). Threshold of ratio between daily maximum of PAR at soil level and at 2 meters that suggest snow presence.
10. daily_max_pardown: Default 75 W/m2). Threshold of daily maximum PAR at soil level that suggest snow presence.
11. SUMMER_MONTHS: Vector of summer months used to exclude snow in modes TS. Example "05": May, "06":June etc...

OUTPUT:

1. Snow_presence_file.RData: in folder data/Output/Snow_Detection_RData/ an .RData file which contains three zoo time series of:

   • Snow presence PAR + Soil Temp: Hourly snow presence time series, detect with model TS+PAR. Value: 0 means NO SNOW, 1 means SNOW
   • Snow presence PAR: Hourly snow presence time series, detect with model PAR. Value: 0 means NO SNOW, 1 means SNOW
   • Snow presence Soil Temp: Hourly snow presence time series, detect with model TS. Value: 0 means NO SNOW, 1 means SNOW

2. Snow_presence_file.csv: in folder data/Output/Snow_Detection/ a.csv file file which contains three zoo time series of:

   • TIMESTAMP: date and time of data
   • Snow presence PAR + Soil Temp: Hourly snow presence time series, detect with model TS+PAR. Value: 0 means NO SNOW, 1 means SNOW
   • Snow presence PAR: Hourly snow presence time series, detect with model PAR. Value: 0 means NO SNOW, 1 means SNOW
   • Snow presence Soil Temp: Hourly snow presence time series, detect with model TS. Value: 0 means NO SNOW, 1 means SNOW

How to use

Open script Snow_detection_TS_PAR.R and:

1. Set git folder, the path where the package is download or used.
2. Run Section 1 to explore data available in folder data/Input data
3. Select file, the station you want process (Section 1.i)
4. Run Section 2 row by row to explore and select the best variables for TS and PAR
5. Run Section 3 to produce output. This section combine results of different models used
6. Run Section 4 to save outputs:
   • .RData contains a list of models zoo series used by Visualize_Snow_detection_TS_PAR.R (a tool for a graphical analysis of snow detection)
   • .csv is a dataframe built using different models time series

References:

Teubner, I.E., L. Haimberger, and M. Hantel, 2015: Estimating Snow Cover Duration from Ground Temperature. J. Appl. Meteor. Climatol., 54, 959-965, https://doi.org/10.1175/JAMC-D-15-0006.1

EURAC-Ecohydro/SnowSeasonAnalysis documentation built on Dec. 6, 2020, 2:05 a.m.