Thermal Infrared Remote Sensing of Snow¶

Learning Objectives

At the conclusion of this tutorial, you will be able to:

understand the differences between sources of thermal IR observations, the advantages and disadvantages of each, and types of research questions we can address with those observations
visualize point and raster thermal infrared datasets together, then compute error statistics between point and raster datasets
access and visualize airborne and satellite thermal infrared imagery, then scale imagery of different spatial resolutions for comparison

Download the sample datasets for this tutorial

For ease of access during the hackweek, sample files are available for download for running the command in the cell below. (See the bonus notebook “thermal-ir-data-download.ipynb” for more details about data access methods)

%%bash 

# Retrieve a copy of data files used in this tutorial from Zenodo.org:
# Re-running this cell will not re-download things if they already exist

mkdir -p /tmp/tutorial-data
cd /tmp/tutorial-data
wget -q -nc -O data.zip https://zenodo.org/record/5504396/files/thermal-ir.zip
unzip -q -n data.zip
rm data.zip

Import the packages we’ll need for this tutorial

# Import some general-purpose packages for handling different data structures
import numpy as np # for working with n-D arrays
import pandas as pd # for reading our csv data file and working with tabular data

# Import matplotlib which we'll use for plotting images and graphs
import matplotlib.pyplot as plt

# Import these packages for working with raster data
import xarray as xr # xarray lets us work with n-D arrays and labeled data, such as NetCDF files
import rioxarray # rioxarray provides capabilities of the rasterio package to xarray, letting us easily work with files such as GeoTIFFs

# Import some packages for working with the SnowEx SQL database
from snowexsql.db import get_db # Import the connection function from the snowexsql library
from  snowexsql.data import SiteData # Import the table classes from our data module which is where our ORM classes are defined 
from datetime import date # Import some tools to build dates 
from snowexsql.conversions import query_to_geopandas # Import a useful function for plotting and saving queries! See https://snowexsql.readthedocs.io/en/latest/snowexsql.html#module-snowexsql.conversions

Part 1: Comparing airborne IR imagery with ground-truth observations¶

Airborne IR imagery¶

Fig. 3 The Naval Postgraduate School Twin Otter aircraft carried the UW APL thermal infrared imager and SWESARR instrument over Grand Mesa for SnowEx 2020. (Photo by Chris Chickadel)¶

Load IR image mosaic geotiff file

TUTORIAL_DATA = '/tmp/tutorial-data/thermal-ir/'
airborne_ir = xr.open_rasterio(f'{TUTORIAL_DATA}/SNOWEX2020_IR_PLANE_2020Feb08_mosaicked_2020-02-08T181915.tif')

Inspect the contents of the file we just opened

airborne_ir

<xarray.DataArray (band: 1, y: 3848, x: 4398)>
[16923504 values with dtype=float32]
Coordinates:
  * band     (band) int64 1
  * y        (y) float64 4.332e+06 4.332e+06 4.332e+06 ... 4.313e+06 4.313e+06
  * x        (x) float64 7.371e+05 7.371e+05 7.371e+05 ... 7.591e+05 7.591e+05
Attributes:
    transform:      (4.9992529765780835, 0.0, 737077.3666563785, 0.0, -4.9992...
    crs:            +init=epsg:32612
    res:            (4.9992529765780835, 4.99925297657805)
    is_tiled:       0
    nodatavals:     (nan,)
    scales:         (1.0,)
    offsets:        (0.0,)
    descriptions:   ('STCtemp',)
    AREA_OR_POINT:  Area
    description:    Airborne thermal infrared imagery from University of Wash...
    grid_mapping:   spatial_ref

airborne_ir.rio.crs # original CRS

CRS.from_epsg(32612)

Under the dataarray’s attributes we can see that we have a coordinate reference system already defined (crs) as EPSG:32612. We can also find this through da.rio.crs.

However, we would like to reproject this into the common projection used by datasets on the SnowEx SQL database: EPSG:26912. We can do that using rioxarray’s reproject method (see an example here).

airborne_ir = airborne_ir.rio.reproject('EPSG:26912') # overwrite itself with new reprojected data array

airborne_ir.rio.crs # new CRS

CRS.from_epsg(26912)

Next, the filename shows us when this imagery was taken in UTC time, “2020-02-08T181915”

We can create a pandas timestamp variable in local time for comparison with other datasets:

# Create a pandas timestamp, subtract 7 hours from UTC time to get local time (MST, UTC-7)
airborne_ir_timestamp = pd.Timestamp(2020,2,8,18,19,15) - pd.Timedelta(hours=7)

What color scale should we use for temperature?

Common advice you may have heard is to avoid using rainbow color scales. Luckily matplotlib gives us lots of options to choose from.

When representing images of temperature, sometimes we want to pick colors that intuitively suggest temperature, such as the “magma” colorbar below. Other times we might be interested in both magnitude and sign, such as temperatures above or below melting point, in which case we could use something like “RdBu_r” below. I often pick simple greyscale, though less visually interesting, it is sometimes easier to pick out details in a continuous color scale like they “Greys” scale below. Make sure to include a labeled colorbar so your plot can be correctly interpreted!

Fig. 4 Some matplotlib color scale options (top to bottom): “magma”, “RdBu_r”, “Greys”¶

Plot the airborne TIR image.

What does our chosen colorscale tell us about temperatures here?
What can we see?

fig, ax = plt.subplots(figsize=(20,10)) # create a new matplotlib figure, set the figure size
ax.set_aspect('equal') # set the aspect ratio to "equal"

# plot the airborne infrared image
airborne_ir.plot(cmap='magma', vmin=-10, vmax=10,ax=ax, 
                 cbar_kwargs={'label': 'Temperature $\degree C$'})

# set axes labels
ax.set_xlabel('Eastings UTM 12N (m)')
ax.set_ylabel('Northings UTM 12N (m)')

ax.set_title('Airborne TIR imagery of Grand Mesa, CO (Feb. 8, 2020)');

../../_images/thermal-ir-tutorial_18_0.png

Image interpretation

What does our chosen colorscale tell us about temperatures here?
- The colorbar on the right shows us that colder temperatures are represented with dark purple to black colors.
- Warmer temperatures are represented with lighter orange/yellow.
What can we see?
- We see a narrow stripe of imagery, this is a mosaic of individual camera images taken from the aircraft along a single flight line over Grand Mesa.
- On the left (west) side of the image we see warmer but varied temperatures (yellow, orange, and purple), then an abrupt transition to uniformly cold temperatures (purple). This is the westernmost edge of the mesa where the snow on top is much colder than the lower elevation slopes of the mountain.
- Continuing to move from left (west) to right (east) we see patches of lighter colors, corresponding with warmer temperatures. These are patches of forest on the central and eastern portions of the mesa.
- Finally near the easternmost side of the image we see a very dark region that means it is much colder than the rest of the scene. What might this cold area or object be?

Ground-based temperature observations¶

To provide a source of “ground truth” for the airborne and satellite thermal infrared images during the SnowEx 2020 Grand Mesa campaign, we can use ground-based snow surface temperature measurements. On February 5, 2020, we installed a thermal infrared radiometer pointing at the snow surface at snow pit #2S10 (left), and buried temperature sensors beneath the snow surface (right). These logged observations at 5-minute intervals until we removed the instrumentation a week later on February 12.

Fig. 5 Snow temperature sensor setup at snow pit 2S10: (left) tripod-mounted thermal ifrared radiometer to measure snow surface, (right) temperature probes to be buried beneath the snow surface.
(Photos by Steven Pestana)¶

What are some of the differences we might expect to see between the ground-based surface temperature data and the thermal IR images?

Emissivity differences?
- Our ground-based radiometer was looking at 45 deg off nadir, versus nadir ASTER versus variable view angle airborne (snow emissivity changes off-nadir)
Different TIR bandwidths (broad versus narrow)?
- Ground-based radiometer: 8-14 μm, Airborne IR cameras: 8-14 μm, ASTER band 14: 10.95-11.65 µm
Different atmospheric path lengths?
- From <1 meter, to 1 km, to entire atmospheric column (~100 km)
“Point” versus area, and geolocation accuracy
- The ground-based radiometer is measuring temperature for a spot (not really a single point) maybe ~1m in diameter. The airborne and ASTER imagers have spatial resolutions of 5m and 90m respectively. How confident are we in the geolocation of individual pixels in the imagery?

Where is snow pit 2S10?

We can find this information through a query to the SnowEx SQL database. First, set up the connection:

# This is what you will use for all of hackweek to access the db
db_name = 'snow:hackweek@db.snowexdata.org/snowex'
# Using the function get_db, we receive 2 ways to interact with the database
engine, session = get_db(db_name)

Then, query SiteData using filter_by to find the entry with the site ID that we want (2S10). Preview the resulting geodataframe.

# Form the query to receive site_id='2S10' from the sites table
qry = session.query(SiteData).filter_by(site_id='2S10')

# Convert the record received into a geopandas dataframe
siteData_df = query_to_geopandas(qry, engine)

# Preview the resulting geopandas dataframe
siteData_df

	site_name	date	time_created	time_updated	id	site_id	doi	date_accessed	latitude	longitude	...	weather_description	precip	sky_cover	wind	ground_condition	ground_roughness	ground_vegetation	vegetation_height	tree_canopy	site_notes
0	Grand Mesa	2020-02-05	2021-06-01 18:41:44.586420+00:00	None	115	2S10	https://doi.org/10.5067/DUD2VZEVBJ7S	None	39.019474	-108.192314	...	Sunny and cold, some blowing snow	None	Few (< 1/4 of sky)	Moderate	Frozen	Rough	Grass, Shrub	20, 30	No Trees	None

1 rows × 30 columns

Inspect of the geodatframe’s metadata

# What is the coordinate reference system used here?
siteData_df.crs

<Projected CRS: EPSG:26912>
Name: NAD83 / UTM zone 12N
Axis Info [cartesian]:
- E[east]: Easting (metre)
- N[north]: Northing (metre)
Area of Use:
- name: North America - between 114°W and 108°W - onshore and offshore. Canada - Alberta; Northwest Territories; Nunavut; Saskatchewan.  United States (USA) - Arizona; Colorado; Idaho; Montana; New Mexico; Utah; Wyoming.
- bounds: (-114.0, 31.33, -108.0, 84.0)
Coordinate Operation:
- name: UTM zone 12N
- method: Transverse Mercator
Datum: North American Datum 1983
- Ellipsoid: GRS 1980
- Prime Meridian: Greenwich

# Preview the geometry of this geodataframe, we should see that it is a POINT
siteData_df.geometry

0    POINT (743076.000 4322689.000)
Name: geom, dtype: geometry

We can now plot our snow pit site from this geodataframe on top of the airborne IR image. (See more tips about plotting geodataframes here)

fig, ax = plt.subplots(figsize=(20,5)) # create a new matplotlib figure, set the figure size
ax.set_aspect('equal') # set the aspect ratio to "equal"

# plot the airborne infrared image
airborne_ir.plot(cmap='magma', vmin=-10, vmax=10, ax=ax, 
                 cbar_kwargs={'label': 'Temperature $\degree C$'})

# plot the location of the snow pit of interest
siteData_df.plot(ax=ax, color='r', marker='x')

# set axes labels
ax.set_xlabel('Eastings UTM 12N (m)')
ax.set_ylabel('Northings UTM 12N (m)')

# set the axes limits, units in meters UTM Zone 12N (I chose these values by just looking at the plot above)
ax.set_xlim((735000, 760000)) # x axis limits
ax.set_ylim((4320000, 4325000)) # y axis limits

# set plot title
ax.set_title('Location of Snow Pit 2S10\nwith Airborne TIR imagery of Grand Mesa, CO (Feb. 8, 2020)');

../../_images/thermal-ir-tutorial_36_0.png

Change the x and y axes limits (set_xlim, and set_ylim) to zoom in to our point of interest. In this case we can use df.geometry.total_bounds to get the x and y values that define the area our geometry takes up. (In this case we have a point so it will return just the point’s location, but this would work if we had a polygon as well)

fig, ax = plt.subplots(figsize=(10,10)) # create a new matplotlib figure, set the figure size
ax.set_aspect('equal') # set the aspect ratio to "equal"

# plot the airborne infrared image
airborne_ir.plot(cmap='magma', vmin=-10, vmax=10, ax=ax, 
                 cbar_kwargs={'label': 'Temperature $\degree C$'})

# plot the location of the snow pit of interest
siteData_df.plot(ax=ax, color='r', marker='x')

# set axes limits
xmin, ymin, xmax, ymax = siteData_df.geometry.total_bounds # get the "total bounds" for our geometry
ax.set_xlim((xmin-1000, xmax+1000)) # x axis limits to +/- 1 km from our point's "total bounds"
ax.set_ylim((ymin-1000, ymax+1000)) # y axis limits to +/- 1 km from our point's "total bounds"

# set axes labels
ax.set_xlabel('Eastings UTM 12N (m)')
ax.set_ylabel('Northings UTM 12N (m)')

# set plot title
ax.set_title('Location of Snow Pit 2S10\nAirborne TIR imagery of Grand Mesa, CO (Feb. 8, 2020)');

../../_images/thermal-ir-tutorial_38_0.png

Import the snow temperature timeseries dataset

This data is available through NSIDC, but we have already downloaded a local copy for this tutorial. (See the bonus notebook thermal-ir-data-download.ipynb for more details about data access methods)

The raw data file doesn’t include the column names, so we need to set the column headers following the dataset’s README file.

!cat {TUTORIAL_DATA}/SNEX20_VPTS_Raw/Level-0/snow-temperature-timeseries/README.txt

Description:

	This folder contains data files from two Campbell Scientific CR10X dataloggers which were installed at two sites on Grand Mesa to measure snow temperature profiles as part of the SnowEx 2020 field campaign.

	"Grand Mesa 1" (GM1) measured snow surface temperature with an Apogee SI-111 thermal infrared radiometer mounted on a tripod at a height of about 127 cm above the snow surface. The radiometer pointed 45 degrees off-nadir at the snow surface just to the west of the tripod. Five Campbell Scientific temperature probes were mounted vertically along a wooden stake with zip ties, at depths of 5, 10, 15, 20, and 30 cm below the snow surface. (CR10X_GM1_final_storage_1.dat) 

	"Grand Mesa 2" (GM2) measured snow temperatures with five temperature probes mounted vertically on a wooden stake with zip ties at depths of 5, 10, 15, 20, and 30 cm below the snow surface. This location was just outside the fence perimeter of the Mesa West met station (northwest of the met station tower). Snow surface temperature is measured at the Mesa West station by its own set of thermal infrared radiometers, which can provide snow surface temperature adjacent to this vertical profile of snow temperature. The Mesa West station should also be able to provide similar snow temperatures below the surface with its own string of temperature sensors.

Date/times: 

	GM1: 2-5-2020 11:12 AM MST (UTC−07:00) - 2-12-2020 3:36 PM MST (UTC−07:00)

	GM2: 2-5-2020 12:35 PM MST (UTC−07:00) - 2-12-2020 2:36 PM MST (UTC−07:00)

Locations: 

	GM1 was located at Pit ID 2S10

	GM2 was located at the Mesa West met station 

File formats:

*.dat - comma-separated values from the Campbell Scientific CR10X Dataloggers

*.xlsx - MS Excel worksheet

Columns:

	GM1:

		'table', 'year', 'doy', 'time',

		'rad_avg', 'rad_max', 'rad_min', 'rad_std',

		'sb_avg', 'sb_max', 'sb_min', 'sb_std',

		'temp1_avg', 'temp1_max', 'temp1_min', 'temp1_std',

		'temp2_avg', 'temp2_max', 'temp2_min', 'temp2_std',

		'temp3_avg', 'temp3_max', 'temp3_min', 'temp3_std',

		'temp4_avg', 'temp4_max', 'temp4_min', 'temp4_std',

		'temp5_avg', 'temp5_max', 'temp5_min', 'temp5_std',

		'batt_a','batt_b'

	GM2:

		'table', 'year', 'doy', 'time',

		'temp1_avg', 'temp1_max', 'temp1_min', 'temp1_std',

		'temp2_avg', 'temp2_max', 'temp2_min', 'temp2_std',

		'temp3_avg', 'temp3_max', 'temp3_min', 'temp3_std',

		'temp4_avg', 'temp4_max', 'temp4_min', 'temp4_std',

		'temp5_avg', 'temp5_max', 'temp5_min', 'temp5_std',

		'batt_a','batt_b'

Resources:

	An example jupyter notebook for reading and plotting the data is available on github:

		https://github.com/spestana/snowex2020-snow-temp

Create a list of column headers according to the readme above (for “GM1” which we can read was the datalogger at snowpit 2S10)

column_headers = ['table', 'year', 'doy', 'time', # year, day of year, time of day (local time, UTC-7)
                  'rad_avg', 'rad_max', 'rad_min', 'rad_std', # radiometer surface temperature
                  'sb_avg', 'sb_max', 'sb_min', 'sb_std', # radiometer sensor body temperature (for calibration)
                  'temp1_avg', 'temp1_max', 'temp1_min', 'temp1_std', # temperature at 5 cm below snow surface
                  'temp2_avg', 'temp2_max', 'temp2_min', 'temp2_std', #               10 cm
                  'temp3_avg', 'temp3_max', 'temp3_min', 'temp3_std', #               15 cm
                  'temp4_avg', 'temp4_max', 'temp4_min', 'temp4_std', #               20 cm
                  'temp5_avg', 'temp5_max', 'temp5_min', 'temp5_std', #               30 cm
                  'batt_a','batt_b', # battery voltage data
                 ]

Open the file as a pandas data frame with read_csv

df = pd.read_csv(f'{TUTORIAL_DATA}/SNEX20_VPTS_Raw/Level-0/snow-temperature-timeseries/CR10X_GM1_final_storage_1.dat',
                 header = None, names = column_headers) 

# After the filepath we specify header=None because the file doesn't contain column headers, 
# then we specify names=column_headers to give our own names for each column.

We need to do some formatting of the data fields, but we can preview what we just loaded fist

df.head() # show the first 5 rows of the dataframe

	table	year	doy	time	rad_avg	rad_max	rad_min	rad_std	sb_avg	sb_max	...	temp4_avg	temp4_max	temp4_min	temp4_std	temp5_avg	temp5_max	temp5_min	temp5_std	batt_a	batt_b
0	101	2020	30	1537	22.963	22.963	22.963	0.000	22.832	22.832	...	-99999.0	-99999.0	-99999.0	0.0	-99999.0	-99999.0	-99999.0	0.0	22.96	22.83
1	101	2020	30	1538	22.962	22.963	22.961	0.001	22.832	22.832	...	-99999.0	-99999.0	-99999.0	0.0	-99999.0	-99999.0	-99999.0	0.0	22.96	22.83
2	101	2020	30	1539	25.080	26.810	22.968	1.484	22.832	22.832	...	-99999.0	-99999.0	-99999.0	0.0	-99999.0	-99999.0	-99999.0	0.0	24.47	22.83
3	101	2020	30	1540	25.320	25.569	25.133	0.184	22.839	22.846	...	-99999.0	-99999.0	-99999.0	0.0	-99999.0	-99999.0	-99999.0	0.0	25.42	22.85
4	102	2020	30	1540	12.050	0.000	NaN	NaN	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

5 rows × 34 columns

Data cleanup and formatting

# Create a zero-padded time string (e.g. for 9:30 AM we are changing '930' into '0930')
df['time_str'] = [('0' * (4 - len(str(df.time[i])))) + str(df.time[i]) for i in range(df.shape[0])]

# locate where rows have time_str == 2400 (midnight), and the whole column 'doy'
# where we are at midnight, we need to shift one day forward
df.loc[df['time_str'] == '2400','doy'] += 1

# and then change midnight from '2400' to '0000'
df.time_str.replace('2400', '0000', inplace=True)

This function lets us convert year and day of year (the format that the datalogger uses) to a pandas datetime index:

def compose_date(years, months=1, days=1, weeks=None, hours=None, minutes=None,
                 seconds=None, milliseconds=None, microseconds=None, nanoseconds=None):
    '''Compose a datetime object from various datetime components. This clever solution is from:
        https://stackoverflow.com/questions/34258892/converting-year-and-day-of-year-into-datetime-index-in-pandas'''
    years = np.asarray(years) - 1970
    months = np.asarray(months) - 1
    days = np.asarray(days) - 1
    types = ('<M8[Y]', '<m8[M]', '<m8[D]', '<m8[W]', '<m8[h]',
             '<m8[m]', '<m8[s]', '<m8[ms]', '<m8[us]', '<m8[ns]')
    vals = (years, months, days, weeks, hours, minutes, seconds,
            milliseconds, microseconds, nanoseconds)
    return sum(np.asarray(v, dtype=t) for t, v in zip(types, vals)
               if v is not None)

# Create a datetime value from the date field and zero-padded time_str field, set this as our dataframe's index
df.index = compose_date(df['year'], 
                        days=df['doy'], 
                        hours=df['time_str'].str[:2],
                        minutes=df['time_str'].str[2:])

# Remove entries that are from table "102" (this contains datalogger battery information we're not interested in at the moment)
df = df[df.table != 102]

# drop the columns we no longer need
df.drop(columns=['table','year','doy','time','time_str','batt_a','batt_b'], inplace=True)

Inspect the contents

df.head()

	rad_avg	rad_max	rad_min	rad_std	sb_avg	sb_max	sb_min	sb_std	temp1_avg	temp1_max	...	temp3_min	temp4_avg	temp4_max	temp4_min	temp5_avg	temp5_max	temp5_min
2020-01-30 15:37:00	22.963	22.963	22.963	0.000	22.832	22.832	22.832	0.000	-99999.0	-99999.0	...	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0
2020-01-30 15:38:00	22.962	22.963	22.961	0.001	22.832	22.832	22.832	0.000	-99999.0	-99999.0	...	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0
2020-01-30 15:39:00	25.080	26.810	22.968	1.484	22.832	22.832	22.832	0.000	-99999.0	-99999.0	...	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0
2020-01-30 15:40:00	25.320	25.569	25.133	0.184	22.839	22.846	22.832	0.005	-99999.0	-99999.0	...	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0
2020-01-30 15:41:00	24.894	25.069	24.611	0.182	22.860	22.860	22.860	0.000	-99999.0	-99999.0	...	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0	-99999.0

5 rows × 28 columns

Make a simple plot of the data. We are interested in the variable rad_avg which is the average temperature measured by the radiometer over each 5 minute period.

plt.figure(figsize=(10,4))

# plot radiometer average temperature
df.rad_avg.plot(linestyle='-', marker='', markersize=1, c='k', label='Ground-based $T_s$')


# set axes limits
plt.ylim((-35,5))
plt.xlim((pd.Timestamp(2020,2,5,11,0),pd.Timestamp(2020,2,12,16,0)))

# add a legend to the plot
plt.legend()

# set axes labels
plt.ylabel('Temperature [$C\degree$]')
plt.xlabel('Time')

# add grid lines to the plot
plt.grid('on')

# set the plot title
plt.title('Snow Surface Temperature at Snow Pit 2S10');

../../_images/thermal-ir-tutorial_55_0.png

Bonus plot: look at snow temperatures below the snow surface

Add the following to the above plot to add lines for temperature recorded at each depth interval below the snow surface:

# plot the snow temperature at each depth it was measured
df.temp1_avg.plot(linestyle='-', marker='.', markersize=1, c=[0.8,0.8,1], label='Ts @ -5 cm')
df.temp2_avg.plot(linestyle='-', marker='.', markersize=1, c=[0.6,0.6,1], label='Ts @ -10 cm')
df.temp3_avg.plot(linestyle='-', marker='.', markersize=1, c=[0.4,0.4,1], label='Ts @ -15 cm')
df.temp4_avg.plot(linestyle='-', marker='.', markersize=1, c=[0.2,0.2,1], label='Ts @ -20 cm')
df.temp5_avg.plot(linestyle='-', marker='.', markersize=1, c=[0,0,1], label='Ts @ -30 cm')

But then we want to focus on the date/time when our IR image was from, so zoom in on Feb 8th by changing our plot’s xlim (using pandas Timestamps for the x axis values).

plt.figure(figsize=(10,4))

# plot radiometer average temperature
df.rad_avg.plot(linestyle='-', marker='', markersize=1, c='k', label='Ground-based $T_s$')

# set axes limits
plt.ylim((-15,0)) # set some temperature y-axis limits for our plot
plt.xlim((pd.Timestamp(2020,2,8,6,0),pd.Timestamp(2020,2,8,20,0))) # zoom in to daytime hours on Feb. 8, 2020

# add a legend to the plot
plt.legend()

# set axes labels
plt.ylabel('Temperature [$C\degree$]')
plt.xlabel('Time')

# add grid lines to the plot
plt.grid('on')

# set the plot title
plt.title('Snow Surface Temperature at Snow Pit 2S10');

../../_images/thermal-ir-tutorial_58_0.png

Next steps:¶

Project ideas with these data:¶

Compare more snow pit temperature data (using snowexsql queries) against airborne and satellite IR imagery
Investigate spatial patterns of snow temperature from open areas to forested areas on the mesa, differences between ASTER and airborne IR imagery
Improved data visualization using hvplot or something similar to create interactive plots of IR and/or visible imagery
Compare thermal infrared and SAR imagery, or snow temperature observations and snow model outputs

Note

Contact Steven Pestana during the hackweek for help accessing more airborne IR or visible imagery and related datasets

Data access/download and pre-processing¶

See the jupyter notebook thermal-ir-data-download.ipynb for more details about data access methods through the NASA EarthData API, and pre-processing ASTER data into geotiff images.

SnowEx Hackweek

Thermal Infrared Remote Sensing of Snow¶

Part 1: Comparing airborne IR imagery with ground-truth observations¶

Airborne IR imagery¶

Bonus activity:¶

Ground-based temperature observations¶

Compare Airborne IR against the “ground truth” snow surface temperature¶

Part 2: Satellite IR remote sensing obsevations¶

Satellite IR imagery with ASTER¶

Bonus activity: comparing two thermal IR rasters¶

Next steps:¶

Project ideas with these data:¶

Data access/download and pre-processing¶

Additional learning resources:¶