Introduction¶

Calibration of a CRNS generally involves finding the so-called "N0" number. This number is the theorised maximum number of neutrons that will be counted by a sensor, over a defined time period, if conditions were completely dry. In neptoon we standardise all our counting rates to counts per hour (cph) and so our N0 number will always be the number of neutrons expected to be counted by a sensor, at a particular site, over a 1 hour integration window, in totally dry conditions.

To convert neutrons to soil moisture we now take the ratio between the actual count rate vs the theorised dry count rate. This ratio places us on the calibration curve which can be seen in Figure 1 below (Franz et al., 2012).

N0 calibration curve showing the relationship between neutron counts and soil moisture.

Figure 1: N0 calibration curve showing the relationship between neutron counts and soil moisture.

Neutron Correction

Neutron count rates are expected to be corrected by the time we get to calibration. The correction removes external influences on the neutron counts (e.g., changes in atmospheric pressure). So the N0 is a corrected term. More on corrections here. This means that if you change your corrections steps you must recalibrate and get a new N0 number. If not your N0 and N numbers are being corrected differently!

Weighting samples¶

The closer a sample is to the sensor, the greater weight it's soil moisture values should be given. The weighting procedure in neptoon is done automatically (Schrön et al., 2017).

Before you calibrate...¶

Before we begin lets describe whats expected at this stage.

You have a CRNSDataHub instance
You have imported your CRNS data into the hub (more on that here)
You have a SensorInformation in the hub (more on that here)
You have collected any external data you need (more on that here)
You have performed some quality assessment on your data (more on that here)
You have corrected your neutron counts so that you have a corrected neutrons column (more on that here)

Your sample data¶

To calibrate a sensor, and calculate the N0 term, we need sample data from around the sensor.

The following columns are required in the calibration data:

Date Time Column (calib_data_date_time_column_name): The column name containing date and time information when samples were collected, formatted according to calib_data_date_time_format (default: '%d.%m.%Y %H:%M')
Profile ID (profile_id_column): Column containing unique identifiers for each soil sampling profile
Distance to Sensor (distance_column): Column with distances from the CRNS sensor to each sampling point in meters
Sample Depth (sample_depth_column): Column with the depth of each soil sample in centimeters
Gravimetric Soil Moisture (soil_moisture_gravimetric_column): Column containing soil moisture measurements in g/g (mass of water per mass of dry soil)
Dry Bulk Density (bulk_density_of_sample_column): Column with soil bulk density values in g/cm³ of each sample
Soil Organic Carbon (soil_organic_carbon_column): Column with soil organic carbon content in g/g of each sample (if missing can be set to 0)
Lattice Water (lattice_water_column): Column with lattice water content in g/g of each sample (if missing can be set to 0)

When multiple days of calibration data are available, these should all be supplied in the same file. Neptoon will automatically split them up using the DateTime column and complete calibration on each day separately, before finding an average N0.

Soil Moisture Units

To be clear in neptoon we expect gravimetric soil moisture values in the calibration data. In the literature describing the weighting (e.g., Schrön et al., 2017), volumetric units are used. We make this conversion internally.

Adding this into the pipeline¶

Bringing this together - we must read in the sample data as a DataFrame, and create a CalibrationConfiguration object. This object lets neptoon know what the different columns are called in the sample data.

from neptoon.calibration import CalibrationConfiguration
import pandas as pd

calib_df = pd.read_csv("example/data/path/calibration_data.csv")
data_hub.calibration_samples_data = calib_df

calibration_config = CalibrationConfiguration(
    calib_data_date_time_column_name='DateTime_utc',
    calib_data_date_time_format='%d.%m.%Y %H:%M', # Also include info on datetime format
    profile_id_column='Profile_ID',
    distance_column='Distance_to_CRNS_m',
    sample_depth_column='Profile_Depth_cm',
    soil_moisture_gravimetric_column='SoilMoisture_g_g',
    bulk_density_of_sample_column='DryBulkDensity_g_cm3',
    soil_organic_carbon_column='SoilOrganicCarbon_g_g',
    lattice_water_column='LatticeWater_g_g',
)
data_hub.calibrate_station(config=calibration_config)

After running this code neptoon will have undertaken calibration using the sample data supplied. Automatically weighting the samples according to the literature and producing an N0 number which is directly saved into the SensorInfo object.

Calibrate without CRNSDataHub¶

It's possible to calibrate your site more directly, without a data hub. For this you will need; 1. A dataframe with pre-corrected CRNS timeseries data and 2. your sample data. For example, you could save your dataframe in your datahub just before calibration and use this to test different variations of your sample data.

Checkout the examples. The example demonstrating this is found under jupyter_notebooks>calibration>example_calibration.ipynb

Koehli et al., (2021) method

If you are supplying your own dataframe, with neutrons already corrected, be sure that the neutrons are corrected appropriately. Humidity correction is done as a standard using the Desilets et al. (2010) method. With the Koehli et al., (2021) method humidity is corrected for in the conversion from neutrons to soil moisture. Given this, when using the Koehli et al., (2021) method, humidity correction using the Rosolem et al., (2013) method should not be applied to corrected neutrons, otherwise double humidity correction occurs.

If you are using neptoon to do the full pipeline - this will be automatically addressed.

Calibrate without data¶

If you already have values for inputs like field average soil moisture, corrected neutron counts, lattice water etc. you can add these directly to the following function to simply optimise the N0.

In the following case it's expected that:

Your soil moisture values are field scale averages using an appropriate weighting scheme
Your corrected neutron counts are corrected for external impacts (e.g., pressure).

Remember that when you find this N0 value, all subsequent neutron corrections and data cleaning methods should match what was done to originally derive your corrected_neutron_counts values. The N0 is a corrected neutron value.

from neptoon.calibration.station_calibration import CalculateN0

n0_calc = CalculateN0()
n0_calc.set_values(
    soil_moisture=[0.4, 0.2],
    corrected_neutron_counts=[1200, 1100],
    lattice_water=0.001,
    water_equiv_soil_organic_carbon=0.02,
)
n0_calc.find_optimal_N0()

or

from neptoon.calibration.station_calibration import CalculateN0

n0_calc = CalculateN0()
n0_calc.set_values(
    soil_moisture=[0.4, 0.2],
    corrected_neutron_counts=[1200, 1100],
    lattice_water=0.001,
    water_equiv_soil_organic_carbon=0.02,
    absolute_humidity=[2.35, 2.50],  # for koehli
    conversion_method="koehli_etal_2021",
)
n0_calc.find_optimal_N0()

References¶

Franz, T. E., Zreda, M., Rosolem, R., and Ferre, T. P. A.: A universal calibration function for determination of soil moisture with cosmic-ray neutrons, Hydrol. Earth Syst. Sci., 17, 453–460, https://doi.org/10.5194/hess-17-453-2013, 2013.

Schrön, M., Köhli, M., Scheiffele, L., Iwema, J., Bogena, H. R., Lv, L., Martini, E., Baroni, G., Rosolem, R., Weimar, J., Mai, J., Cuntz, M., Rebmann, C., Oswald, S. E., Dietrich, P., Schmidt, U., and Zacharias, S.: Improving calibration and validation of cosmic-ray neutron sensors in the light of spatial sensitivity, Hydrol. Earth Syst. Sci., 21, 5009–5030, https://doi.org/10.5194/hess-21-5009-2017, 2017.