**4.2. Soil characteristics**

The characterisation of the soil is based on the measurement of its texture on the surface and some physical parameters such as electric conductivity, pH, soil moisture, % of saturation, field capacity (CC) and permanent wilt point (PWP) (see **Table 1**). CC is the largest amount of water that this type of soil will retain under conditions of complete humidity *CC* = (%clay) ∗ *a* + (%sil) ∗ *b* + (%sand) ∗ *c*, and the PWP is the minimum water content where the plants usually die; for each of the sampling soils *PWP* <sup>=</sup> \_\_\_\_ *CC* 1.84, the coefficients a, b and c are determined for each region and type of floor, and in this case, the coefficients used are *a* = 0.555, *b* = 0.187 and *c* = 0.027 [42].

Modesto Ángel where the three techniques were implemented has a more constant type of soil: the first 5 cm is a sandy soil, from 10 to 80 cm is Frank and more than 90 cm is loamy clay. In all the other sites, soil type varies as the depth increases.


**Table 1.** Soil characteristics measured at 10 cm depth in the sites.

**Figure 2.** Distribution of head of precipitation recorded from August 2013 to October 2014 in the Modesto Ángel station and the Automatic Weather Station (EMA)-CONANP station at the north and south of the study of the CBR, respectively.

is the output of the devices (FDR and TDR readings) and *Obsi*

observed gravimetric soil moisture. *RMSE* minimum value is zero under the hypothetical situation that the model is capable of perfect (long-term) readings of the system, and there are no data errors being small values desirable. Mean bias error (MBE) measures the average magnitude of the errors in a set of readings. It is the average over the test sample of the absolute differences between prediction and actual observations having the differences an equal

The results present the climatic variations of the studied area that make it a complex system to analyse the variability of soil moisture in the sampling sites. The test of the TDR and FDR systems can offer more than the evaluation of the accuracy of each system if they are used as well as a complementary study. Also, a datalogger was tested to work in a complex environment in order to guarantee a constant data register in order to monitor water requirements

As it was mentioned, the climate that predominates in the region is warm humid and warm sub-humid, with an average maximum temperature of 36°C during the months of May and June and average minimum temperature of 18°C during January. However, in some occasions during August, the so-called *dog days* is presented, which is the year period where heat

weight. According to [41] an acceptable value for volumetric soil moisture is 0.04 m<sup>3</sup>

is the

 m−<sup>3</sup> .

where subscripts *Di*

64 Soil Moisture

**4.1. Climatic data**

**4. Results and discussion**

and its supply for the different uses in the area.

### **4.3. Soil moisture analysis**

Soil moisture instruments tested report changes in time or frequency related indirectly to the dielectric permittivity to the volumetric water contents. Results are presented per site and per type of technique, and this means that TDR were analysed with Campbell sensors (TDR\_Campbell), FDR using the Decagon sensors (FDR\_Dec) and FDR using Diviner 2000 (FDR\_Div). A specific analysis for the field conditions under the different operating sensors was not necessary since there were the same conditions at the sites (Modesto Ángel with three methods and the other two sites). Individual calibrations per depth offer equations that improve the sensor performance. Then all together were compared in order to know the sensitivity of each one.

#### *4.3.1. FDR and TDR devices at the Modesto Ángel station*

In the Modesto Ángel station, the three devices to measure soil moisture were installed: TDR\_ CS, FDR\_Dec and FDR\_Div. TDR\_CS calibration process includes a first analysis using data collected from the datalogger with the default equations of the device. Both lineal and quadratic equations were tested founding that the lineal equation offered better results than the quadratic one with RE of 0.22 and 0.41, respectively. Secondly, the calibration using the gravimetric data measured at the field was done in terms of volumetric water moisture, θgv. The gravimetric measurements were 10 samples per site from 2012 to 2013 years, at 2.4, 5, 10, 20 30 cm depth. Six more gravimetric samples per site were measured during 2014–2015 to confirm the reliability of the calibration for a different weather, soil and vegetation conditions. The TDR\_CS provides a R<sup>2</sup> of 0.96, with RMSE, MBE and RE values of 0.101 cm<sup>3</sup> cm−<sup>3</sup> , 0.107 cm<sup>3</sup> cm−<sup>3</sup> and 0.12, respectively. The major deviation was observed at the 2.5 and 5 cm, this is because the place where the sensors were installed was not disturbed but the place where the sample was taken, even if it was close to the area of the station, was more susceptible to the surface soil conditions at this specific time. In the case of the FDR\_Dec, R<sup>2</sup> once calibrated the readings was 0.90 with RMSE, MBE and RE values of 0.18 cm<sup>3</sup> cm−<sup>3</sup> , 0.407 cm<sup>3</sup> cm−<sup>3</sup> and 0.083, respectively. For FDR\_Div, results showed that the first 30 cm have the same texture; thus, a calibration equation was obtained for it with an R<sup>2</sup> of 0.86, RMSE 0.086 cm<sup>3</sup> cm−<sup>3</sup> , MBE 0.079 cm<sup>3</sup> cm−<sup>3</sup> and RE 0.069. FDR\_Div was the best device, but it is important to mention that only three depths were tested at 10, 20 and 30 cm. Between TDR\_CS and FDR\_Dec, it is quite difficult to analyse since results are not conclusive, but TDR\_CS could be expected to provide a more consistent value. This is because Decagon devices demonstrated to be more sensible to the weather (see temperature and precipitation results) and to soil conditions at the time of the sampling. However, the RE is major for TDR\_CS. Despite the previous results, the calibration equations were applied to 400 daily θs records with a latency of 20 min for the 2012–2015 period for both TDR\_CS and FDR\_Dec. **Figure 3** plots the behaviour of the devices at a 10 cm depth for 2013–2014.

cases until reaching almost 1.0 of water content. The maximum difference perceived from January to August 2014 was close to 30%. Looking at the FDR system, it is a major agreement for both Decagon and Diviner 2000, although during the dry period, FDR\_Div overestimated more than 50% the soil moisture. The gravimetric measurements for October 31, 2013 and May 05, 2014 were also included in the plot demonstrating that TDR\_CS is closer than FDR\_Dec. The life of the experimental THINNK datalogger without change of battery was from 2013 to 2015. In the case of the Campbell datalogger, it was required to change the battery since the extreme conditions at the field lowered its energy every 6 or 8 months. Also, it was necessary to protect the battery of the Campbell datalogger, whereas in the THINNK, one could be

Correlation between TDR and FDR Soil Moisture Measurements at Different Scales to Establish…

http://dx.doi.org/10.5772/intechopen.81477

67

**Figure 3.** 10 cm depth soil moisture values from July 2013 to October 2014 at the Modesto Ángel station.

The Diviner 2000 allows monitoring different areas once the accessed pipe in each site was installed. Thus, eight aguadas were monitored; Carlos A Madrazo was analysed at the La Ceiba site only. More than 30 readings were registered in the period of 2012–2015, as well as several gravimetric analyses were performed a less two per year. As soil moisture is function of the texture along the profile, this implicates different water aggregation and, in consequence, a different behaviour. For that, along the profile, one could have more than one calibration equations in order to represent what actually happened to soil moisture in the profile. For each site, a graph was developed as shown in **Figure 4** for the Modesto Ángel site with

attached to a tree without more protection.

*4.3.2. FDR diviner for the other sampling sites*

120 cm depth. For this place, three equations were established:

As one can observe, there is a good agreement between TDR\_CS and FDR\_Dec, following a similar pattern taking into account the accuracy of each device. However, there is a major response of the Decagon sensor when rain is presented, being evident an increase in some Correlation between TDR and FDR Soil Moisture Measurements at Different Scales to Establish… http://dx.doi.org/10.5772/intechopen.81477 67

**Figure 3.** 10 cm depth soil moisture values from July 2013 to October 2014 at the Modesto Ángel station.

cases until reaching almost 1.0 of water content. The maximum difference perceived from January to August 2014 was close to 30%. Looking at the FDR system, it is a major agreement for both Decagon and Diviner 2000, although during the dry period, FDR\_Div overestimated more than 50% the soil moisture. The gravimetric measurements for October 31, 2013 and May 05, 2014 were also included in the plot demonstrating that TDR\_CS is closer than FDR\_Dec.

The life of the experimental THINNK datalogger without change of battery was from 2013 to 2015. In the case of the Campbell datalogger, it was required to change the battery since the extreme conditions at the field lowered its energy every 6 or 8 months. Also, it was necessary to protect the battery of the Campbell datalogger, whereas in the THINNK, one could be attached to a tree without more protection.
