Limits...
Assessing inter-sensor variability and sensible heat flux derivation accuracy for a large aperture scintillometer.

Rambikur EH, Chávez JL - Sensors (Basel) (2014)

Bottom Line: Average H fluxes were compared between LAS units and between LAS and EC.It is possible that the LAS physical misalignment may have caused edge-of-beam signal noise as well as vulnerability to signal noise from wind-induced vibrations, both having an impact on the solution of H.In addition, there were some uncertainties in the solutions of H from the LAS and EC instruments, including lack of energy balance closure with the EC unit.

View Article: PubMed Central - PubMed

Affiliation: Department of Civil and Environmental Engineering Colorado State University, 1372 Campus Delivery, Fort Collins, CO 80523, USA. evan.rambikur@gmail.com.

ABSTRACT
The accuracy in determining sensible heat flux (H) of three Kipp and Zonen large aperture scintillometers (LAS) was evaluated with reference to an eddy covariance (EC) system over relatively flat and uniform grassland near Timpas (CO, USA). Other tests have revealed inherent variability between Kipp and Zonen LAS units and bias to overestimate H. Average H fluxes were compared between LAS units and between LAS and EC. Despite good correlation, inter-LAS biases in H were found between 6% and 13% in terms of the linear regression slope. Physical misalignment was observed to result in increased scatter and bias between H solutions of a well-aligned and poorly-aligned LAS unit. Comparison of LAS and EC H showed little bias for one LAS unit, while the other two units overestimated EC H by more than 10%. A detector alignment issue may have caused the inter-LAS variability, supported by the observation in this study of differing power requirements between LAS units. It is possible that the LAS physical misalignment may have caused edge-of-beam signal noise as well as vulnerability to signal noise from wind-induced vibrations, both having an impact on the solution of H. In addition, there were some uncertainties in the solutions of H from the LAS and EC instruments, including lack of energy balance closure with the EC unit. However, the results obtained do not show clear evidence of inherent bias for the Kipp and Zonen LAS to overestimate H as found in other studies.

No MeSH data available.


Related in: MedlinePlus

Regression plots for sensible heat flux (H, W·m−2) of LAS-1, LAS-2, and LAS-3, for 2–7 July (a,b), 8–21 July (c,d), and 22 July–3 August (e,f); LAS-1 is reference. Results filtered for precipitation only. Dashed line represents best-fit linear regression; solid line represents the 1:1 relationship. Regression slopes were statistically significantly different from 1.0 in each case (Table A1).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3958211&req=5

f4-sensors-14-02150: Regression plots for sensible heat flux (H, W·m−2) of LAS-1, LAS-2, and LAS-3, for 2–7 July (a,b), 8–21 July (c,d), and 22 July–3 August (e,f); LAS-1 is reference. Results filtered for precipitation only. Dashed line represents best-fit linear regression; solid line represents the 1:1 relationship. Regression slopes were statistically significantly different from 1.0 in each case (Table A1).

Mentions: The results for the inter-LAS H comparison are presented for three data subsets, as shown in Table 1, based on the above discussion of the alignment issues over the period of record. Figure 4 shows the relationship between the H solutions from each LAS unit using scatter plots and Figure 5 shows the H solution time series for sample days. Comparison of H was also facilitated using summary statistics including the mean bias error divided by mean of the absolute value of observation H (MBE//Ō/). This parameter is an alternative relative deviation parameter which can be used if mean relative deviation is considered unrealistic due to effects of unbounded relative deviation for near zero values of (e.g.,) H. Reference (/Ō/) data were taken from LAS-1 for the inter-LAS comparison, since LAS-1 did not become misaligned during the study period. MBE and /Ō/ represent period averages in W·m−2. The inter-comparison results represent LAS data processed using the u*1-L method. Additional dataset statistics beyond those reported here are provided in a table in the Appendix, referenced as Table A1 in the report body. For the first week when all units were well aligned, there was very little scatter between the H solutions for any of the LAS units (Figure 4a,b). However, there was a mean bias (MBE//Ō/) of approximately +11% for LAS-2 relative to LAS-1 and +9% for LAS-3 relative to LAS-1. Units LAS-2 and LAS-3 were very well correlated with little bias. Following the decrease in signal strength of LAS-2 and LAS-3 observed on 8 July, scatter increased between all LAS units and bias increased between LAS-2 and LAS-1 (Figure 4c,d). The MBE//Ō/ between LAS-2 and LAS-1 increased to 24% and, in addition, disagreement in H pattern between LAS units was apparent for afternoon/nighttime periods associated with larger wind speeds, generally about 8 m s−1. Further, an MBE//Ō/ of 7% was observed between LAS-2 and LAS-3, making apparent that the slip in alignment did not affect LAS-2 and LAS-3 the same. Notably, the trend between LAS-3 and LAS-1 did not appear to change after the 8 July slip in alignment despite the observed increase in scatter (Figure 4d). After the complete misalignment of LAS-2 late 21 July along with the improved alignment of LAS-3, the level of scatter and bias between LAS-2 and LAS-1 remained similar to the prior subset, but the scatter and bias between LAS-3 and LAS-1 were reduced (Figure 4e,f). The MBE//Ō/ value between LAS-3 and LAS-1 was reduced from 18% to 6%, which was lower even than the 9% bias observed during the first subset. It is notable that despite the signal strength of LAS-2 being near zero, the general diurnal pattern in HLAS-2 was similar to that of LAS-1 and LAS-3 (Figure 5b). Furthermore, the deviation between LAS-2 and LAS-1 H was not larger than for the prior period when LAS-2 had approximately 20% signal strength. For each comparison period, mean H values between all LAS units were significantly different (Table A1).


Assessing inter-sensor variability and sensible heat flux derivation accuracy for a large aperture scintillometer.

Rambikur EH, Chávez JL - Sensors (Basel) (2014)

Regression plots for sensible heat flux (H, W·m−2) of LAS-1, LAS-2, and LAS-3, for 2–7 July (a,b), 8–21 July (c,d), and 22 July–3 August (e,f); LAS-1 is reference. Results filtered for precipitation only. Dashed line represents best-fit linear regression; solid line represents the 1:1 relationship. Regression slopes were statistically significantly different from 1.0 in each case (Table A1).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3958211&req=5

f4-sensors-14-02150: Regression plots for sensible heat flux (H, W·m−2) of LAS-1, LAS-2, and LAS-3, for 2–7 July (a,b), 8–21 July (c,d), and 22 July–3 August (e,f); LAS-1 is reference. Results filtered for precipitation only. Dashed line represents best-fit linear regression; solid line represents the 1:1 relationship. Regression slopes were statistically significantly different from 1.0 in each case (Table A1).
Mentions: The results for the inter-LAS H comparison are presented for three data subsets, as shown in Table 1, based on the above discussion of the alignment issues over the period of record. Figure 4 shows the relationship between the H solutions from each LAS unit using scatter plots and Figure 5 shows the H solution time series for sample days. Comparison of H was also facilitated using summary statistics including the mean bias error divided by mean of the absolute value of observation H (MBE//Ō/). This parameter is an alternative relative deviation parameter which can be used if mean relative deviation is considered unrealistic due to effects of unbounded relative deviation for near zero values of (e.g.,) H. Reference (/Ō/) data were taken from LAS-1 for the inter-LAS comparison, since LAS-1 did not become misaligned during the study period. MBE and /Ō/ represent period averages in W·m−2. The inter-comparison results represent LAS data processed using the u*1-L method. Additional dataset statistics beyond those reported here are provided in a table in the Appendix, referenced as Table A1 in the report body. For the first week when all units were well aligned, there was very little scatter between the H solutions for any of the LAS units (Figure 4a,b). However, there was a mean bias (MBE//Ō/) of approximately +11% for LAS-2 relative to LAS-1 and +9% for LAS-3 relative to LAS-1. Units LAS-2 and LAS-3 were very well correlated with little bias. Following the decrease in signal strength of LAS-2 and LAS-3 observed on 8 July, scatter increased between all LAS units and bias increased between LAS-2 and LAS-1 (Figure 4c,d). The MBE//Ō/ between LAS-2 and LAS-1 increased to 24% and, in addition, disagreement in H pattern between LAS units was apparent for afternoon/nighttime periods associated with larger wind speeds, generally about 8 m s−1. Further, an MBE//Ō/ of 7% was observed between LAS-2 and LAS-3, making apparent that the slip in alignment did not affect LAS-2 and LAS-3 the same. Notably, the trend between LAS-3 and LAS-1 did not appear to change after the 8 July slip in alignment despite the observed increase in scatter (Figure 4d). After the complete misalignment of LAS-2 late 21 July along with the improved alignment of LAS-3, the level of scatter and bias between LAS-2 and LAS-1 remained similar to the prior subset, but the scatter and bias between LAS-3 and LAS-1 were reduced (Figure 4e,f). The MBE//Ō/ value between LAS-3 and LAS-1 was reduced from 18% to 6%, which was lower even than the 9% bias observed during the first subset. It is notable that despite the signal strength of LAS-2 being near zero, the general diurnal pattern in HLAS-2 was similar to that of LAS-1 and LAS-3 (Figure 5b). Furthermore, the deviation between LAS-2 and LAS-1 H was not larger than for the prior period when LAS-2 had approximately 20% signal strength. For each comparison period, mean H values between all LAS units were significantly different (Table A1).

Bottom Line: Average H fluxes were compared between LAS units and between LAS and EC.It is possible that the LAS physical misalignment may have caused edge-of-beam signal noise as well as vulnerability to signal noise from wind-induced vibrations, both having an impact on the solution of H.In addition, there were some uncertainties in the solutions of H from the LAS and EC instruments, including lack of energy balance closure with the EC unit.

View Article: PubMed Central - PubMed

Affiliation: Department of Civil and Environmental Engineering Colorado State University, 1372 Campus Delivery, Fort Collins, CO 80523, USA. evan.rambikur@gmail.com.

ABSTRACT
The accuracy in determining sensible heat flux (H) of three Kipp and Zonen large aperture scintillometers (LAS) was evaluated with reference to an eddy covariance (EC) system over relatively flat and uniform grassland near Timpas (CO, USA). Other tests have revealed inherent variability between Kipp and Zonen LAS units and bias to overestimate H. Average H fluxes were compared between LAS units and between LAS and EC. Despite good correlation, inter-LAS biases in H were found between 6% and 13% in terms of the linear regression slope. Physical misalignment was observed to result in increased scatter and bias between H solutions of a well-aligned and poorly-aligned LAS unit. Comparison of LAS and EC H showed little bias for one LAS unit, while the other two units overestimated EC H by more than 10%. A detector alignment issue may have caused the inter-LAS variability, supported by the observation in this study of differing power requirements between LAS units. It is possible that the LAS physical misalignment may have caused edge-of-beam signal noise as well as vulnerability to signal noise from wind-induced vibrations, both having an impact on the solution of H. In addition, there were some uncertainties in the solutions of H from the LAS and EC instruments, including lack of energy balance closure with the EC unit. However, the results obtained do not show clear evidence of inherent bias for the Kipp and Zonen LAS to overestimate H as found in other studies.

No MeSH data available.


Related in: MedlinePlus