A neural network approach to smarter sensor networks for water quality monitoring.

Environmental monitoring is evolving towards large-scale and low-cost sensor networks operating reliability and autonomously over extended periods of time. Sophisticated analytical instrumentation such as chemo-bio sensors present inherent limitations because of the number of samples that they can take. In order to maximize their deployment lifetime, we propose the coordination of multiple heterogeneous information sources. We use rainfall radar images and information from a water depth sensor as input to a neural network (NN) to dictate the sampling frequency of a phosphate analyzer at the River Lee in Cork, Ireland. This approach shows varied performance for different times of the year but overall produces output that is very satisfactory for the application context in question. Our study demonstrates that even with limited training data, a system for controlling the sampling rate of the nutrient sensor can be set up and can improve the efficiency of the more sophisticated nodes of the sensor network.

Repeatability of gaseous measurements across consecutive days in sheep using portable accumulation chambers.

Portable accumulation chambers (PACs) enable gaseous emissions from small ruminants to be measured over a 50-min period; to date, however, the repeatability of consecutive days of measurement in the PAC has not been investigated. The objectives of this study were 1) to investigate the repeatability of consecutive days of gaseous measurements in the PAC, 2) to determine the number of days required to achieve precise gaseous measurements, and 3) to develop a prediction equation for gaseous emissions in sheep. A total of 48 ewe lambs (c. 10 to 11 mo of age) were randomly divided into four measurement groups each day, for 17 consecutive days. Gaseous measurements were conducted between 0800 and 1200 hours daily. Animals were removed from perennial ryegrass silage for at least 1 h before measurements in the PAC, and animals were assigned randomly to each of the 12 chambers. Methane (CH4; ppm) concentration, oxygen (O2; %), and carbon dioxide (CO2; %) were measured at three time points (0, 25, and 50 min after entry of the first animal into the first chamber). To quantify the effect of animal and day variation on gaseous emissions, between-animal, between-day, and error variances were calculated for each gaseous measurement using a linear mixed model. The number of days required to gain a certain precision (defined as the 95% confidence interval range) for each gaseous measurement was also calculated. For all three gases, the between-day variance (39% to 40%) accounted for a larger proportion of total variance compared with between-animal variance, while the repeatability of 17 consecutive days of measurement was 0.36, 0.31, and 0.23 for CH4, CO2, and O2, respectively. Correlations between consecutive days of measurement were strong for all three gases; the strongest correlation between day 1 and the remaining days for CH4, CO2, and O2 was 0.71 (days 1 and 6), 0.77 (days 1 and 2), and 0.83 (days 1 and 5), respectively. A high level of precision was achieved when gaseous measurements from PAC were taken over three consecutive days. The prediction equation overestimated gaseous production for all three gases: the correlations between actual and predicted gaseous output ranged from 0.67 to 0.71, with the r2 ranging from 0.45 to 0.71. The results from this study will aid the refinement of the protocol for the measurement of gaseous emissions in sheep using the PAC.

Investigation of intra-day variability of gaseous measurements in sheep using portable accumulation chambers.

Portable accumulation chambers (PAC) enable short-term spot measurements of gaseous emissions including methane (CH4), carbon dioxide (CO2), and oxygen (O2) consumption from small ruminants. To date the differences in morning and evening gaseous measurements in the PAC have not been investigated. The objectives of this study were to investigate: 1) the optimal measurement time in the PAC, 2) the appropriate method of accounting for the animal's size when calculating the animal's gaseous output, and 3) the intra-day variability of gaseous measurements. A total of 12 ewe lambs (c. 10 to 11 months of age) were randomly selected each day from a cohort of 48 animals over nine consecutive days. Methane emissions from the 12 lambs were measured in 12 PAC during two measurement runs daily, AM (8 to 10 h) and PM (14 to 16 h). Animals were removed from Perennial ryegrass silage for at least 1 h prior to measurements in the PAC and animals were assigned randomly to each of the 12 chambers. Methane (ppm) concentration, O2 and CO2 percentage were measured at 5 time points (T1 = 0.0 min, T2 = 12.5 min, T3 = 25.0 min, T4 = 37.5 min, and T5 = 50.0 min from entry of the first animal into the first chamber) using an Eagle 2 monitor. The correlation between time points T5-T1 (i.e., 50 min minus 0 min after entry of the animal to the chamber) and T4-T1 was 0.95, 0.92, and 0.77 for CH4, O2, and CO2, respectively (P < 0.01). The correlation between CH4 and CO2 output and O2 consumption, calculated with live-weight and with body volume was 0.99 (P < 0.001). The correlation between the PAC measurement recorded on the same animal in the AM and PM measurement runs was 0.73. Factors associated with CH4 production included: day and time of measurement, the live-weight of the animal and the hourly relative humidity. Results from this study suggest that the optimal time for measuring an animal's gaseous output in the PAC is 50 min, that live-weight should be used in the calculation of gaseous output from an animal and that the measurement of an animal's gaseous emissions in either the AM or PM does not impact on the ranking of animals when gaseous emissions are measured using the feeding and measurement protocol outlined in the present study.