drones: application of novel technologies to increase ... · drones offer significant potential to...
TRANSCRIPT
Drones: Application of novel technologies to increase efficiency of
water sampling methodologies
Heather Lally, Ian O’Connor, Liam Broderick, Mark Broderick, Olaf Jensen, Conor Graham
IntroductionLarge scale hydrological monitoring programmes of large open lakes(e.g. WFD Lake Monitoring programme) require for followingconsiderations:
Introduction
Drones offer significant potential to collect water samples &hydrochemical data
May prove significantly safer, cheaper, and time and resourceefficient
Significant limitations were highlighted:
• Volume of water collected is low (60-330 ml)
• Successful and consistent water capture rates are low (60-83%)
• Significant differences in hydrochemical parameters obtained via drone samples versus boat samples
• No cost benefit analyses conducted
Research Aims
Assess the applicability of drones for openlake sampling
Evaluate whether water samples andphysico-chemical data collected using
drones satisfies the WFD objectives
Examine whether drones can offer a
quicker, cost effective, less labour
intensive and safer lake sampling protocol
DJI Matrice 600 Pro
Methods & Materials
Zodiac inflatable boat
Water sampling payload
Experimental Field Trials
• Paired water samples (3 boat and 3 drone) at each sampling site
• 4 in-situ real-time water chemistry parameters: pH, dissolved oxygen, conductivity, temperature
• >10 water chemistry parameters: chlorophyll a, nitrate, nitrite, total oxidised nitrogen, ammonia, ortho-phosphate, alkalinity, colour, total phosphorus, heavy metals
• Sample capture success rates recorded
No. of sampling
sites
Surface area (ha)
Included in the EPA WFD
Lake MP
WFD Alkalinity Status*
WFD Typology
Class*
WFD Status**
Lough Fee 1 174 - - -Lough Inagh 1 310 - - -
Lough Conn 2 4,704 ✓ High 12 ModerateLough Derg 3 13,000 ✓ High 12 Moderate
Lough Mask 3 8,218 ✓ High 12 GoodBallyquirkelake
2 73.6 ✓ Moderate 6 Bad
Results – Experimental Field Trials
1. Volume of water sampled
Previous Research:• 60 ml (Ore et al. 2013; 2015, Detweiler et al. 2015; Chung et al. 2015, Song
et al. 2017)
• 130 ml (Koparan & Koc 2016; Koparan et al., 2018a,b)
• 250-330 ml (Terada et al. 2018)
• 390 ml (Koparan et al. 2020)
This study: 2 L
Results – Experimental Field Trials
2. Sampling success rates
Previous Research:• 90% (Ore et al. 2013) for indoor trials but 69-83% outdoor trials
(Ore et al. 2013; 2015)
• 60-66% (Koparan & Koc 2016; Koparan et al., 2018)
• 96% (Koparan et al. 2020)
This study: 100 %
Results – Experimental Field Trials
3. Comparison of variables
Previous Research: Majority didn’t compare results• Temperature change of ~1°C (Ore et al. 2013,2015; Detweiler et al.
2015)
• Temperature change of ~0.5°C (Chung et al. 2015)
• Temperature different? (Koparan & Koc 2016)
• Temperature similar (Song et al. 2017)
• Koporan et al. (2018) incorporated statistical tests• Temperature (t=0.12,d.f.=18,p=0.91); Conductivity (t=1.59,
d.f.=18,p=0.13)• Oxygen (t=10.1, df=18, p<0.001, mean diff.= 0.25mg/L) (7.18
mg/L vs. 6.93mg/L)• pH (t=3.3,d.f.18, p=0.004, mean diff.= 0.04)• Chloride (t=-12.1,d.f,=18, p<0.001, mean diff.=1.49mg/L)
(5.46mg/L vs.3.97mg/L)
Boat
Drone
Results – Experimental Field Trials
10
30
50
70
90
110
130
150
170
10 30 50 70 90 110 130 150 170
Only used sites if results were
above LOD for all 6 replica samples
This study
Boat
Results – Experimental Field Trials
Paired t = -0.416, d.f. = 9, p = 0.69
Mean diff. = -1.27
Lough Derg Site 1
Lough Derg Site 2
Lough Derg Site 3
Lough Mask Site 1
Lough Mask Site 2
Lough Mask Site 3
Ballyquirke Site 1
Ballyquirke Site 2
Lough Inagh
Lough Fee
Lough Conn Site 1
Lough Conn Site 2
This study
Drone
Boat
Drone
Results – Experimental Field Trials
Paired t = -1.99, d.f. = 7, p = 0.09
Mean diff. = -0.25
Lough Derg Site 1
Lough Derg Site 2
Lough Derg Site 3
Lough Mask Site 1
Lough Mask Site 2
Lough Mask Site 3
Ballyquirke Site 1
Ballyquirke Site 2
Lough Inagh
Lough Fee
Lough Conn Site 1
Lough Conn Site 2
This study
Boat
Drone
Results – Experimental Field Trials
Paired t = 1.19, d.f.= 4, p = 0.3
Mean diff. = 0.0005
Lough Derg Site 1
Lough Derg Site 2
Lough Derg Site 3
Lough Mask Site 1
Lough Mask Site 2
Lough Mask Site 3
Ballyquirke Site 1
Ballyquirke Site 2
Lough Inagh
Lough Fee
Lough Conn Site 1
Lough Conn Site 2
This study
Boat
Drone
Results – Experimental Field Trials
Paired t = -2.46, d.f.= 11, p = 0.031
Mean diff.=-0.048
Lough Derg Site 1
Lough Derg Site 2
Lough Derg Site 3
Lough Mask Site 1
Lough Mask Site 2
Lough Mask Site 3
Ballyquirke Site 1
Ballyquirke Site 2
Lough Inagh
Lough Fee
Lough Conn Site 1
Lough Conn Site 2
This study
Results – Experimental Field Trials
Variable Test statistic
No. of pairs p value Mean difference
True colour* -0.872 12 0.41 -0.78
Hardness* 0.85 6 0.43 1.67
Silica* 0.89 12 0.39 0.06
TON* 0.775 4 0.5 0.002
Chloride# -0.614 10 0.54 0.04
Temperature# -0.94 12 0.35 -0.017
Conductivity* 1.89 12 0.09 12.2
Dissolved oxygen#
-0.63 12 0.53 0.056
*Paired t-test #Wilcoxon signed rank testTON = Total oxidised nitrogen
This study
Results – Experimental Field Trials
4. Comparison of variability
Assessment of precision between boat & drone• Calculated the coefficient of variation for each variable, for each
method, at each site (n=3)
• Firstly, tested overall CoV of all variables between both methods:
Z = -0.197, p = 0.85 (Averages: boat 3.29%; drone 3.76%)
• Secondly, compared CoV for each variable and compared each statistically • Gives assessment of precision between boat & drone
Assessment of precision between boat & drone
*Square root transformed, #Wilcoxon signed rank test, TON= Total oxidised nitrogen, TP = Total phosphorous
Variable Test statistic
No. of pairs
p value Average CV%
Boat Drone
True colour* -1.54 12 0.15 2.9 4.3
Hardness 2.87 6 0.043 7.3 3.9
Silica* -0.89 12 0.4 7.6 8.7
TON* -0.19 4 0.86 1.3 1.6
Chloride -0.15 10 0.89 0.9 0.9
Alkalinity# -0.77 10 0.44 6.6 9.4
Chlorophyll-a -1.25 8 0.25 6.4 8.9
TP# -0.41 5 0.69 6.9 5.9
pH# -1.81 12 0.07 1.1 0.6
Temperature# -1.73 12 0.08 0.6 0.4
Conductivity* 0.08 12 0.94 0.6 0.6
Dissolved oxygen 0.2 12 0.84 0.5 0.5
Results - Cost Benefit Analyses
How much time does it take to get a sample?
Estimated time for a
boat sample
Estimated time for a drone
sample
Key differences
Sampling time(3 reps)
0:49:21 0:22:38 Inflatable boat set up & dismantling
No. of persons 2 2
Total person minutes 1:38:42 0:45:16
Biosecurity measures 0:26:49 0:11:03 More equipment needing disinfecting with boat sampling
No. of persons 2 2
Total person minutes 0:53:38 0:22:06
Total estimated person minutes per sampling site
2:32:20 1:07:22
Results - Cost Benefit Analyses
Drone sampling X 2.3 times
faster
Drone sampling X 3 times faster
Drone sampling X 3.4 times
faster
Boat sampling X 2.6 times
cheaper
In Conclusion• Water samples & physio-chemical data collected by
drones satisfy Ireland’s WFD objectives for lake watermonitoring
• Drone is approximately 3 times faster than boat
• Boat is approximately half the capital outlay
compared to the drone system
• Drone is far safer than boat, for both personnel and
reduces the biosecurity risk
Overall, use of drone reduces the cost of sampling, is
safer and reduces the risk of invasive species being
introduced
Acknowledgements This research (2017–W–MS-28) is funded by the Environmental
Protection Agency (EPA) Ireland as part of their EPA Research Programme
2014-2020
EPA Steering Committee (Dr Ruth Little, Ray Smith, Gerard Hannon, Dr
Iq Mead & Cecilia Hegarty)
Alan Stephens & team EPA Castlebar
Research Assistants: Ashley Johnson & Nicole Perich
Maura O’Connor & Colin Folan – Lough Inagh
Bowen Ormsby - Lough Feeagh
Thank you!!Twitter: @DroPLEtS18, @heatherLally;
@MFRCGMIT; @modelhelisrvcs; @ioconn; @TheConorGraham
Facebook: Marine and Freshwater Research Centre; Model Heli Services
Website: dronesforlakesampling.com