multiple sensors -- fetchSCAN about soildb

[dylanbeaudette]

There are apparently some SCAN sites with multiple below-ground sensors too.

[dylanbeaudette]

Several possible solutions:

1. include all sensors and add an ID (done)
   • include a message when there are > 2 IDs per sensor type
2. keep the first sensor (current approach for above-ground sensors)
3. keep the sensor with the most data associated with it (by year)
4. apply an aggregation function

[dylanbeaudette]

Examples to follow-up on.

##  station 482
##
##  WTEQ.I WTEQ.I-2 PREC.I PREC.I-2 TOBS.I TOBS.I-2 TOBS.I-3 TMAX.D TMIN.D TAVG.D SNWD.I SMS.I_8 STO.I_8

## example of multiple versions of the below-ground sensors
## due to sensor repair, replacement, upgrades, loss of function, ???
##
##   station 2196
##
##  "STO.I-1:-2", "STO.I-1:-4", "STO.I-1:-8", "STO.I-1:-20", "STO.I-1:-40",
##  "STO.I-2:-2", "STO.I-2:-4", "STO.I-2:-8", "STO.I-2:-20", "STO.I-2:-40"

[brownag]

## example of multiple versions of the below-ground sensors
## due to sensor repair, replacement, upgrades, loss of function, ???

RE: site 2196

I am pretty sure that these are actually duplicate below-ground sensors. Both sensors are active as of 2021. The values are very similar but not identical.

library(soilDB)

f <- fetchSCAN(2196, 2021)
#> multiple sensors per depth [site 2196] STO.I_2, STO.I_4, STO.I_8, STO.I_20, STO.I_40, STO.I-2_2, STO.I-2_4, STO.I-2_8, STO.I-2_20, STO.I-2_40
#> 7301 records (0.33 Mb transferred)
f1 <- subset(f$STO, f$STO$sensor.id == "STO.I_2")
f2 <- subset(f$STO, f$STO$sensor.id == "STO.I-2_2")

plot(f1$value~f1$Date, type="l", lty = 2)
lines(f2$value~f2$Date, lty = 3)

summary(lm(f1$value ~ f2$value))
#> 
#> Call:
#> lm(formula = f1$value ~ f2$value)
#> 
#> Residuals:
#>       Min        1Q    Median        3Q       Max 
#> -0.214075 -0.051135 -0.000856  0.044210  0.185177 
#> 
#> Coefficients:
#>               Estimate Std. Error t value Pr(>|t|)    
#> (Intercept) -0.0991436  0.0058147  -17.05   <2e-16 ***
#> f2$value     1.0093109  0.0004542 2222.12   <2e-16 ***
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> 
#> Residual standard error: 0.06935 on 363 degrees of freedom
#> Multiple R-squared:  0.9999, Adjusted R-squared:  0.9999 
#> F-statistic: 4.938e+06 on 1 and 363 DF,  p-value: < 2.2e-16

[dylanbeaudette]

Good, those are simpler to account for. Keep in mind those notes and this issue has to do with all of the potential reasons for multiple sensors / type / depth. These were examples to follow-up on.

[brownag]

Do you have an example site ID where a sensor was replaced? I'd like to see one of those

[brownag]

An example might be sitenum=783