Environmental Data Analysis BC ENV 3017
Indoor - Outdoor PM lab
The goal of this lab is to build a mathematical model of the experiment
that was conducted innorder to better understand the processes controlling
the indoor PM concentrations.
Part 2
-
download the PM data
-
you are looking at the particle numbers from 0.3 to 0.5 mm
in 1000 parts/m3
-
convert the date and time into hours > midnight of 7/2/99
-
plot the timeseries of indoor and outdoor particle concentrations, note
that the data have slightly different time scales (tip: plot one time series
first and than past the second one in)
-
describe the graph
-
in the next two columns, add l and R as
a function of time. Assume that R is constant for the whole time period
and that l before SF6 experiment
#1 is the same as during experiment #2. Also assume that l
was
constant between experiment 1 and 3 and use the average l
for
this period
-
now calculate the Cin(t) model as we discussed in class
using the equation that we derived:
Cin(t+Dt) = Cin(t)*
(1-Dt *(l + R)) +
(Cout(t)+Cout(t+Dt))/2
*Dt *l
-
you will see unrealistic oszillations in the model output => we will need
to make the timestep smaller
-
the following formula reduces the timestep by half:
-
Cin(t+Dt) = Cin(t)* (1-Dt/2
*(l + R))2 + (1-Dt/2
*(l + R))*(Cout(t)+Cout(t+Dt))/2
*Dt/2 *l + (Cout(t)+Cout(t+Dt))/2
*Dt/2 *l
-
(alternatively, you can download the modified data
set in which the time step was reduced by a factor of 2 and use the same
equation as you used initially)
-
play with R and see when you get the best fit between modeled and observed
in door PM concentration
-
if you adssume that the measured air exchange coefficients are not correct,
how would they have to be adjusted in order to improve the fit of the model?
-
highlight on the graph when the windows were open or sealed (see for example
the PM
timeseries for the 2 to 5 mm size fraction)
-
how can you explain differences between modeled and observed data?
