Decompose time-series signal into different components

Question

I am working with time series data from a sensor that measures particulate matter. I know that there is an ambient/ background concentration (which always exists and can naturally vary due to various conditions) and concentration due to human/other activity (for example vehicular emissions or drilling/mining in the area). I want to take the particulate matter concentration from my sensor and split into two components:

1. Background Concentration
2. Concentration due to human activity/ other activity

I am looking for theory/ any algorithm that can help me solve this problem. I am having trouble to figure where to start. For the purpose of this question I generated data in R.

set.seed(100)
true_background <- rnorm(n = 1000, mean = 5, sd = 0.5) #Assumption here, true background has a mean 5

spike_signal <- rgamma(n = 1000, shape = 2.7, rate = 0.08)

observed_signal <- data.frame(true_background + spike_signal)

My desired output would be a data frame that splits the observed_signal into observed_background and observed_spike. Thanks.