a flooding problem

#1
I am a little stuck here with my stats analysis, wondering if anyone could shed some light?

I am testing whether long levee banks (used to protect irrigated land from flooding) have an effect on flood frequency. The null hypothesis would be that the floodwater flows on one side of the levee bank are the same as on the other side.

I have taken 15 random samples of levee banks on the floodplain. For each levee bank sample, I took two 500x100m "paired" adjacent quadrats, one on the river-side and one on the uplands-side with the levee bank in between. Within each quadrat, I counted the percentage of area of flow frequency (for example, 10% of the quadrat area was flooded every year, 39% flooded every 3 years, 18% flooded every 5 years, 5% flooded every 10 years, 28% never flooded).

From my data observations, the quadrats on the uplands-side of the levee bank receive much less water than the quadrats on the river-side. What statistical analysis should I use to quantify this and investigate further into how the flow frequency is affected?

any help would be much appreciated!
cheers
 
#2
The ideal story is taking those 15 random sample of levee banks as 15 blocks, as we are more interested in the effect of long levee bank. within each block, there are two treatments randomly assigned to two experiment units. So it's a RCBD, with the block effect random. (Check any stats book about methods on block design) Then the problem of interest is to test if there's any difference between treatment effects.


But the ideal story needs strong assumption , such as independence, normality, etc. However, our observations may not have those assumptions satisfied, like does not follow from normal distribution. In this case, it'll be helpful to consider GLMM (Generalized Linear Mixed effects Model).