I used WorldView 2 image for extraction of tree crown in farmed parkland where trees are sparse, there may be some errors between extracted tree crown and actual tree crown size. I also have Above ground biomass (AGB) data for some reference plots. Now I want to build a model using AGB as the dependent variable and Tree Crown as independent variable based on 50 reference plots, each of reference plot is 50 by 50 meter. I will use 30 plots for training model and 20 for validating a model.

For simple least square regression, there is an assumption the independent variable (tree crown in my case) should be free from error.

Which method should I use for relating AGB with tree crown cover, either simple linear regression or RMA in my study area?

Thanks and regards