So I'm trying to run our California carbon & flood models on a slightly modified context - a new San Joaquin Valley context. I created the context and try a simple model -d (same results for model -o)
model -d core.models.carbon-ca/vegetation-type core.contexts.beta/san_joaquin512
and get:
"error: resource not found: cannot observe any of the 1 contingencies of model core.models.carbon-ca/vegetation-type"
I can hack it by creating a new model statement that removes vegetation-type from the context list, but I've always been under the impression that the whole point of using Bayesian models is that they'll use data where we have it and priors where we don't. It appears that right now the system isn't smart enough to still run the models on priors - it needs full coverage of all layers in the context list. Correct me if I'm wrong, but this seems like a really clunky way to run our models at the geographic margins of our case studies where one or more data layers may be missing and it would be amazing to seamlessly sub in priors. Am I missing something obvious?
Ciao Ferd -
So I'm trying to run our California carbon & flood models on a slightly modified context - a new San Joaquin Valley context. I created the context and try a simple model -d (same results for model -o)
model -d core.models.carbon-ca/vegetation-type core.contexts.beta/san_joaquin512
and get:
"error: resource not found: cannot observe any of the 1 contingencies of model core.models.carbon-ca/vegetation-type"
I can hack it by creating a new model statement that removes vegetation-type from the context list, but I've always been under the impression that the whole point of using Bayesian models is that they'll use data where we have it and priors where we don't. It appears that right now the system isn't smart enough to still run the models on priors - it needs full coverage of all layers in the context list. Correct me if I'm wrong, but this seems like a really clunky way to run our models at the geographic margins of our case studies where one or more data layers may be missing and it would be amazing to seamlessly sub in priors. Am I missing something obvious?
Ken