thurber
commented
2 years ago notes from 1/12
Farmer ABM
files
- nldas_ids.p - not scenario specific
- nldas.txt - not scenario specific
- water_constraints_by_farm_pyt278.p - not used
- hist_avail_bias_correction_20201102.csv - based on ABM calibration
- hist_avail_bias_correction_live.csv - written each time
- hist_dependent_storage.csv - based on ABM calibration (historic run -> average historic storage levels)
- MOSART_WM_PMP_inputs_20201028.xlsx - ABM calibration / scenario conditions (could change based on scenario)
- has some duplicated data with the below
- crop_ids_by_farm.p - join of farms to crops (probably can be done all in code)
- max_land_constr_20201102_protocol2.p - scenario specific
- gammas_new_20201102_protocol2.p - calibration
- net_prices_new_20201102_protocol2.p - calibration
- 25583.64 - sqft to acre -
- abmresults - these outputs have individual agent crop land allocations for each year -> ultimately converted to demand
- lines 274 - 277 - alignment of grid cell with nldas id
demand flags / returnflow /groundwater
- needs to be ready in tandem with ABM
near term experiments:
- groundwater capability: mid-2022
- cover historical period
wishlist:
- class instance for each agent (i.e. grid cell)
- currently, all agent decisions are solved in single optimization problem, but pyomo may not be optimal for solving single problems (but may be efficient at batches of ~1000 agents)
erexer
Transfer Jim's Agent Based Model of farmer crop rotation into
mosartwmpy(it is currently only implemented iniwmm)