-
Notifications
You must be signed in to change notification settings - Fork 0
/
simple_model_env.py
244 lines (217 loc) · 13.4 KB
/
simple_model_env.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
# Copyright 2020-2021, The Johns Hopkins University Applied Physics Laboratory LLC
# All rights reserved.
# Distributed under the terms of the Apache 2.0 License.
import copy
import numpy as np
import datetime
import gym
from simple_model_functions import calc_arid, calc_yield, calc_ref_evapotranspiration, calc_transpiration, \
calc_f_temp, calc_f_co2, calc_f_heat, calc_f_water, calc_plant_available_water, calc_f_solar_water, calc_f_solar, \
calc_biomass_rate, calc_cumulative_biomass, calc_rad_50p_senescence, delta_cumulative_temp
class SimpleCropModelEnv(gym.Env):
def __init__(self, sowing_date,
num_growing_days,
weather_schedule,
weather_forecast_stds,
latitude,
elevation,
crop_parameters,
initial_biomass=0, # Kg
initial_cumulative_temp=0, # degrees C day
# initial_f_solar=0, # unit-less; todo: paper says this is an init param, but where would it be used?
initial_available_water_content=None, # not an initial parameter in SIMPLE, but reasonable to vary,
# https://acsess.onlinelibrary.wiley.com/doi/full/10.2134/agronj2011.0286 assumes initial value to
# be equal to the available water capacity (AWC) in their crop model comparison
seed=None,
cumulative_biomass_std=1.0,
plant_available_water_std=1.0):
# simulation parameters
self.num_growing_days = num_growing_days
# location parameters
self.latitude = latitude
self.elevation = elevation
# crop parameters
self.crop_temp_base = crop_parameters.temp_base
self.crop_temp_opt = crop_parameters.temp_opt
self.crop_RUE = crop_parameters.RUE
self.crop_rad_50p_growth = crop_parameters.rad_50p_growth
self.crop_rad_50p_senescence = crop_parameters.rad_50p_senescence
self.crop_maturity_temp = crop_parameters.maturity_temp
self.crop_rad_50p_max_heat = crop_parameters.rad_50p_max_heat
self.crop_rad_50p_max_water = crop_parameters.rad_50p_max_water
self.crop_heat_stress_thresh = crop_parameters.heat_stress_thresh
self.crop_heat_stress_extreme = crop_parameters.heat_stress_extreme
self.crop_drought_stress_sensitivity = crop_parameters.drought_stress_sensitivity
self.crop_deep_drainage_coef = crop_parameters.deep_drainage_coef
self.crop_water_holding_capacity = crop_parameters.water_holding_capacity
self.crop_runoff_curve_number = crop_parameters.runoff_curve_number
self.crop_root_zone_depth = crop_parameters.root_zone_depth
self.crop_co2_sensitivity = crop_parameters.co2_sensitivity
self.crop_harvest_index = crop_parameters.harvest_index
# variables for reset
self.init_cumulative_temp = initial_cumulative_temp
self.init_crop_rad_50p_senescence = crop_parameters.rad_50p_senescence
self.init_biomass = initial_biomass
self.sowing_date = sowing_date
# root zone available water (W_i) = root zone available water content (theta^{ad}_{a, i}) times the root
# zone depth (RZD), i.e. W_i = theta^{ad}_{a, i} * RZD. See:
# https://acsess.onlinelibrary.wiley.com/doi/full/10.2134/agronj2011.0286, which also justifies the initial
# condition shown here in the crop comparison section
if initial_available_water_content is None:
initial_available_water_content = self.crop_water_holding_capacity
self.initial_available_water_content = initial_available_water_content # unused, but saved for reference
self.initial_plant_available_water = initial_available_water_content * self.crop_root_zone_depth
# tracking variables
self.cumulative_mean_temp = None # TT variable in paper
self.cumulative_biomass = None
self.plant_available_water = None
self.date = None
self.day = None
# Randomization parameters
self.cumulative_biomass_sigma = cumulative_biomass_std
self.plant_available_water_sigma = plant_available_water_std
self.rng = np.random.default_rng(seed)
# weather data
self.weather_schedule = weather_schedule
self.weather_forecast_stds = weather_forecast_stds
# gym parameters
self.observation_space = gym.spaces.Box(-np.inf, np.inf, shape=(30,), dtype=np.float64)
self.action_space = gym.spaces.Box(0, 100, shape=(1,), dtype=np.float64)
self.reward_range = (-np.inf, np.inf)
self.metadata = {}
def weather_info(self, day, noisy=False):
"""return a weather info for a certain day"""
if day >= self.num_growing_days:
return np.array([0, 0, 0, 0, 0, 0, 0, 0])
if not noisy:
return np.array(
[self.weather_schedule.max_temp[day], # degrees C
self.weather_schedule.min_temp[day], # degrees C
self.weather_schedule.precipitation[day], # mm
self.weather_schedule.radiation[day], # MJ/(m**2 * day)
self.weather_schedule.co2[day], # PPM
self.weather_schedule.avg_vapor_pressure[day], # hPa
self.weather_schedule.mean_temp[day], # degrees C; not in SIMPLE model
self.weather_schedule.avg_wind[day] # m/s; not in SIMPLE model
]
)
else:
return np.array(
[self.rng.normal(self.weather_schedule.max_temp[day], self.weather_forecast_stds.max_temp),
self.rng.normal(self.weather_schedule.min_temp[day], self.weather_forecast_stds.min_temp),
max(0.0, self.rng.normal(self.weather_schedule.precipitation[day],
self.weather_forecast_stds.precipitation)),
max(0.0, self.rng.normal(self.weather_schedule.radiation[day], self.weather_forecast_stds.radiation)),
max(0.0, self.rng.normal(self.weather_schedule.co2[day], self.weather_forecast_stds.co2)),
max(0.0, self.rng.normal(self.weather_schedule.avg_vapor_pressure[day],
self.weather_forecast_stds.avg_vapor_pressure)),
self.rng.normal(self.weather_schedule.mean_temp[day], self.weather_forecast_stds.mean_temp),
max(0.0, self.rng.normal(self.weather_schedule.avg_wind[day], self.weather_forecast_stds.avg_wind))
]
)
def create_dict_state(self):
"""
Things to include in the state would be:
- Actual cumulative biomass
- Noisy reading of cumulative biomass (such as might be estimated from an image of the plant?)
- Cumulative mean temp (seems relatively measurable)
- Days since sowing date
- Weather from last day
- Next day's weather forecast (perfect)
- Noisy "prediction" of next day's weather forecast
- Root zone available water reading (like a soil measurement?)
- Noisy reading of root zone available water
"""
return {'cumulative_biomass': np.array(self.cumulative_biomass).reshape((1,)),
'cumulative_biomass_noisy':
np.array(max(0, self.rng.normal(self.cumulative_biomass,
self.cumulative_biomass_sigma))).reshape((1,)),
'cumulative_mean_temp': np.array(self.cumulative_mean_temp).reshape((1,)),
'day': np.array(self.day).reshape((1,)),
'weather_today': self.weather_info(self.day),
'weather_tomorrow': self.weather_info(self.day + 1),
'weather_tomorrow_forecast': self.weather_info(self.day + 1, noisy=True),
'plant_available_water': np.array(self.plant_available_water).reshape((1,)),
'plant_available_water_noisy':
np.array(max(0,
self.rng.normal(self.plant_available_water,
self.plant_available_water_sigma))).reshape((1,))
}
@staticmethod
def create_numpy_state(dict_state: dict):
return np.concatenate([v for v in dict_state.values()])
@staticmethod
def create_info(dict_state: dict, additional_info: dict):
info = copy.deepcopy(dict_state)
info.update(additional_info)
return info
def irrigation_reward(self, irrigation):
# assume irrigation is for a non-towable center pivot, and consider
# http://h2oinitiative.com/wp-content/uploads/2018/05/Estimating-Irrigation-Costs-Tacker-et-al.pdf
# even though it is from Arkansas...
# 9 inch irrigation season is $116/acre ~ $286.64/hectare assuming 2.471 acres per hectare
# 9 inches is 22.86 cm, or 0.2286 m, and 1 hectare = 10,000 m**2, so price per mm for 1 square meter plot is
# $286.64/(10,000 m**2 * 228.6 mm) = $0.000125 / mm / m**2
return -0.000125 * irrigation
def step(self, action):
rad_day = self.weather_schedule.radiation[self.day]
mean_temp_day = self.weather_schedule.mean_temp[self.day]
max_temp_day = self.weather_schedule.max_temp[self.day]
min_temp_day = self.weather_schedule.min_temp[self.day]
avg_vapor_pressure = self.weather_schedule.avg_vapor_pressure[self.day]
avg_wind = self.weather_schedule.avg_wind[self.day]
precipitation = self.weather_schedule.precipitation[self.day]
irrigation = action[0]
co2_day = self.weather_schedule.co2[self.day]
if irrigation < 0:
raise ValueError("Action cannot be negative! Got {}".format(irrigation))
self.cumulative_mean_temp += delta_cumulative_temp(mean_temp_day, self.crop_temp_base)
f_heat = calc_f_heat(max_temp_day, self.crop_heat_stress_thresh, self.crop_heat_stress_extreme)
ref_evapotranspiration = calc_ref_evapotranspiration(self.date, self.latitude, self.elevation, min_temp_day,
max_temp_day, rad_day, avg_vapor_pressure, avg_wind)
transpiration = calc_transpiration(ref_evapotranspiration, self.plant_available_water) # use PAW from today
# update PAW
self.plant_available_water = calc_plant_available_water(self.plant_available_water,
precipitation, irrigation,
transpiration,
self.crop_deep_drainage_coef,
self.crop_root_zone_depth,
self.crop_water_holding_capacity,
self.crop_runoff_curve_number)
arid_index = calc_arid(transpiration, ref_evapotranspiration)
f_water = calc_f_water(self.crop_drought_stress_sensitivity, arid_index)
f_solar, senescence = calc_f_solar(self.cumulative_mean_temp, self.crop_rad_50p_growth, self.crop_maturity_temp,
self.crop_rad_50p_senescence)
if senescence: # Note: Not sure if this is how to correctly designate growth vs senescence periods
self.crop_rad_50p_senescence = calc_rad_50p_senescence(self.crop_rad_50p_senescence,
self.crop_rad_50p_max_heat,
self.crop_rad_50p_max_water,
f_heat,
f_water)
f_co2 = calc_f_co2(self.crop_co2_sensitivity, co2_day)
f_temp = calc_f_temp(mean_temp_day, self.crop_temp_base, self.crop_temp_opt)
f_solar_water = calc_f_solar_water(f_water)
biomass_rate = calc_biomass_rate(rad_day, f_solar, f_solar_water, self.crop_RUE, f_co2, f_temp, f_heat, f_water)
self.cumulative_biomass = calc_cumulative_biomass(self.cumulative_biomass, biomass_rate)
# do this before updating day to get accurate states
dict_state = self.create_dict_state()
state = self.create_numpy_state(dict_state)
info = self.create_info(dict_state, {'arid_index': arid_index, 'f_solar': f_solar})
if self.day >= self.num_growing_days - 1:
reward = calc_yield(self.cumulative_biomass, self.crop_harvest_index) + self.irrigation_reward(irrigation)
return state, reward, True, info
self.day += 1
self.date += datetime.timedelta(days=1)
return state, self.irrigation_reward(irrigation), False, info
def reset(self):
self.cumulative_mean_temp = self.init_cumulative_temp # TT variable in paper
self.cumulative_biomass = self.init_biomass
self.crop_rad_50p_senescence = self.init_crop_rad_50p_senescence
# root zone available water (W_i) = root zone available water content (theta^{ad}_{a, i} or PAW) times the root
# zone depth (RZD)
self.plant_available_water = self.initial_plant_available_water
self.date = self.sowing_date
self.day = 0
return self.create_numpy_state(self.create_dict_state())
def render(self, mode='human'):
pass