1 | """Classes for implementing damage curves and economic damage |
---|
2 | |
---|
3 | Duncan Gray, Ole Nielsen, Jane Sexton, Nick Bartzis |
---|
4 | Geoscience Australia, 2006 |
---|
5 | """ |
---|
6 | |
---|
7 | from math import sqrt |
---|
8 | from Scientific.Functions.Interpolation import InterpolatingFunction |
---|
9 | from Numeric import array, ravel, Float, zeros |
---|
10 | from random import choice |
---|
11 | |
---|
12 | try: |
---|
13 | import kinds |
---|
14 | except ImportError: |
---|
15 | # Hand-built mockup of the things we need from the kinds package, since it |
---|
16 | # was recently removed from the standard Numeric distro. Some users may |
---|
17 | # not have it by default. |
---|
18 | class _bunch: |
---|
19 | pass |
---|
20 | |
---|
21 | class _kinds(_bunch): |
---|
22 | default_float_kind = _bunch() |
---|
23 | default_float_kind.MIN = 2.2250738585072014e-308 #smallest +ve number |
---|
24 | default_float_kind.MAX = 1.7976931348623157e+308 |
---|
25 | |
---|
26 | kinds = _kinds() |
---|
27 | |
---|
28 | |
---|
29 | from utilities.numerical_tools import ensure_numeric |
---|
30 | from pyvolution.data_manager import Exposure_csv |
---|
31 | from pyvolution.util import file_function |
---|
32 | from geospatial_data.geospatial_data import ensure_absolute |
---|
33 | from utilities.numerical_tools import INF |
---|
34 | from anuga_config import epsilon |
---|
35 | depth_epsilon = epsilon |
---|
36 | |
---|
37 | def inundation_damage(sww_file, exposure_file_in, |
---|
38 | exposure_file_out=None, |
---|
39 | ground_floor_height=0.3, |
---|
40 | overwrite=False, verbose=True, |
---|
41 | use_cache = True): |
---|
42 | """ |
---|
43 | This is the main function for calculating tsunami damage due to |
---|
44 | inundation. It gets the location of structures from the exposure |
---|
45 | file and gets the inundation of these structures from the |
---|
46 | sww file. |
---|
47 | |
---|
48 | It then calculates the damage loss. |
---|
49 | """ |
---|
50 | |
---|
51 | csv = Exposure_csv(exposure_file_in) |
---|
52 | geospatial = csv.get_location() |
---|
53 | max_depths, max_momentums = calc_max_depth_and_momentum(sww_file, |
---|
54 | geospatial, |
---|
55 | ground_floor_height=ground_floor_height, |
---|
56 | verbose=verbose, |
---|
57 | use_cache=use_cache) |
---|
58 | edm = EventDamageModel(max_depths, |
---|
59 | csv.get_column('SHORE_DIST'), |
---|
60 | csv.get_column('WALLS'), |
---|
61 | csv.get_column('STR_VALUE'), |
---|
62 | csv.get_column('C_VALUE') |
---|
63 | ) |
---|
64 | results_dic = edm.calc_damage_and_costs(verbose_csv=True, verbose=verbose) |
---|
65 | for title, value in results_dic.iteritems(): |
---|
66 | csv.set_column(title, value, overwrite=overwrite) |
---|
67 | |
---|
68 | # Save info back to csv file |
---|
69 | if exposure_file_out == None: |
---|
70 | exposure_file_out = exposure_file_in |
---|
71 | csv.save(exposure_file_out) |
---|
72 | |
---|
73 | def add_depth_and_momentum2csv(sww_file, exposure_file_in, |
---|
74 | exposure_file_out=None, |
---|
75 | overwrite=False, verbose=True, |
---|
76 | use_cache = True): |
---|
77 | """ |
---|
78 | """ |
---|
79 | |
---|
80 | csv = Exposure_csv(exposure_file_in) |
---|
81 | geospatial = csv.get_location() |
---|
82 | max_depths, max_momentums = calc_max_depth_and_momentum(sww_file, |
---|
83 | geospatial, |
---|
84 | verbose=verbose, |
---|
85 | use_cache=use_cache) |
---|
86 | csv.set_column("MAX INUNDATION DEPTH (m)",max_depths, overwrite=overwrite) |
---|
87 | csv.set_column("MOMENTUM (m^2/s) ",max_momentums, overwrite=overwrite) |
---|
88 | csv.save(exposure_file_out) |
---|
89 | |
---|
90 | def calc_max_depth_and_momentum(sww_file, points, |
---|
91 | ground_floor_height=0.0, |
---|
92 | verbose=True, |
---|
93 | use_cache = True): |
---|
94 | """ |
---|
95 | Calculate the maximum inundation height above ground floor (mihagf) for a list |
---|
96 | of locations. |
---|
97 | |
---|
98 | The mihagf value is in the range -ground_floor_height to overflow errors. |
---|
99 | """ |
---|
100 | quantities = ['stage', 'elevation', 'xmomentum', 'ymomentum'] |
---|
101 | points = ensure_absolute(points) |
---|
102 | point_count = len(points) |
---|
103 | callable_sww = file_function(sww_file, |
---|
104 | quantities=quantities, |
---|
105 | interpolation_points=points, |
---|
106 | verbose=verbose, |
---|
107 | use_cache=use_cache) |
---|
108 | # initialise the max lists |
---|
109 | max_depths = [-ground_floor_height]*point_count |
---|
110 | max_momentums = [-ground_floor_height]*point_count |
---|
111 | for point_i, point in enumerate(points): |
---|
112 | for time in callable_sww.get_time(): |
---|
113 | quantities = callable_sww(time,point_i) |
---|
114 | #print "quantities", quantities |
---|
115 | |
---|
116 | w = quantities[0] |
---|
117 | z = quantities[1] |
---|
118 | uh = quantities[2] |
---|
119 | vh = quantities[3] |
---|
120 | |
---|
121 | # -ground_floor_height is the minimum value. |
---|
122 | depth = w - z - ground_floor_height |
---|
123 | |
---|
124 | if depth > max_depths[point_i]: |
---|
125 | max_depths[point_i] = depth |
---|
126 | if w == INF or z == INF or uh == INF or vh == INF: |
---|
127 | continue |
---|
128 | momentum = sqrt(uh*uh + vh*vh) |
---|
129 | if momentum > max_momentums[point_i]: |
---|
130 | max_momentums[point_i] = momentum |
---|
131 | return max_depths, max_momentums |
---|
132 | |
---|
133 | class EventDamageModel: |
---|
134 | """ |
---|
135 | |
---|
136 | |
---|
137 | """ |
---|
138 | STRUCT_LOSS_TITLE = "Structure Loss ($)" |
---|
139 | CONTENTS_LOSS_TITLE = "Contents Loss ($)" |
---|
140 | CONTENTS_DAMAGE_TITLE = "Contents damaged (fraction)" |
---|
141 | STRUCT_DAMAGE_TITLE = "Structure damaged (fraction)" |
---|
142 | COLLAPSE_CSV_INFO_TITLE = "Calculation notes" |
---|
143 | MAX_DEPTH_TITLE = "Inundation height above ground floor (m)" |
---|
144 | STRUCT_COLLAPSED_TITLE = "collapsed structure if 1" |
---|
145 | STRUCT_INUNDATED_TITLE = "inundated structure if 1" |
---|
146 | double_brick_damage_array = array([[-kinds.default_float_kind.MAX, 0.0], |
---|
147 | [0.0-depth_epsilon, 0.0], |
---|
148 | [0.0,0.016], |
---|
149 | [0.1,0.150], |
---|
150 | [0.3,0.425], |
---|
151 | [0.5,0.449], |
---|
152 | [1.0,0.572], |
---|
153 | [1.5,0.582], |
---|
154 | [2.0,0.587], |
---|
155 | [2.5,0.647], |
---|
156 | [kinds.default_float_kind.MAX,64.7]]) |
---|
157 | double_brick_damage_curve = InterpolatingFunction( \ |
---|
158 | (ravel(double_brick_damage_array[:,0:1]),), |
---|
159 | ravel(double_brick_damage_array[:,1:])) |
---|
160 | |
---|
161 | brick_veeer_damage_array = array([[-kinds.default_float_kind.MAX, 0.0], |
---|
162 | [0.0-depth_epsilon, 0.0], |
---|
163 | [0.0,0.016], |
---|
164 | [0.1,0.169], |
---|
165 | [0.3,0.445], |
---|
166 | [0.5,0.472], |
---|
167 | [1.0,0.618], |
---|
168 | [1.5,0.629], |
---|
169 | [2.0,0.633], |
---|
170 | [2.5,0.694], |
---|
171 | [kinds.default_float_kind.MAX,69.4]]) |
---|
172 | brick_veeer_damage_curve = InterpolatingFunction( \ |
---|
173 | (ravel(brick_veeer_damage_array[:,0:1]),), |
---|
174 | ravel(brick_veeer_damage_array[:,1:])) |
---|
175 | struct_damage_curve = {'Double Brick':double_brick_damage_curve, |
---|
176 | 'Brick Veneer':brick_veeer_damage_curve} |
---|
177 | default_struct_damage_curve = brick_veeer_damage_curve |
---|
178 | |
---|
179 | contents_damage_array = array([[-kinds.default_float_kind.MAX, 0.0], |
---|
180 | [0.0-depth_epsilon, 0.0], |
---|
181 | [0.0,0.013], |
---|
182 | [0.1,0.102], |
---|
183 | [0.3,0.381], |
---|
184 | [0.5,0.500], |
---|
185 | [1.0,0.970], |
---|
186 | [1.5,0.976], |
---|
187 | [2.0,0.986], |
---|
188 | [kinds.default_float_kind.MAX,98.6]]) |
---|
189 | contents_damage_curve = InterpolatingFunction( \ |
---|
190 | (ravel(contents_damage_array[:,0:1]),), |
---|
191 | ravel(contents_damage_array[:,1:])) |
---|
192 | |
---|
193 | #building collapse probability |
---|
194 | # inundation depth above ground floor, m |
---|
195 | depth_upper_limits = [depth_epsilon, 1.0, 2.0, 3.0, 5.0, kinds.default_float_kind.MAX] |
---|
196 | # shore mistance, m |
---|
197 | shore_upper_limits = [125,200,250, kinds.default_float_kind.MAX] |
---|
198 | # Building collapse probability |
---|
199 | collapse_probability = [[0.0, 0.0, 0.0, 0.0], #Code below assumes 0.0 |
---|
200 | [0.05, 0.02, 0.01, 0.0], |
---|
201 | [0.6, 0.3, 0.1, 0.05], |
---|
202 | [0.8, 0.4, 0.25, 0.15], |
---|
203 | [0.95, 0.7, 0.5, 0.3], |
---|
204 | [0.99, 0.9, 0.65, 0.45]] |
---|
205 | |
---|
206 | def __init__(self,max_depths, shore_distances, walls, |
---|
207 | struct_costs, content_costs): |
---|
208 | """ |
---|
209 | max depth is Inundation height above ground floor (m), so |
---|
210 | the ground floor has been taken into account. |
---|
211 | """ |
---|
212 | self.max_depths = [float(x) for x in max_depths] |
---|
213 | self.shore_distances = [float(x) for x in shore_distances] |
---|
214 | self.walls = walls |
---|
215 | self.struct_costs = [float(x) for x in struct_costs] |
---|
216 | self.content_costs = [float(x) for x in content_costs] |
---|
217 | |
---|
218 | self.structure_count = len(self.max_depths) |
---|
219 | #Fixme expand |
---|
220 | assert self.structure_count == len(self.shore_distances) |
---|
221 | assert self.structure_count == len(self.walls) |
---|
222 | assert self.structure_count == len(self.struct_costs) |
---|
223 | assert self.structure_count == len(self.content_costs) |
---|
224 | #assert self.structure_count == len(self.) |
---|
225 | |
---|
226 | def calc_damage_and_costs(self, verbose_csv=False, verbose=False): |
---|
227 | self.calc_damage_percentages() |
---|
228 | collapse_probability = self.calc_collapse_probability() |
---|
229 | self._calc_collapse_structures(collapse_probability, |
---|
230 | verbose_csv=verbose_csv) |
---|
231 | self.calc_cost() |
---|
232 | results_dict = {self.STRUCT_LOSS_TITLE:self.struct_loss |
---|
233 | ,self.STRUCT_DAMAGE_TITLE:self.struct_damage |
---|
234 | ,self.CONTENTS_LOSS_TITLE:self.contents_loss |
---|
235 | ,self.CONTENTS_DAMAGE_TITLE:self.contents_damage |
---|
236 | ,self.MAX_DEPTH_TITLE:self.max_depths |
---|
237 | ,self.STRUCT_COLLAPSED_TITLE:self.struct_collapsed |
---|
238 | ,self.STRUCT_INUNDATED_TITLE:self.struct_inundated |
---|
239 | } |
---|
240 | if verbose_csv: |
---|
241 | results_dict[self.COLLAPSE_CSV_INFO_TITLE] = self.collapse_csv_info |
---|
242 | return results_dict |
---|
243 | |
---|
244 | def calc_damage_percentages(self): |
---|
245 | |
---|
246 | # the data being created |
---|
247 | struct_damage = zeros(self.structure_count,Float) |
---|
248 | contents_damage = zeros(self.structure_count,Float) |
---|
249 | self.struct_inundated = ['']* self.structure_count |
---|
250 | |
---|
251 | for i,max_depth,shore_distance,wall in map(None, |
---|
252 | range(self.structure_count), |
---|
253 | self.max_depths, |
---|
254 | self.shore_distances, |
---|
255 | self.walls): |
---|
256 | #print "i",i |
---|
257 | #print "max_depth",max_depth |
---|
258 | #print "shore_distance",shore_distance |
---|
259 | #print "wall",wall |
---|
260 | ## WARNING SKIP IF DEPTH < 0.0 |
---|
261 | if 0.0 > max_depth: |
---|
262 | continue |
---|
263 | |
---|
264 | # The definition of inundated is if the max_depth is > 0.0 |
---|
265 | self.struct_inundated[i] = 1.0 |
---|
266 | |
---|
267 | #calc structural damage % |
---|
268 | damage_curve = self.struct_damage_curve.get(wall, |
---|
269 | self.default_struct_damage_curve) |
---|
270 | struct_damage[i] = damage_curve(max_depth) |
---|
271 | contents_damage[i] = self.contents_damage_curve(max_depth) |
---|
272 | |
---|
273 | self.struct_damage = struct_damage |
---|
274 | self.contents_damage = contents_damage |
---|
275 | |
---|
276 | |
---|
277 | def calc_cost(self): |
---|
278 | # ensure_numeric does not cut it. |
---|
279 | self.struct_loss = self.struct_damage * \ |
---|
280 | ensure_numeric(self.struct_costs) |
---|
281 | self.contents_loss = self.contents_damage * \ |
---|
282 | ensure_numeric(self.content_costs) |
---|
283 | |
---|
284 | def calc_collapse_probability(self): |
---|
285 | """ |
---|
286 | return a dict of which structures have x probability of collapse. |
---|
287 | key is collapse probability |
---|
288 | value is list of struct indexes with key probability of collapse |
---|
289 | """ |
---|
290 | # I could've done this is the calc_damage_percentages and |
---|
291 | # Just had one loop. |
---|
292 | # But for ease of testing and bug finding I'm seperating the loops. |
---|
293 | # I'm make the outer loop for both of them the same though, |
---|
294 | # so this loop can easily be folded into the other loop. |
---|
295 | |
---|
296 | # dict of which structures have x probability of collapse. |
---|
297 | # key of collapse probability |
---|
298 | # value of list of struct indexes |
---|
299 | struct_coll_prob = {} |
---|
300 | |
---|
301 | for i,max_depth,shore_distance,wall in map(None, |
---|
302 | range(self.structure_count), |
---|
303 | self.max_depths, |
---|
304 | self.shore_distances, |
---|
305 | self.walls): |
---|
306 | #print "i",i |
---|
307 | #print "max_depth",max_depth |
---|
308 | #print "shore_distance",shore_distance |
---|
309 | #print "wall",wall |
---|
310 | # WARNING ASSUMING THE FIRST BIN OF DEPTHS GIVE A ZERO PROBABILITY |
---|
311 | depth_upper_limits = self.depth_upper_limits |
---|
312 | shore_upper_limits = self.shore_upper_limits |
---|
313 | collapse_probability = self.collapse_probability |
---|
314 | if max_depth <= depth_upper_limits[0]: |
---|
315 | continue |
---|
316 | start = 1 |
---|
317 | for i_depth, depth_limit in enumerate(depth_upper_limits[start:]): |
---|
318 | #Have to change i_depth so it indexes into the lists correctly |
---|
319 | i_depth += start |
---|
320 | if max_depth <= depth_limit: |
---|
321 | for i_shore, shore_limit in enumerate(shore_upper_limits): |
---|
322 | if shore_distance <= shore_limit: |
---|
323 | coll_prob = collapse_probability[i_depth][i_shore] |
---|
324 | if 0.0 == collapse_probability[i_depth][i_shore]: |
---|
325 | break |
---|
326 | struct_coll_prob.setdefault(coll_prob,[]).append(i) |
---|
327 | break |
---|
328 | break |
---|
329 | |
---|
330 | return struct_coll_prob |
---|
331 | |
---|
332 | def _calc_collapse_structures(self, collapse_probability, |
---|
333 | verbose_csv=False): |
---|
334 | """ |
---|
335 | Given the collapse probabilities, throw the dice |
---|
336 | and collapse some houses |
---|
337 | """ |
---|
338 | |
---|
339 | self.struct_collapsed = ['']* self.structure_count |
---|
340 | if verbose_csv: |
---|
341 | self.collapse_csv_info = ['']* self.structure_count |
---|
342 | #for a given 'bin', work out how many houses will collapse |
---|
343 | for probability, house_indexes in collapse_probability.iteritems(): |
---|
344 | collapse_count = round(len(house_indexes) *probability) |
---|
345 | |
---|
346 | if verbose_csv: |
---|
347 | for i in house_indexes: |
---|
348 | # This could be sped up I think |
---|
349 | self.collapse_csv_info[i] = str(probability) + ' prob.( ' \ |
---|
350 | + str(int(collapse_count)) + ' collapsed out of ' \ |
---|
351 | + str(len(house_indexes)) + ')' |
---|
352 | for _ in range(int(collapse_count)): |
---|
353 | house_index = choice(house_indexes) |
---|
354 | self.struct_damage[house_index] = 1.0 |
---|
355 | self.contents_damage[house_index] = 1.0 |
---|
356 | house_indexes.remove(house_index) |
---|
357 | self.struct_collapsed[house_index] = 1 |
---|
358 | |
---|
359 | # Warning, the collapse_probability list now lists |
---|
360 | # houses that did not collapse, (though not all of them) |
---|
361 | #print "",self.collapse_csv_info |
---|
362 | ############################################################################# |
---|
363 | if __name__ == "__main__": |
---|
364 | from Scientific.Functions.Interpolation import InterpolatingFunction |
---|
365 | from Numeric import array, ravel |
---|
366 | a = array([[0,0],[1,10]]) |
---|
367 | c = array([0,1]) |
---|
368 | |
---|
369 | axis = ravel(a[:,0:1]) |
---|
370 | values = ravel(a[:,1:]) |
---|
371 | |
---|
372 | #axis = array([0,1]) |
---|
373 | #values = array([0,10]) |
---|
374 | print "axis",axis |
---|
375 | print "values",values |
---|
376 | i = InterpolatingFunction((axis,), values) |
---|
377 | print "value",i(0.5) |
---|
378 | |
---|
379 | |
---|
380 | i = InterpolatingFunction((array([0,1]),),array([0,10])) |
---|
381 | print "value",i(0.5) |
---|
382 | |
---|
383 | dict = {1:10, 2:20} |
---|
384 | print dict.get(100,dict[2]) |
---|
385 | |
---|
386 | inundation_damage('source.sww', 'augmented_buildings_high.csv', |
---|
387 | 'augmented_buildings_high_dsg.csv') |
---|