13 | | Available options: |
14 | | |
15 | | Sub-types (Parameter subtype) |
16 | | 0: Evolve unit modes with finite difference approx. of TLM |
17 | | 1: like 0 with modes scaled by eigenvalues, unit U |
18 | | 2: Fixed basis vectors; variable U matrix |
19 | | 3: Fixed covariance matrix (V and U kept constant) |
20 | | 5: Offline mode |
21 | | |
22 | | Integer parameters (Array param_int) |
23 | | param_int(1): Dimension of state vector (>0), required |
24 | | param_int(2): Ensemble size (>0), required |
25 | | param_int(3): Interval for re-diagonalization of P (>0); optional: default 1 |
26 | | param_int(4): 1 for incremental updating, 0 else; optional: default 0 |
27 | | |
28 | | Floating point parameters (Array param_real) |
29 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
30 | | param_real(2): epsilon for finite-difference approx. of TLM, required |
31 | | |
32 | | Further parameters |
33 | | n_modeltasks: Number of parallel model integration tasks |
34 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
35 | | =1 required for subtypes 2 and 3 |
36 | | screen: Control verbosity of PDAF |
37 | | 0: no outputs |
38 | | 1: basic output (default) |
39 | | 2: 1 plus timing output |
40 | | 3: 2 plus debug output |
41 | | |
42 | | +++++++++ End of option overview for the SEEK filter ++++++++++ |
| 13 | PDAF Available options for SEEK: |
| 14 | PDAF --- Sub-types (Parameter subtype) --- |
| 15 | PDAF 0: Evolve unit modes with finite difference approx. of TLM |
| 16 | PDAF 1: like 0 with modes scaled by eigenvalues, unit U |
| 17 | PDAF 2: Fixed basis vectors; variable U matrix |
| 18 | PDAF 3: Fixed covariance matrix (V and U kept constant) |
| 19 | PDAF 5: Offline mode |
| 20 | PDAF --- Integer parameters (Array param_int) --- |
| 21 | PDAF param_int(1): Dimension of state vector (>0), required |
| 22 | PDAF param_int(2): Ensemble size (>0), required |
| 23 | PDAF param_int(3): Interval for re-diagonalization of P (>0); optional: default 1 |
| 24 | PDAF param_int(4): 1 for incremental updating, 0 else; optional: default 0 |
| 25 | PDAF --- Floating point parameters (Array param_real) --- |
| 26 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 27 | PDAF param_real(2): epsilon for finite-difference approx. of TLM, required |
| 28 | PDAF --- Further parameters --- |
| 29 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 30 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 31 | PDAF =1 required for subtypes 2 and 3 |
| 32 | PDAF screen: Control verbosity of PDAF |
| 33 | PDAF 0: no outputs |
| 34 | PDAF 1: basic output (default) |
| 35 | PDAF 2: 1 plus timing output |
| 36 | PDAF 3: 2 plus debug output |
| 37 | PDAF +++++++++ End of option overview for the SEEK filter ++++++++++ |
49 | | Available options: |
50 | | |
51 | | Sub-types (Parameter subtype) |
52 | | 0: full ensemble integration; left-sided application of T |
53 | | 1: full ensemble integration; right-sided application of T |
54 | | 2: Fixed error space basis |
55 | | 3: Fixed state covariance matrix |
56 | | 4: Implementation with explicit ensemble transformation |
57 | | 5: Offline mode |
58 | | |
59 | | Integer parameters (Array param_int) |
60 | | param_int(1): Dimension of state vector (>0), required |
61 | | param_int(2): Ensemble size (>0), required |
62 | | param_int(3): not used |
63 | | param_int(4): 1 for incremental updating, 0 else; optional, default: 0 |
64 | | param_int(5): Type of forgetting factor; optional, default: 0 |
65 | | 0: fixed forgetting factor |
66 | | 1: adaptive forgetting factor (experimental) |
67 | | param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
68 | | 0: deterministic omega |
69 | | 1: random orthonormal omega orthogonal to (1,...,1)^T |
70 | | 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
71 | | (experimental; for random transformations, 1 is recommended) |
72 | | param_int(7): Type of transformation matrix square root; optional, default: 0 |
73 | | (Only relevant for subtype=4) |
74 | | 0: symmetric square root |
75 | | 1: Cholesky decomposition |
76 | | |
77 | | Floating point parameters (Array param_real) |
78 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
79 | | |
80 | | Further parameters |
81 | | n_modeltasks: Number of parallel model integration tasks |
82 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
83 | | =1 required for subtypes 2 and 3 |
84 | | screen: Control verbosity of PDAF |
85 | | 0: no outputs |
86 | | 1: basic output (default) |
87 | | 2: 1 plus timing output |
88 | | 3: 2 plus debug output |
89 | | |
90 | | Internal parameter (defined inside PDAF) |
91 | | Nm1vsN: Normalization of covariance matrix; default: 1 |
92 | | 0: normalization with 1/(Ensemble size) |
93 | | (original SEIK, mainly for compatibility with older studies) |
94 | | 1: normalization with 1/(Ensemble size - 1) |
95 | | (sample covariance matrix consistent with other EnKFs) |
96 | | |
97 | | +++++++++ End of option overview for the SEIK filter ++++++++++ |
| 44 | PDAF Available options for SEIK: |
| 45 | PDAF --- Sub-types (Parameter subtype) --- |
| 46 | PDAF 0: full ensemble integration; left-sided application of T |
| 47 | PDAF 1: full ensemble integration; right-sided application of T |
| 48 | PDAF 2: Fixed error space basis |
| 49 | PDAF 3: Fixed state covariance matrix |
| 50 | PDAF 4: Implementation with explicit ensemble transformation |
| 51 | PDAF 5: Offline mode |
| 52 | PDAF --- Integer parameters (Array param_int) --- |
| 53 | PDAF param_int(1): Dimension of state vector (>0), required |
| 54 | PDAF param_int(2): Ensemble size (>0), required |
| 55 | PDAF param_int(3): not used |
| 56 | PDAF param_int(4): 1 for incremental updating, 0 else; optional, default: 0 |
| 57 | PDAF param_int(5): Type of forgetting factor; optional, default: 0 |
| 58 | PDAF 0: fixed forgetting factor |
| 59 | PDAF 1: adaptive forgetting factor (experimental) |
| 60 | PDAF param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
| 61 | PDAF 0: deterministic omega |
| 62 | PDAF 1: random orthonormal omega orthogonal to (1,...,1)^T |
| 63 | PDAF 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
| 64 | PDAF (experimental; for random transformations, 1 is recommended) |
| 65 | PDAF param_int(7): Type of transformation matrix square root; optional, default: 0 |
| 66 | PDAF (Only relevant for subtype/=3) |
| 67 | PDAF 0: symmetric square root |
| 68 | PDAF 1: Cholesky decomposition |
| 69 | PDAF --- Floating point parameters (Array param_real) --- |
| 70 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 71 | PDAF --- Further parameters --- |
| 72 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 73 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 74 | PDAF =1 required for subtypes 2 and 3 |
| 75 | PDAF screen: Control verbosity of PDAF |
| 76 | PDAF 0: no outputs |
| 77 | PDAF 1: basic output (default) |
| 78 | PDAF 2: 1 plus timing output |
| 79 | PDAF 3: 2 plus debug output |
| 80 | PDAF --- Internal parameter (defined inside PDAF) --- |
| 81 | PDAF Nm1vsN: Normalization of covariance matrix; default: 1 |
| 82 | PDAF 0: normalization with 1/(Ensemble size) |
| 83 | PDAF (original SEIK, mainly for compatibility with older studies) |
| 84 | PDAF 1: normalization with 1/(Ensemble size - 1) |
| 85 | PDAF (sample covariance matrix consistent with other EnKFs) |
| 86 | PDAF +++++++++ End of option overview for the SEIK filter ++++++++++ |
104 | | Available options: |
105 | | |
106 | | Sub-types (Parameter subtype) |
107 | | 0: Full ensemble integration; analysis for 2*dim_obs>dim_ens |
108 | | 1: Full ensemble integration; analysis for 2*dim_obs<=dim_ens |
109 | | 5: Offline mode |
110 | | |
111 | | Integer parameters (Array param_int) |
112 | | param_int(1): Dimension of state vector (>0), required |
113 | | param_int(2): Ensemble size (>0), required |
114 | | param_int(3): maximum rank for inversion of HPH^T, optional, default=0 |
115 | | (for =0, HPH is inverted by solving the representer equation) |
116 | | (if set to >=ensemble size, it is reset to ensemble size - 1) |
117 | | param_int(4): not used |
118 | | param_int(5): Size of lag for smoothing |
119 | | |
120 | | Floating point parameters (Array param_real) |
121 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
122 | | |
123 | | Further parameters |
124 | | n_modeltasks: Number of parallel model integration tasks |
125 | | (>=1; not larger than total number of processors) |
126 | | screen: Control verbosity of PDAF |
127 | | 0: no outputs |
128 | | 1: basic output (default) |
129 | | 2: 1 plus timing output |
130 | | 3: 2 plus debug output |
131 | | |
132 | | +++++++++ End of option overview for the EnKF ++++++++++ |
| 93 | PDAF Available options for EnKF: |
| 94 | PDAF --- Sub-types (Parameter subtype) --- |
| 95 | PDAF 0: Full ensemble integration; analysis for 2*dim_obs>dim_ens |
| 96 | PDAF 1: Full ensemble integration; analysis for 2*dim_obs<=dim_ens |
| 97 | PDAF 5: Offline mode |
| 98 | PDAF --- Integer parameters (Array param_int) --- |
| 99 | PDAF param_int(1): Dimension of state vector (>0), required |
| 100 | PDAF param_int(2): Ensemble size (>0), required |
| 101 | PDAF param_int(3): maximum rank for inversion of HPH^T, optional, default=0 |
| 102 | PDAF (for =0, HPH is inverted by solving the representer equation) |
| 103 | PDAF (if set to >=ensemble size, it is reset to ensemble size - 1) |
| 104 | PDAF param_int(4): not used |
| 105 | PDAF param_int(5): Size of lag for smoothing |
| 106 | PDAF --- Floating point parameters (Array param_real) --- |
| 107 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 108 | PDAF --- Further parameters --- |
| 109 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 110 | PDAF (>=1; not larger than total number of processors) |
| 111 | PDAF screen: Control verbosity of PDAF |
| 112 | PDAF 0: no outputs |
| 113 | PDAF 1: basic output (default) |
| 114 | PDAF 2: 1 plus timing output |
| 115 | PDAF 3: 2 plus debug output |
| 116 | PDAF +++++++++ End of option overview for the EnKF ++++++++++ |
139 | | Available options: |
140 | | |
141 | | Sub-types (Parameter subtype) |
142 | | 0: full ensemble integration; left-sided application of T |
143 | | 2: Fixed error space basis |
144 | | 3: Fixed state covariance matrix |
145 | | 5: Offline mode |
146 | | |
147 | | Integer parameters (Array param_int) |
148 | | param_int(1): Dimension of state vector (>0), required |
149 | | param_int(2): Ensemble size (>0), required |
150 | | param_int(3): not used |
151 | | param_int(4): 1 for incremental updating, 0 else; optional, default: 0 |
152 | | param_int(5): Type of forgetting factor; optional, default: 0 |
153 | | 0: fixed forgetting factor |
154 | | 1: adaptive forgetting factor for full domain (experimental) |
155 | | 2: locally adaptive forgetting factor (experimental) |
156 | | param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
157 | | 0: deterministic omega |
158 | | 1: random orthonormal omega orthogonal to (1,...,1)^T |
159 | | 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
160 | | (experimental; for random transformations, 1 is recommended) |
161 | | param_int(7): Type of transformation matrix square root; optional, default: 0 |
162 | | (Only relevant for subtype=4) |
163 | | 0: symmetric square root |
164 | | 1: Cholesky decomposition |
165 | | |
166 | | Floating point parameters (Array param_real) |
167 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
168 | | |
169 | | Further parameters |
170 | | n_modeltasks: Number of parallel model integration tasks |
171 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
172 | | =1 required for subtypes 2 and 3 |
173 | | screen: Control verbosity of PDAF |
174 | | 0: no outputs |
175 | | 1: basic output (default) |
176 | | 2: 1 plus timing output |
177 | | 3: 2 plus debug output |
178 | | |
179 | | Internal parameter (defined inside PDAF) |
180 | | Nm1vsN: Normalization of covariance matrix; default: 1 |
181 | | 0: normalization with 1/(Ensemble size) |
182 | | (original SEIK, mainly for compatibility with older studies) |
183 | | 1: normalization with 1/(Ensemble size - 1) |
184 | | (sample covariance matrix consistent with other EnKFs) |
185 | | |
186 | | +++++++++ End of option overview for the LSEIK filter ++++++++++ |
| 123 | PDAF Available options for LSEIK: |
| 124 | PDAF --- Sub-types (Parameter subtype) --- |
| 125 | PDAF 0: full ensemble integration; left-sided application of T |
| 126 | PDAF 2: Fixed error space basis |
| 127 | PDAF 3: Fixed state covariance matrix |
| 128 | PDAF 5: Offline mode |
| 129 | PDAF --- Integer parameters (Array param_int) --- |
| 130 | PDAF param_int(1): Dimension of state vector (>0), required |
| 131 | PDAF param_int(2): Ensemble size (>0), required |
| 132 | PDAF param_int(3): not used |
| 133 | PDAF param_int(4): 1 for incremental updating, 0 else; optional, default: 0 |
| 134 | PDAF param_int(5): Type of forgetting factor; optional, default: 0 |
| 135 | PDAF 0: fixed forgetting factor |
| 136 | PDAF 1: adaptive forgetting factor for full domain (experimental) |
| 137 | PDAF 2: locally adaptive forgetting factor (experimental) |
| 138 | PDAF param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
| 139 | PDAF 0: deterministic omega |
| 140 | PDAF 1: random orthonormal omega orthogonal to (1,...,1)^T |
| 141 | PDAF 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
| 142 | PDAF (experimental; for random transformations, 1 is recommended) |
| 143 | PDAF param_int(7): Type of transformation matrix square root; optional, default: 0 |
| 144 | PDAF (Only relevant for subtype/=3) |
| 145 | PDAF 0: symmetric square root |
| 146 | PDAF 1: Cholesky decomposition |
| 147 | PDAF --- Floating point parameters (Array param_real) --- |
| 148 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 149 | PDAF --- Further parameters --- |
| 150 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 151 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 152 | PDAF =1 required for subtypes 2 and 3 |
| 153 | PDAF screen: Control verbosity of PDAF |
| 154 | PDAF 0: no outputs |
| 155 | PDAF 1: basic output (default) |
| 156 | PDAF 2: 1 plus timing output |
| 157 | PDAF 3: 2 plus debug output |
| 158 | PDAF --- Internal parameter (defined inside PDAF) --- |
| 159 | PDAF Nm1vsN: Normalization of covariance matrix; default: 1 |
| 160 | PDAF 0: normalization with 1/(Ensemble size) |
| 161 | PDAF (original SEIK, mainly for compatibility with older studies) |
| 162 | PDAF 1: normalization with 1/(Ensemble size - 1) |
| 163 | PDAF (sample covariance matrix consistent with other EnKFs) |
| 164 | PDAF +++++++++ End of option overview for the LSEIK filter ++++++++++ |
193 | | Available options: |
194 | | |
195 | | Sub-types (Parameter subtype) |
196 | | 0: full ensemble integration; apply T-matrix analogously to SEIK |
197 | | 1: full ensemble integration; formulation without T matrix |
198 | | 2: Fixed error space basis; analysis with T-matrix |
199 | | 3: Fixed state covariance matrix; analysis with T-matrix |
200 | | 5: Offline mode; analysis with T-matrix |
201 | | |
202 | | Integer parameters (Array param_int) |
203 | | param_int(1): Dimension of state vector (>0), required |
204 | | param_int(2): Ensemble size (>0), required |
205 | | param_int(3): Size of lag for smoothing |
206 | | param_int(4): not used |
207 | | param_int(5): Type of forgetting factor; optional, default: 0 |
208 | | 0: fixed forgetting factor |
209 | | 1: adaptive forgetting factor (experimental) |
210 | | param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
211 | | 0: deterministic transformation |
212 | | 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
213 | | |
214 | | Floating point parameters (Array param_real) |
215 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
216 | | |
217 | | Further parameters |
218 | | n_modeltasks: Number of parallel model integration tasks |
219 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
220 | | =1 required for subtypes 2 and 3 |
221 | | screen: Control verbosity of PDAF |
222 | | 0: no outputs |
223 | | 1: basic output (default) |
224 | | 2: 1 plus timing output |
225 | | 3: 2 plus debug output |
226 | | |
227 | | +++++++++ End of option overview for the ETKF ++++++++++ |
| 171 | PDAF Available options for ETKF: |
| 172 | PDAF --- Sub-types (Parameter subtype) --- |
| 173 | PDAF 0: full ensemble integration; apply T-matrix analogously to SEIK |
| 174 | PDAF 1: full ensemble integration; formulation without T matrix |
| 175 | PDAF 2: Fixed error space basis; analysis with T-matrix |
| 176 | PDAF 3: Fixed state covariance matrix; analysis with T-matrix |
| 177 | PDAF 5: Offline mode; analysis with T-matrix |
| 178 | PDAF --- Integer parameters (Array param_int) --- |
| 179 | PDAF param_int(1): Dimension of state vector (>0), required |
| 180 | PDAF param_int(2): Ensemble size (>0), required |
| 181 | PDAF param_int(3): Size of lag for smoothing |
| 182 | PDAF param_int(4): not used |
| 183 | PDAF param_int(5): Type of forgetting factor; optional, default: 0 |
| 184 | PDAF 0: fixed forgetting factor |
| 185 | PDAF 1: adaptive forgetting factor (experimental) |
| 186 | PDAF param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
| 187 | PDAF 0: deterministic transformation |
| 188 | PDAF 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
| 189 | PDAF --- Floating point parameters (Array param_real) --- |
| 190 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 191 | PDAF --- Further parameters --- |
| 192 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 193 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 194 | PDAF =1 required for subtypes 2 and 3 |
| 195 | PDAF screen: Control verbosity of PDAF |
| 196 | PDAF 0: no outputs |
| 197 | PDAF 1: basic output (default) |
| 198 | PDAF 2: 1 plus timing output |
| 199 | PDAF 3: 2 plus debug output |
| 200 | PDAF +++++++++ End of option overview for the ETKF ++++++++++ |
234 | | Available options: |
235 | | |
236 | | Sub-types (Parameter subtype) |
237 | | 0: full ensemble integration; apply T-matrix analogously to SEIK |
238 | | 2: Fixed error space basis; analysis with T-matrix |
239 | | 3: Fixed state covariance matrix; analysis with T-matrix |
240 | | 5: Offline mode; analysis with T-matrix |
241 | | |
242 | | Integer parameters (Array param_int) |
243 | | param_int(1): Dimension of state vector (>0), required |
244 | | param_int(2): Ensemble size (>0), required |
245 | | param_int(3): Size of lag for smoothing |
246 | | param_int(4): not used |
247 | | param_int(5): Type of forgetting factor; optional, default: 0 |
248 | | 0: fixed forgetting factor |
249 | | 1: adaptive forgetting factor for full domain (experimental) |
250 | | 2: locally adaptive forgetting factor (experimental) |
251 | | param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
252 | | 0: deterministic transformation |
253 | | 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
254 | | |
255 | | Floating point parameters (Array param_real) |
256 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
257 | | |
258 | | Further parameters |
259 | | n_modeltasks: Number of parallel model integration tasks |
260 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
261 | | =1 required for subtypes 2 and 3 |
262 | | screen: Control verbosity of PDAF |
263 | | 0: no outputs |
264 | | 1: basic output (default) |
265 | | 2: 1 plus timing output |
266 | | 3: 2 plus debug output |
267 | | |
268 | | +++++++++ End of option overview for the LETKF ++++++++++ |
| 207 | PDAF Available options for LETKF: |
| 208 | PDAF --- Sub-types (Parameter subtype) --- |
| 209 | PDAF 0: full ensemble integration; apply T-matrix analogously to SEIK |
| 210 | PDAF 2: Fixed error space basis; analysis with T-matrix |
| 211 | PDAF 3: Fixed state covariance matrix; analysis with T-matrix |
| 212 | PDAF 5: Offline mode; analysis with T-matrix |
| 213 | PDAF --- Integer parameters (Array param_int) --- |
| 214 | PDAF param_int(1): Dimension of state vector (>0), required |
| 215 | PDAF param_int(2): Ensemble size (>0), required |
| 216 | PDAF param_int(3): Size of lag for smoothing |
| 217 | PDAF param_int(4): not used |
| 218 | PDAF param_int(5): Type of forgetting factor; optional, default: 0 |
| 219 | PDAF 0: fixed forgetting factor |
| 220 | PDAF 1: adaptive forgetting factor for full domain (experimental) |
| 221 | PDAF 2: locally adaptive forgetting factor (experimental) |
| 222 | PDAF param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
| 223 | PDAF 0: deterministic transformation |
| 224 | PDAF 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
| 225 | PDAF --- Floating point parameters (Array param_real) --- |
| 226 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 227 | PDAF --- Further parameters --- |
| 228 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 229 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 230 | PDAF =1 required for subtypes 2 and 3 |
| 231 | PDAF screen: Control verbosity of PDAF |
| 232 | PDAF 0: no outputs |
| 233 | PDAF 1: basic output (default) |
| 234 | PDAF 2: 1 plus timing output |
| 235 | PDAF 3: 2 plus debug output |
| 236 | PDAF +++++++++ End of option overview for the LETKF ++++++++++ |
275 | | Available options: |
276 | | |
277 | | Sub-types (Parameter subtype) |
278 | | 0: Standard implementation with ensemble integration |
279 | | 2: Fixed error space basis |
280 | | 3: Fixed state covariance matrix |
281 | | 5: Offline mode |
282 | | |
283 | | Integer parameters (Array param_int) |
284 | | param_int(1): Dimension of state vector (>0), required |
285 | | param_int(2): Ensemble size (>0), required |
286 | | param_int(3): Size of lag for smoothing |
287 | | param_int(4): not used |
288 | | param_int(5): Type of forgetting factor; optional, default: 0 |
289 | | 0: fixed forgetting factor |
290 | | 1: adaptive forgetting factor (experimental) |
291 | | param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
292 | | 0: deterministic omega |
293 | | 1: random orthonormal omega orthogonal to (1,...,1)^T |
294 | | 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
295 | | (experimental; for random transformations, 0 or 1 are recommended) |
296 | | param_int(7): Type of transformation matrix square root; optional, default: 0 |
297 | | 0: symmetric square root |
298 | | 1: Cholesky decomposition |
299 | | |
300 | | Floating point parameters (Array param_real) |
301 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
302 | | |
303 | | Further parameters |
304 | | n_modeltasks: Number of parallel model integration tasks |
305 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
306 | | =1 required for subtypes 2 and 3 |
307 | | screen: Control verbosity of PDAF |
308 | | 0: no outputs |
309 | | 1: basic output (default) |
310 | | 2: 1 plus timing output |
311 | | 3: 2 plus debug output |
312 | | |
313 | | +++++++++ End of option overview for the ESTKF ++++++++++ |
| 243 | PDAF Available options for ESTKF: |
| 244 | PDAF --- Sub-types (Parameter subtype) --- |
| 245 | PDAF 0: Standard implementation with ensemble integration |
| 246 | PDAF 2: Fixed error space basis |
| 247 | PDAF 3: Fixed state covariance matrix |
| 248 | PDAF 5: Offline mode |
| 249 | PDAF --- Integer parameters (Array param_int) --- |
| 250 | PDAF param_int(1): Dimension of state vector (>0), required |
| 251 | PDAF param_int(2): Ensemble size (>0), required |
| 252 | PDAF param_int(3): Size of lag for smoothing |
| 253 | PDAF param_int(4): not used |
| 254 | PDAF param_int(5): Type of forgetting factor; optional, default: 0 |
| 255 | PDAF 0: fixed forgetting factor |
| 256 | PDAF 1: adaptive forgetting factor (experimental) |
| 257 | PDAF param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
| 258 | PDAF 0: deterministic omega |
| 259 | PDAF 1: random orthonormal omega orthogonal to (1,...,1)^T |
| 260 | PDAF 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
| 261 | PDAF (experimental; for random transformations, 0 or 1 are recommended) |
| 262 | PDAF param_int(7): Type of transformation matrix square root; optional, default: 0 |
| 263 | PDAF 0: symmetric square root |
| 264 | PDAF 1: Cholesky decomposition |
| 265 | PDAF --- Floating point parameters (Array param_real) --- |
| 266 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 267 | PDAF --- Further parameters --- |
| 268 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 269 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 270 | PDAF =1 required for subtypes 2 and 3 |
| 271 | PDAF screen: Control verbosity of PDAF |
| 272 | PDAF 0: no outputs |
| 273 | PDAF 1: basic output (default) |
| 274 | PDAF 2: 1 plus timing output |
| 275 | PDAF 3: 2 plus debug output |
| 276 | PDAF +++++++++ End of option overview for the ESTKF ++++++++++ |
320 | | Available options: |
321 | | |
322 | | Sub-types (Parameter subtype) |
323 | | 0: Standard implementation with ensemble integration |
324 | | 2: Fixed error space basis |
325 | | 3: Fixed state covariance matrix |
326 | | 5: Offline mode |
327 | | |
328 | | Integer parameters (Array param_int) |
329 | | param_int(1): Dimension of state vector (>0), required |
330 | | param_int(2): Ensemble size (>0), required |
331 | | param_int(3): Size of lag for smoothing |
332 | | param_int(4): not used |
333 | | param_int(5): Type of forgetting factor; optional, default: 0 |
334 | | 0: fixed forgetting factor |
335 | | 1: adaptive forgetting factor for full domain (experimental) |
336 | | 2: locally adaptive forgetting factor (experimental) |
337 | | param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
338 | | 0: deterministic omega |
339 | | 1: random orthonormal omega orthogonal to (1,...,1)^T |
340 | | 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
341 | | (experimental; for random transformations, 1 is recommended) |
342 | | param_int(7): Type of transformation matrix square root; optional, default: 0 |
343 | | 0: symmetric square root |
344 | | 1: Cholesky decomposition |
345 | | |
346 | | Floating point parameters (Array param_real) |
347 | | param_real(1): Forgetting factor (usually >0 and <=1), required |
348 | | |
349 | | Further parameters |
350 | | n_modeltasks: Number of parallel model integration tasks |
351 | | >=1 for subtypes 0 and 1; not larger than total number of processors |
352 | | =1 required for subtypes 2 and 3 |
353 | | screen: Control verbosity of PDAF |
354 | | 0: no outputs |
355 | | 1: basic output (default) |
356 | | 2: 1 plus timing output |
357 | | 3: 2 plus debug output |
358 | | |
359 | | +++++++++ End of option overview for the LESTKF ++++++++++ |
360 | | }}} |
| 283 | PDAF Available options for LESTKF: |
| 284 | PDAF --- Sub-types (Parameter subtype) --- |
| 285 | PDAF 0: Standard implementation with ensemble integration |
| 286 | PDAF 2: Fixed error space basis |
| 287 | PDAF 3: Fixed state covariance matrix |
| 288 | PDAF 5: Offline mode |
| 289 | PDAF --- Integer parameters (Array param_int) --- |
| 290 | PDAF param_int(1): Dimension of state vector (>0), required |
| 291 | PDAF param_int(2): Ensemble size (>0), required |
| 292 | PDAF param_int(3): Size of lag for smoothing |
| 293 | PDAF param_int(4): not used |
| 294 | PDAF param_int(5): Type of forgetting factor; optional, default: 0 |
| 295 | PDAF 0: fixed forgetting factor |
| 296 | PDAF 1: adaptive forgetting factor for full domain (experimental) |
| 297 | PDAF 2: locally adaptive forgetting factor (experimental) |
| 298 | PDAF param_int(6): Type of ensemble transformation matrix; optional, default: 0 |
| 299 | PDAF 0: deterministic omega |
| 300 | PDAF 1: random orthonormal omega orthogonal to (1,...,1)^T |
| 301 | PDAF 2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T |
| 302 | PDAF (experimental; for random transformations, 1 is recommended) |
| 303 | PDAF param_int(7): Type of transformation matrix square root; optional, default: 0 |
| 304 | PDAF 0: symmetric square root |
| 305 | PDAF 1: Cholesky decomposition |
| 306 | PDAF --- Floating point parameters (Array param_real) --- |
| 307 | PDAF param_real(1): Forgetting factor (usually >0 and <=1), required |
| 308 | PDAF --- Further parameters --- |
| 309 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 310 | PDAF >=1 for subtypes 0 and 1; not larger than total number of processors |
| 311 | PDAF =1 required for subtypes 2 and 3 |
| 312 | PDAF screen: Control verbosity of PDAF |
| 313 | PDAF 0: no outputs |
| 314 | PDAF 1: basic output (default) |
| 315 | PDAF 2: 1 plus timing output |
| 316 | PDAF 3: 2 plus debug output |
| 317 | PDAF +++++++++ End of option overview for the LESTKF ++++++++++ |
| 318 | }}} |