Closed dvictori closed 6 years ago
Table on Perrin et all (2003) mentions the median and a 80% confidence interval for parameters estimated over several watersheds. Maybe use those values as starting point and range
param | median | 80% interval |
---|---|---|
X1 (mm) | 350 | 100 to 1200 |
X2 (mm) | 0 | -5 to 3 |
X3 (mm) | 90 | 20 to 300 |
X4 (days) | 1.7 | 1.1 to 2.9 |
airGR package mentions that parameter interval in transformed parameter space should be from -9.99 to +9.99
param | min | max |
---|---|---|
X1 | 4.58e-05 | 21807.29 |
x2 | -10903.6 | 8798230131 |
x3 | 4.58e-05 | 21807.29 |
x4 | 0.50004 | 21807.79 |
This range looks a bit too broad for me. Will try to use the previous 80% range first
https://cran.r-project.org/web/packages/airGR/vignettes/V02.1_param_optim.html
Place a limit on the values each parameter, so optimization does not go to far