[hana waga]

hi how are you? I use both version of IHACRES,GR4J and AWBM model for watershed characterization but the result in AWBM for all routing fit , calibration,simulation and validation is underestimated with out calibration using fit byDream and fit bySCE. which shows satisfactory result for calibration ONLY. how to correct ? i use the followings;

model specification

```{r}

Achemmodel<-hydromad(data89, sma = "awbm", routing = "expuh", rfit = list("sriv", order = c(2, 1)),warmup = 365)
print(Achemmodel)

```

for routing fit

```{r}

 ihSpec <- hydromad(data89, sma = "awbm", routing = "armax",cap.ave=500,tau_s=9,etmult =1)
 osumm <- tryModelOrders(update(ihSpec, rfit = "sriv"), n = 1:3, m = 1:3, delay = 1)
 Arfit<-summary(osumm)

```

```{r}

```{r,echo=FALSE,warning=FALSE,message=FALSE,results='hide'}
AWchemmodel<-update(Achemmodel,routing = "armax",rfit=list("sriv",
           order=c(n=2, m=1)),cap.ave =100,etmult=0.7,d=1)
```

```{r,echo=FALSE,warning=FALSE,message=FALSE,results='hide'}
OAchemFit <- fitByOptim(Achemmodel, samples=100, method="PORT")
OAchemFit
library(DEoptim)
DAchemFit <- fitByDE(Achemmodel, control = DEoptim.control(itermax = 5))
DAchemFit
SCAchemFit <- fitBySCE(Achemmodel,control = list(maxit = 5, ncomplex = 2))
SCAchemFit
 SaAchemFit <- fitBySampling(Achemmodel)
 SaAchemFit
 library(dream)
 Adream<-fitByDream(GModel, control = list(ndraw = 500))
 Adream
```

|            |   rel.bias|  r.squared|  r.sq.sqrt|   r.sq.log|
|:-----------|----------:|----------:|----------:|----------:|
|calibration |  1.2228643| -0.4877669| -0.0183647| -0.2351904|
|DAchemFit   |  1.1133293| -1.1914489|  0.1600318|  0.5047655|
|SCAchemFit  | -0.0719441|  0.5160834|  0.5950572|  0.2283745|
|SaAchemFit  |  1.0804228| -1.2220687|  0.1577858|  0.4352051|
|Adream      | -0.0138886|  0.5027873|  0.6883075|  0.6199139|
|sim89       |  1.2228643| -0.4877669| -0.0183647| -0.2351904|
|sim2000     |  1.4336536| -2.7678647| -0.4332218| -0.2597431|
|simVerif    |  1.8523780| -3.4038095| -0.5779129| -0.3562355|

i try to find good result but untilknow underestimate by all fit options. how to correct this?

[Willem Vervoort]

Dear Hana,

I am not really sure what you answer you are trying to get in relation to hydromad. There is nothing wrong with your code, it does exactly what it should be doing.

If you are worried about the results in the validation and verification being different, I think you should read up on the hydrological literature. What you are doing is what is called a differential split sample test (Klemes, 1986) and there are lots of issues why this might or might not be different. The paper by Fowler et al. (2018) is probably a good starting point

Regards

Willem

Fowler, K., et al. (2018). "Simulating Runoff Under Changing Climatic Conditions: A Framework for Model Improvement." Water Resources Research 54(12): 9812-9832.
Klemes, V. (1986). "Dilettantism in hydrology: Transition or destiny." Water Resources Research 22(9): 177S-188S.

--
You received this message because you are subscribed to the Google Groups "hydromad users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hydromad-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hydromad-users/CAOmnh4ty3_PmmPUn9EMEsUpKQ80GHtzYH84b46xzrWdTwEAUvw%40mail.gmail.com.