R help - Apr 2007 - zoo merge() method

R users:

I'd like to get some insight on an error I encounter when attempting to
work with two moderately sized sets of time series data.  FYI - I'm
using the following versions of R and supporting packages on a Windows
2000 OS:

- R version 2.4.1 (2006-12-18) 
- zoo version 1.2-2
- chron version 2.3-10

The two time series I'm working with are from the summer of 2004 and
are: 
1.) wet.bulb.air.temp: air temperatures recorded on an hourly basis, and

2.) creek.temperature: surface water body temperatures collected every
12 minutes.

I would ultimately like to observe the difference in temperatures and
attempted to get at this by merging the two time series (by union),
interpolating the NAs, and finally, subtracting one vector from the
other.  The problem is that I can not combine the two zoo time series
objects using the merge() or cbind() functions.  I get the following
error:

Error in z[match0(index(a), indexes), ] <- a[match0(indexes, index(a))]
: 
        number of items to replace is not a multiple of replacement
length

The input/output from a recent R Console session might help, so I've
included it as follows:
> summary(creek.temperature)
     Index                     creek.temperature
 Min.   :(07/21/04 00:03:00)   Min.   :12.82    
 1st Qu.:(08/11/04 03:00:00)   1st Qu.:16.28    
 Median :(09/01/04 03:45:00)   Median :18.53    
 Mean   :(09/01/04 04:45:13)   Mean   :18.87    
 3rd Qu.:(09/22/04 06:37:00)   3rd Qu.:21.48    
 Max.   :(10/13/04 09:22:00)   Max.   :27.72    
> length(creek.temperature)
[1] 10140
> summary(wet.bulb.air.temp)
     Index                     wet.bulb.air.temp
 Min.   :(07/01/04 00:00:00)   Min.   : 3.889   
 1st Qu.:(07/31/04 12:00:00)   1st Qu.:12.778   
 Median :(08/31/04 00:00:00)   Median :14.444   
 Mean   :(08/31/04 00:00:00)   Mean   :14.469   
 3rd Qu.:(09/30/04 12:00:00)   3rd Qu.:16.667   
 Max.   :(10/31/04 00:00:00)   Max.   :22.222   
> length(wet.bulb.air.temp)
[1] 2929
> class(creek.temperature)
[1] "zoo"
> class(wet.bulb.air.temp)
[1] "zoo"
> merge(wet.bulb.air.temp, creek.temperature)
Error in z[match0(index(a), indexes), ] <- a[match0(indexes, index(a))]
: 
        number of items to replace is not a multiple of replacement
length
> cbind(wet.bulb.air.temp, creek.temperature)
Error in z[match0(index(a), indexes), ] <- a[match0(indexes, index(a))]
: 
        number of items to replace is not a multiple of replacement
length

The really puzzling part about this error is that it does not occur when
I pare down the data sets and only look at a 24 hour window of data (for
brevity, the input/output from that exercise has not been included in
this e-mail).

My question to the R user community is as follows:  What is this error?
How do I get past it and get these two data sets to play nice with each
other?

Thanks,

Matt Findley

-------------------------------------
Matthew C. Findley, CPSSc
Environmental Scientist
 
CH2M HILL
2300 NW Walnut Blvd
Corvallis, OR 97330-3538
 
Tel: 541.768.3504
Fax: 541.752.0276
 
mfindley at ch2m.com

On Thu, 12 Apr 2007 Matthew.Findley at ch2m.com wrote:
> attempted to get at this by merging the two time series (by union),
> interpolating the NAs, and finally, subtracting one vector from the
> other.  The problem is that I can not combine the two zoo time series
> objects using the merge() or cbind() functions.  I get the following
> error:
>
> Error in z[match0(index(a), indexes), ] <- a[match0(indexes, index(a))]
> :
>         number of items to replace is not a multiple of replacement
> length
Usually, problems like this occur when the time stamps in one of your time
series are not unique. Maybe we should improve the error message by
explicitely trying to catch this error.

You can easily check this, e.g., via
  any(table(time(zoo_object)) > 1)
> The input/output from a recent R Console session might help,
Not really, the data itself would have been much more helpful...and even
better some simplified artificial data set.

hth,
Z