[R] spacetime stConstruct gives wrong dimensions for long table

[Bebber, Dan]

I have a large spatiotemporal database (131 spatial locations, 9 years of
daily weather data) in long table format which I would like to convert to
an ST object.

> head(tempmean)
           site  lat    lon  alt                              var year mth
day value       date doy numdate
518941 27015260 6.02 -75.37 1680 AVERAGE DAILY TEMPERATURE VALUES 2006   1
  1  16.9 2006-01-01 001   13149
518942 27015260 6.02 -75.37 1680 AVERAGE DAILY TEMPERATURE VALUES 2006   1
  2  16.4 2006-01-02 002   13150
518943 27015260 6.02 -75.37 1680 AVERAGE DAILY TEMPERATURE VALUES 2006   1
  3  16.9 2006-01-03 003   13151
518944 27015260 6.02 -75.37 1680 AVERAGE DAILY TEMPERATURE VALUES 2006   1
  4  16.1 2006-01-04 004   13152
518945 27015260 6.02 -75.37 1680 AVERAGE DAILY TEMPERATURE VALUES 2006   1
  5  16.6 2006-01-05 005   13153
518946 27015260 6.02 -75.37 1680 AVERAGE DAILY TEMPERATURE VALUES 2006   1
  6  16.7 2006-01-06 006   13154


The data.frame is ‘ragged’, i.e. not all sites are represented by all
dates, and there are some NA values in the data.

I took a subset of the data (4 sites, 31 days in January 2006, for which
all site/date combinations are present) and tried

> st <- stConstruct(tempmean, 3:2, 10)
> summary(st)
Object of class STIDF
 with Dimensions (s, t, attr): (124, 124, 9)
[[Spatial:]]
Object of class SpatialPoints
Coordinates:
       min    max
lon -76.45 -73.11
lat   2.11   7.07
Is projected: NA 
proj4string : [NA]
Number of points: 124
[[Temporal:]]
     Index              timeIndex
 Min.   :2006-01-01   Min.   :  1.00
 1st Qu.:2006-01-08   1st Qu.: 31.75
 Median :2006-01-16   Median : 62.50
 Mean   :2006-01-16   Mean   : 62.50
 3rd Qu.:2006-01-24   3rd Qu.: 93.25
 Max.   :2006-01-31   Max.   :124.00


This is incorrect: There should be 4 spatial dimensions and 31 timepoints.

What am I doing wrong?

Thanks,
Dan

Dr Dan Bebber
Senior Research Fellow
Biosciences
University of Exeter