This RHtestV3 software package can be used to detect, and adjust for, multiple changepoints (shifts) that could exist in a data series that may have first order autoregressive errors [but excluding daily precipitation data series, for which the RHtest_dlyPrcp package (given below, after the RClimDex package) should be used]. It is based on the penalized maximal t test (Wang et al. 2007) and the penalized maximal F test (Wang 2008b), which are embedded in a recursive testing algorithm (Wang 2008a), with the lag-1 autocorrelation (if any) of the time series being empirically accounted for. The problem of uneven distribution of false alarm rate and detection power is also greatly alleviated by using empirical penalty functions (Wang et al. 2007, Wang 2008b). The time series being tested may have zero-trend or a linear trend throughout the whole period of record. A homogenous time series that is well correlated with the base series may be used as a reference series. However, detection of changepoints is also possible with the RHtestV3 package when a homogenous reference series is not available.
The RHtestV3 package is an extended version of the RHtestV2 package. The extension includes: (1) provision of Quantile-Match (QM) adjustments ( Wang 2009 ) in addition to the mean-adjustments that were provided in the RHtestV2; (2) choice of the segment to which the base series is to be adjusted (referred to as the base segment); and (3) choices of the nominal level of confidence at which to conduct the test. This package has been developed and maintained by Xiaolan Wang and Yang Feng at Climate Research Division. Equivalent FORTRAN programs are also available by sending a request to Xiaolan Wang.