Using the monthly classification of global SSTS and 500 HPA height anomalies to predict temperature and precipitation regimes one to two seasons in advance for the mid-Mississippi region.(Report)

Using the results from previous studies of the interannual variability of local mean monthly temperature and precipitation by this group, long range forecasts were generated for summer and winter season temperatures and precipitation four to five months in advance using analogs and/or contingency tables. These forecasts also included information about El Nino phase and east Pacific region blocking events. Summer and winter season forecasts of temperature and precipitation are of interest to the local and regional media as well as the agricultural communities in the mid-Mississippi River Valley. A simple forecast verification scheme was borrowed and used to score the long range forecasts, and skill scores were used to compare and evaluate the forecasts against climatology. The results show that these forecasts have been better than climatology, in general, especially in the summer season and for seasonal temperature forecasts.

The research results used here have demonstrated that Pacific region SSTs and SST anomalies can be separated into seven general synoptic classifications ("clusters", A-G). Some of these clusters are shown to have a distinct impact on the barotropic component of the mean tropospheric height distributions as well. Clusters A, B, E, and G (C, D, and F) have been shown to be representative of La Nina (El Nino) type SST distributions by previous studies. Further, an analysis of the SST patterns from 1955-2007 demonstrated that certain clusters were prominent from 1955-1977, and from 1999 to the present; others dominated the period in between. This shift in prominent patterns during 1977 and 1999 corresponded roughly with a change in phase of the Pacific Decadal Oscillation (PDO). Some SST anomalies were correlated with warmer or cooler than normal conditions in the mid-Mississippi region, while others did not produce definitive results.

1. Introduction

Many recent studies have attempted to link variations in global circulation changes (e.g., Wallace and Gutzler 1981; Gershanov and Barnett 1998; Enfield and Mestas-Nunez 1999; Mestas-Nunez and Enfield 1999, 2001; Wiedenmann et al. 2002), or local (and regional) climate variations (e.g., Kung and Chern 1995 [hereafter KC95]; Kunkel and Angel 1999; Lupo et al. 2007) with inter annual and interdecadal variations in sea surface temperatures (SSTs) and pressures in the Pacific Ocean basin and/or the changes in the character of the atmospheric and oceanic circulations in the Atlantic Ocean basin (e.g., Hu et al. 1998). The interactions between the atmosphere and oceans are important processes to consider when attempting to either understand the relevant physics of the earth's climate system or to make long-range forecasts (e.g., Anderson et al. 1999; Barnston et al. 2005).

The dominant interannual variations in global and regional climate characteristics are largely influenced by El Nino and Southern Oscillation (ENSO) modes (e.g., Mokhov et al. 2000, 2004). Diagnosing regional and local climate variability has been a topic of interest lately, since global circulation models are used heavily to study the potential for climate change (Houghton et al. 2001) and in long range forecasting. Thus, it is critically important that these models be able to demonstrate that they can faithfully simulate not only the range of regional and local climates, but the interannual and interdecadal variations as well. It is well known that tropical SST distributions and "anomalous" SST distributions have a large impact on the weather and climate by changing heat and mass distributions of the troposphere. Through this influence, SSTs can ultimately alter the prevailing wind patterns over a large portion of the globe (e.g., Namias 1982, 1983; Enfield and Mestas-Nunez 1999; Mestas-Nunez and Enfield 1999, 2001; Wiedenman et al. 2002). This in turn can impact the frequency, occurrence, and intensity of such phenomena as mid-latitude cyclones (e.g., Key and Chan 1999), tropical cyclones (e.g., Gray 1984) and blocking anticyclones (e.g., Wiedenmann et al. 2002). However, there are studies (e.g., Enfield and Mestas-Nunez 1999; Mestas-Nunez and Enfield, 1999, 2001; Kushnir et al. 2002) that point out that mid-latitude SSTs may not be very influential on mid-latitude circulations. Nonetheless, the influence of tropical SSTs on the mid-Missouri area regional climate have been demonstrated, albeit indirectly, via the impacts on snowfall regimes (e.g., Lupo et al. 2005), tornado occurrences (e.g., Akyuz et al. 2004), and temperature and precipitation regimes (e.g., Hu et al. 1998; Lupo et al. 2007).

KC95 used principal component analysis (e.g., Wilks 2006) to extract the large-scale modes of monthly mean global SST anomalies and the Northern Hemisphere tropospheric circulation anomalies during the period 1955-1993. A similar analysis was performed by Enfield and Mestas-Nunez (1999) and Mestas-Nunez and Enfield (1999, 2001), but using data covering the period from 1870-1991. The KC95 study provided an archive which can be used as guidance for long-range forecasting applications (e.g., forecasting by the use of analogs and/or contingency tables). This analysis was then extended to 2005 by Lupo et al. (2007). A by-product of these analyses demonstrates that global SST anomalies could be classified into one of seven distinct pattern types (A-G). Each of these was correlated with corresponding Northern Hemisphere tropospheric mass distributions or flow anomalies, and in subsequent work, correlated with surface climatic characteristics in mid-Missouri (Lee and Kung 200). It is noted here, however, that correlations between SSTs and atmospheric flow patterns do not imply any cause and effect relationship between these quantities. KC95 and Lupo et al. (2007) also noted that anomaly types (clusters) A, B, E, and G (C, D, and F) are representative of La Nina or neutral (El Nino) type SST distributions within the Pacific Ocean basin. They also demonstrated that clusters A-D dominated the negative phase of the Pacific Decadal Oscillation (PDO) (1955-1977, and 1999-present), while E and F type clusters dominated the middle portion (1977-1998). Lupo et al. (2007) then demonstrated that some of these are associated strongly with certain temperature and precipitation regimes in the mid-Mississippi valley region (e.g., D-type patterns are strongly correlated with dry months in the region).

Thus, this work has two primary objectives. First, the work of Lupo et al. (2007) will be discussed (sections 2 and 3) and this includes an analysis of SST anomaly types and their correlation to monthly temperatures and precipitation in the mid-Mississippi valley region as represented by a time series from the Columbia Regional Airport. Their work is then extended here (section 4) to discuss the synoptic-scale flow regimes associated with prolonged SST anomaly distributions of each of the seven types discussed above. This paper then examines the usefulness of these results in making long range forecasts made by the long range prediction group at the University of Missouri-Columbia, and the verification of these forecasts (section 5). This includes a discussion of summer season blocking in the East Pacific and the relationship to temperatures and precipitation in our study region. We will demonstrate that these forecasts are better than a commonly used baseline forecast (climatology).