Information on the California Climate Tracker


Rationale:

The state of California is dominated by its diverse topography, ranging from the coastal environs, to the great Central Valley, the Sierras and the Mohave-Sonoran Desert.  California's varying landscape gives way for a number of physical mechanisms that not only influence the average climate, but also climate variability across the state.  The analysis of climate variability and trends is a crucial and necessary component in understanding the role of climate change.  Although there exists clear indication of changes in the global surface temperature, the regional manifestation of climate change is not well quantified at the present time.  Driven by the interests of the governmental, economic, and scientific communities it is pertinent to develop an objective method to define and monitor climate not only for the state as a whole, but also for its distinct climate regions.

The National Climatic Data Center (NCDC) has developed climate divisions that span the contiguous United States, whereby each state has been subdivided into 10 or fewer climate divisions.  Across much of the western United States climate divisions were guided mostly by watershed and river basins, as opposed to climatological patterns.  Consequently, the divisions suffer from a number of problems in the western United States, where complex topography plays a strong role in dictating regional climate patterns.  For example, the current climate divisions in California make no distinction between the Sierra Nevada Mountains and the Central Valley.  Not only does the Central Valley and Sierra differ greatly in terms of average temperature and precipitation, but also in terms of variations in temperature and precipitation. 


Data:

Monthly station data, taken from cooperative observers (COOP), along with gridded data from the PRISM database, are used to assess climate across the state.  The primary variables that are considered in this process are monthly average mean temperatures and monthly precipitation totals.  We consider COOP stations across the state that reported over 75% of observations over the time period 1949-2005, and continued to report in 2006.  A total of 195 stations across the state are included in this analysis.  We consider COOP station data along with the PRISM database dating back to January of 1895. Temperature data from the COOP stations have been adjusted for inhomogeneities, a procedure used to "correct" for non-climate shifts in temperature.  No effort is made to adjust for urbanization or land-use changes. Inhomogeneity detection includes the entire period of record; however we caution that the dataset contains larger uncertainties prior to 1918 due to the limited number of stations reporting statewide.


Methodology:

Variations in climate are at the forefront of both the public mind and of climate researchers. The initial steps in creating the California Climate Tracker was to identify cohesive regions of climate variability within the state. Using an infilled dataset we perform an EOF analysis on the COOP station data using both monthly precipitation and average monthly mean temperature. This analysis focuses on how stations vary with one another.  We identify 11 distinct regions across the state wherein stations located within a region vary with one another in a similar fashion.  An analogous analysis is performed with the PRISM data, resulting in striking similar results.  These 11 regions hereafter define our climate regions.

The collection of data from both station and PRISM data from these regions is used to create a single value for each variable for each month. This is a two step process, dependent on the timing of data availability of both COOP station based data and the gridded PRISM based data. An effort is made to create a seamless translation between these two datasets. At the beginning of each month only COOP data is available, and often from around only 60% of the 195 stations statewide. Data is first screened for outliers (defined as a data anomaly that exceeds more than two standard deviations from any other anomaly within the state). Temperature datasets are also screened for inhomogeneities, that lead to an abrupt, non-climatic, change in the time series of a given station. An effort is then made to estimate missing stations from anomaly regression with highly correlated reference stations.At this point, the regional value is computed by the average of the collection of COOP stations within a region. Further adjustment is then made to adjust for inherent biases between the COOP based value and the PRISM-based areal average (e.g., COOP stations in mountainous terrain are generally located at elevations lower than the mean topography of the region, and are regularly warmer than the areal average of the domain of interest). Note that for the first couple weeks of each month that the most updated value is generated by station based data.

The PRISM group at Oregon State calculates monthly datasets within the first few weeks of each month. As the monthly data becomes available it is then incorporated into the California Climate Tracker by taking the areal average of the gridded data with respect to each region. Concurrent with the updating of the PRISM based values, we rerun the COOP based dataset as values continue to be ingested throughout the month. The monthly value reported at this second and final stage is a hybrid value that is weighted equally for PRISM-based data and station-based data.

The statewide average is computed by weighting the regional value by the area covered by each region. An extensive time series dating back to the late 19th century is formed for each region, and for the state as a whole. These time series are used to both categorize and track climate across the spatially diverse state of California in order to place the present climate in context to climate variations back to the late 19th century.

Products:

Statewide, regional, and station based graphics and time series are produced each month. Preliminary products for the last available month will be updated by the 3rd of each month, once the initial COOP station reports come in. A second update will appear near the 10th of each month once the PRISM dataset is updated. The COOP and PRISM products are combined into a single dataset to improve difficulties with sampling errors and the insufficiencies with each dataset. Data from up to twelve months ago is considered provisional, as data continues to filter in and undergoes quality control procedures.

If you acquire data from the California Climate Tracker, we ask that you acknowledge use of the data by citing the reference below:
Citation: Abatzoglou, J.T., K.T. Redmond, L.M. Edwards, 2009, Classification of Regional Climate Variability in the State of California, Journal of Applied Meteorology and Climatology,48, 1527-1541

Return to California Climate Tracker