Revised method to detect the onset and demise dates of the rainy season in the South American Monsoon areas

This paper proposes the detection of dates for the onset and demise of the rainy season (ONR and DER, respectively) for some areas within the South American monsoon system (SAMS) region using procedures based on indices which could be easily calculated and subjected to very simple criterion. Previously, the time series of the averaged antisymmetric outgoing longwave radiation ([AOLR]) in each region of interest was used by some authors to detect ONR and DER. This method is referred to as OLD. First, it was proposed in this paper to smooth this time series with the five-pentad running mean ([RM_AOLR]). This modified method is called NEW. Then, an additional modification was adopted by smoothing the [AOLR] with the three-pentad running mean but with the [AOLR] in the most recent pentad being replaced by its climatological value ([RM_AOLRCLI]). This method is called NEW_CLI. For these methods, the criterion was that the first pentad with negative (positive) [AOLR] (or [RM_AOLR] or [RM_AOLRCLI]) is the ONR (DER) date in the region of interest. All these methods were applied for the Central Amazon (CAM) and Western-Central Brazil (WCB) regions during the rainy seasons in the 2001-2013 period. The efficiency of these methods was analyzed using the precipitation estimated from the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis. The spatial average of the TRMM ([TRMM]) time series with threshold values, which depends on the area and season, was used to determine the ONR and DER dates. In this case, the single method based on the threshold value is called RAIN, and the method which considers the persistence of dry or wet condition is referred to as NEW_RAIN. Comparisons between the methods were performed by calculating the root mean square error (RMSE) when ONR and DER dates obtained from a given method were considered as the true dates. The advantages and disadvantages of the methods were discussed here. It was previously noted that the [AOLR] time series during the transition from dry (wet) to wet (dry) season showed several signs of oscillation. This was a weakness when detecting the ONR and DER dates; however, using the NEW and NEW_CLI methods eliminated this problem. The NEW_CLI method has more advantages than the other methods and will be useful for monitoring the ONR and DER dates of the SAMS regions.