SPEI and hurst analysis of precipitation in the Amazonian area of Brazil

Abstract

Amazon rainforest controls, to some extent, the global hydrological cycle. The objectives of the present work were (i) to investigate hydrologicalpatterns in a larger regionof the Brazilian rainforestthrough 9-month SPEI series and (ii) to search for long or short-term correlation within the number of days with precipitation by monthand accumulated monthly rainfall. Data sets were collected from 15 meteorological stations spanning a large area of the Amazonian rainforest in Brazil. We computed SPEI values from monthly precipitation and monthly meantemperature time series and determined Hurst exponents from detrendedtime series of days with precipitationand accumulated monthly rainfall. In the first case SPEI was determined on a 9-month timescale while Hurst exponents were calculated from rescaled range analysis. Percentage of SPEI values in the near normal class (-1 ≤SPEI ≤ 1) ranged from 59.8% (Peixe, TO) to 69.7% (Tarauaca, AC). The Hurst exponent varied from 0.382 (Diamantino, MT) to 0.636 (Tefé, AM) and correlated positively with the monthly meanrainfall. This indicates a persistenttrend of wet patterns in the futurein some areas. Hurst analysis of days with precipitation and monthly rainfallcould be an additional tool for interpreting rainfall data.