Online Resource_1On the figure top left the South America map indicating the case study region. Just below RGs before the consistency tests (634) and on the bottom RGs after the consistency analysis (207). On the larger mapa zoom of the case study region with the 207 RGs. The Brazilian Biomes are also represented along with the Arch of Deforestation in Southern Amazonia
Online Resource_2 A Cumulative difference precipitation for RG 0074900 (ANA) for the 40-year period. X-axis represents months of the year with their respective lines for the period. The calculus was obtained from Leibmann and Mechoso (year) method.
Online Resource_2 BSpatial distribution of the 89 RGs used in the Rainy Season’s Onset and Offset analysis. The Brazilian Biomes are also represented along with the Arch of Deforestation in Southern Amazonia.
Online Resource_3 A In the graph RGs illustrating ruptures in the historical series after the statistical analysis in Pettitt’s test. In red: negative ruptures and in blue: positive. In the “x” axis: the years where the rupture was identified and in the “y” axis on the right, the number of RGs with rupture. The colored lines indicate the total rainfall in mm during the period for all RGs. On the right y-axis: Total rainfall in mm for Southern Amazonia (dark green line includes RGs in the Amazon Biome, Cerrado+Pantanal Biome), Amazon RGs (light green line), Cerrado RGs (orange line), and Pantanal RGs (violet line).
Online Resources_3 BPettit’s test depicting results for RGs with ruptures. Yellow dots represent negative ruptures and blue triangles represent positive ruptures during 1971-2010 period
Statistical Tests / Ruptures/TendenciesPositive (+) / Ruptures/Tendencies
Negative (-) / No Rupture/Tendency
Pettitt’s Test / 6 / 34 / 167
Mann-Kendall’s Annual Test / 10 / 35 / 162
Mann-Kendall’s Monthly Test / 27 / 94 / 86
Pettitt’s Test + Mann-Kendall’s Annual Test / 4 / 19 / 184
Online Resources_4 Table 1 207 RGs according to the Mann-Kendall’s tendency test and the Pettitt’s rupture test
Online Resource_4 A RGs results showing positive and negative tendencies in the historical series according to the Mann-Kendall’s Monthly test (1970-2010). Note that stations in the graph can account for more than 1 month since Mann-Kendall’s test can show tendencies in several months. Stations with no tendency are not showed
Online Resource_4 B Mann-Kendall Monthly Seasonal test depicting average results for RGs with tendencies. Yellow dots represent negative tendencies and blue triangles represent positive tendencies during 1971-2010 period
Online Resources_5 A Ordinary Kriging Interpolation Methodology
There are many types of Kriging Interpolation and each method has many options. The Kriging method is used whenever spatial variation is too high and needs a stochastic function to be described (Burrough and McDonnell, 1998). Kriging method choices and options require decision-making what provides the correct method to tuning the interpolation strategy for rainfall studies. Kriging uses a variogram model to characterize spatial correlation. The variogram describes in terms of variances how spatial variability changes as a function of distance and direction (Isaaks and Srivastava, 1989).
To assess the 207 RGs and 89 RGs used in the Linear Regression and Rainy Season’s Onset and Offset analysis respectively, we have decided to use the Exponential Ordinary Kriging method(see Fig 1).The ordinary Kriging is the basic form of Kriging, which is composed by a linear combination of measured values. Thespatial correlation data described by the variogram determines its weight (Hengl, 2009). The Kriging maps in our analyses were derived from the Linear Regression and Mann-Kendall’s results originated from the script’s run for the season’s onset and offset. We used ArcGIS 10 for map point interpolation. An example of a variogram used in the 207 RGs Linear Regression analyses is illustrated below:
Online Resources_5 AOrdinary Exponential Kriging Variogram example composed of 207 RGs assessed in the Linear Regression Analysis. Empty dots close to the curve indicate good spatial neighborhood correlation among RGs data. On the x-axis the distance (between points) and on the y-axis the semivariance among points. Parameters found for the variogram were: Nugget - 7.85, Range - 23.55 km and PhilSill - 28
References
Burrough, P.A. & R.A. McDonnell (1998) Principles of Geographical Information Systems. Oxford: Oxford University Press.
Hengl, T (2009) A Practical Guide to Geostatistical Mapping. 2nd ed. Amsterdam: University of Amsterdam.
Isaaks, E.H. & R.M. Srivastava (1989) An Introduction to Applied Geostatistics.
Online Resource_5 B Linear tendency interpolated through LR analysis for the 207 RGs(black dots) in the states of Mato Grosso and Rondônia (Southern Amazonia). The red colors indicate reduction tendencies in precipitation and blue an increase. The data is depicted in mm(-1) for the period of 1971-2010
Rain Gauges (ANA)Averages Onset – Offset
Biome / Onset / Offset
Amazonia
70S/120S / Cerrado
130S/180S / Amazonia
IBAMA / Cerrado IBAMA / Amazonia
70S/120S / Cerrado
130S/180S / Amazonia
IBAMA / Cerrado
IBAMA
Number of RGs / 37 / 52 / 38 / 51 / 37 / 52 / 38 / 51
Day Value / 274 / 280 / 275 / 279 / 121 / 108 / 121 / 108
Day Period / 30/09 / 6/10 / 1/10 / 5/10 / 30/04 / 17/04 / 30/4 / 17/4
Online Resource_6 A Table 2 Accumulated daily averages for the 89 RGs. Monthly and daily averages for the onset and offset of the rainy season in the Amazon and Cerrado between 7oS and 12oS, and 130S and 180Sof latitude
Online Resource_6 B Onset and offset dates for the RG 0074900 (ANA) during the last 40 years. The graph illustrates the most recurrent data among RGs. Both onset and offset phases have approximated their dates during the last 40 years
Online Resource_7A MK test tendencies for the onset and offset of the rainy season (1971-2010). Top map: yellow circles represent tendencies intensity for a later onset and the blue triangles represent a premature onset. Bottom map: larger yellow circles represent tendencies intensity for a premature offset and blue triangles represent a later offset
Online Resource_7B LRA analysis for the onset and offset of the rainy season (1971-2010).Top map: yellow circles represent tendencies for a later onset and the blue triangles represent a premature onset. Bottom map: yellow circles represent tendencies for a premature offset and blue triangles represent a later offset