Precipitation Variability for Protected Areas of Primary Forest and Pastureland in Southwestern Amaz(o)over-capnia

Daily and monthly rainfall data provided by surface rain gauges in the Amazon Basin are sparse and defective, making it difficult to monitor rainfall patterns for certain portions of its territory, in this sense, estimations of precipitation from remote sensing calibrated with rain gauge data are key to overcome this problem. This paper presents a spatiotemporal analysis of the precipitation distribution for Rondonia State, in southwestern Amazonia. Data from Climate Hazards Group InfraRed Precipitation and Station (CHIRPS) were analyzed, using a pooled time analysis of a forty-year period (1981-2020). Data obtained from remote sensing were validated by rain gauges distributed over the study region. Pixel-by-pixel trend analyzes were developed by applying the Mann-Kendall test and Sen's slope test to study the magnitude of the trend. The analysis revealed that CHIRPS presents a tendency to underestimate precipitation values in most cases. Among the metrics, mean values between very good (<+/- 15%) and good (+/- 15-+/- 35%) were observed using PBIAS; mean RMSE values range from 57.8 mm to 107.9 mm; an average agreement level of 0.9 and an average SES of 0.5; and good fit for the linear regression model (average R-2 > 0.70) for about 64.7% of the stations. Sen' slope spatialization results show a reduction of approximately -15 mm year(-1), with decrease mainly in the Northern Region of Rondonia, which has extensive areas where the native forest has been replaced by pasture.

Automated summary

Rainfall data in the Amazon Basin provided by surface rain gauges are limited and flawed, making it challenging to monitor rainfall patterns in certain areas. To address this issue, this study utilizes remote sensing calibrated with rain gauge data to estimate precipitation distribution in Rondonia State, southwest Amazonia. The analysis reveals that CHIRPS tends to underestimate precipitation values in most cases.