A Counting Process Approach for Trend Assessment of Drought Condition

Abstract

This paper discusses some methodological aspects of the historical analysis of drought, particularly the trend assessment. The Standardized Evapotranspiration Index (SPEI) is widely used as a measure of drought condition. Since different SPEI thresholds allow classifying the risk into moderate, severe, and extreme, the drought occurrence becomes a counting process. In this framework, would a statistical trend test based on a Non-Homogeneous Poisson Process (NHPP) give a similar result of the nonparametric Mann–Kendall (M-K) test? In this paper, we demonstrate that the NHPP approach is able to characterize the information given by the classical M-K approach in term of drought risk classes. Furthermore, we show how it can be used to reinforce the framework of drought trend analysis in combination with a standard non-parametric approach. At a global scale, we find that: (1) areas under increasing risk of drought identified by the NHPP approach are considerably larger in comparison to those identified by M-K; and (2) the results of the two tests are different during crucial periods such as hydrological droughts in winter and spring.

Spatiotemporal climate variability and extremes in Middle Awash Afar region Ethiopia: implications to pastoralists and agro-pastoralists food security

“…This study calculated the SPI and the SPEI indices using a 3–12-month time scale. The three-month time scale is crucial for understanding agricultural drought events, while the 12- and 24-month time scales are exclusively used to analyze hydrological drought events or annual water deficits (Di Giuseppe et al , 2019; Alsafadi et al , 2020; Potop et al , 2012).…”

Variation of Ground Surface Freezing/Thawing Index in China under the CMIP6 Warming Scenarios

“…The non-parametric test method is also called the non-distribution test. The advantage is that the known data do not need to follow certain distribution rules (e.g., Gaussian distribution) when participating in the calculation, and it is not affected by individual outliers [46,47]. The second method is rescaled range (R/S) analysis, which can distinguish a random sequence from a non-random sequence, and it can also perform the exploration of long-term memory processes in nonlinear systems.…”

Analyzing asymmetries in the response of European beech to precipitation anomalies in various stand and site conditions using decadal diameter censuses

“…We also calculated the significance of the trend in the SPI in the period between two measurements using the Mann-Kendall Trend Test and Sen's slope method to evaluate the direction of the trend (MK_SSE) using the sens.slope() function in the trend R package (v1.1.4; Di Giuseppe et al, 2019;Pohlert, 2020). A negative MK_SSE indicates a trend of decreasing wetness/increasing drought, whereas a positive MK_SSE indicates a trend of increasing wetness/decreasing drought (Gocic and Trajkovic, 2013;Sen, 1968).…”