SPI-based drought category prediction using loglinear models

Abstract

Loglinear modeling for three-dimensional contingency tables was used with data from 14 rainfall stations located in Alentejo and Algarve region, southern of Portugal, for short term prediction of drought severity classes. Loglinear models were fitted to drought class transitions derived from Standardized Precipitation Index (SPI) time series computed in a 12-month time scale. Quasi-association loglinear models proved to be the most adequate in fitting all the 14 data series. Odds and respective confidence intervals were calculated in order to understand the drought evolution and to estimate the drought class transition probabilities. The validation of the predictions was performed for the 2004-2006 drought, particularly for periods when the drought was initiating and establishing, and when it was dissipating. Despite the contingency tables of drought class transitions present a strong diagonal tendency, results of three-dimensional loglinear modeling present good results when comparing predicted and observed drought classes with 1 and 2 months lead for those 14 sites. Only for a few cases predictions did not fully match the observed drought severity, mainly for 2-month lead and when the SPI values are near the limit of the severity class. It could be concluded that loglinear prediction of drought class transitions is a useful tool for short term drought warning. ª

Monthly Prediction of Drought Classes Using Log-Linear Models under the Influence of NAO for Early-Warning of Drought and Water Management

“…However, to avoid problems in the model fitting [64] caused by too many class transitions with value zero, several classes were grouped. We have considered a total of four classes in our modeling analysis as in recent studies [65] albeit not necessarily exactly defined in the same way.…”

“…Among the techniques used for drought forecasting, statistical models are chosen many times, since they are simple to implement, do not have a high computational burden, and produce useful predictions [58]. There are a variety of statistical methodologies available which can be applied for the intended purpose, namely autoregressive integrated moving average (ARIMA)-type approaches [59,60], artificial neural network (ANN) models [61,62] or even other types of stochastic and probability models, such as Markov chains [63], log-linear models [64,65], and others [66,67]. A thorough discussion on various methodologies used for drought modeling and prediction showing the limitations and advantages of each modeling/technique was done by Mishra and Singh [58].…”

“…Taking into account the NAO state in the current month, the most probable drought class transition for the next month and their confidence intervals are computed. This approach has proved its usefulness in predicting the SPI drought class one or two months in advance [64,65,69].…”

“…Additionally, we aim to go a step further from previous studies using log-linear models [64,65,69], identifying the predictors that can better predict changes of drought classes one or two months in advance, namely with the use of a four-month NAO average as a predictor instead of the single lagged NAO value as in a previous study [65]. An analysis of the extended winter months (November-May) and the complete annual period was performed, as well a study of the use of different SPI (SPEI) drought class intervals, aiming at a better assessment of the Prut basin climate.…”

Identifying the relationships of climate and physiological responses of a beech forest using the Standardised Precipitation Index: a case study for Slovakia

“…In Table 1 are described values of the SPI with corresponding cumulative probability and with appropriate classification of the severity by McKee et al (1995). The SPI has been applied in a number of international works dealing with the research of drought impacts on ecological environmental conditions (Kim et al, 2014;Moreira et al, 2008;Nam et al, 2015). …”