Seasonal prediction of Standardized Precipitation Index (SPI) using fuzzy inference system

Master's Thesis in Civil Engineering - Water (M.Sc)

Abstract

Recognizing the pattern of drought occurrence and its monitoring is important in determining the optimal approach to water resources management, especially regarding the water supply sources of big cities that are exposed to drought events from a climatic point of view. In the meantime, the city of Tehran is exposed to drought and its damages by using five catchment areas and related dams (Amir Kabir, Lar, Letian, Taleghan and Mamlo). In this article, by using atmospheric data in the geographical range [?0, ?60] North and [?0, ?90] East (with an accuracy of 10x10 degrees) and including monthly information of temperature and pressure equivalent height from 1948 to 2008 AD at the levels of 1000, 850, 700, 500 and 300 millibars to forecast the medium-term meteorological drought (with forecast time) 2.5 and 4.5 months) using the standardized precipitation index (SPI) in the precipitation of winter, spring, autumn and the sum of winter and spring, autumn and winter, autumn and winter and spring. In this research, after identifying the point-atmospheric parameters affecting the rainfall pattern in the target areas and using the appropriate statistical criteria, a fuzzy inference system has been developed to predict the index (SPI). The selected parameters in this research are the height equivalent to the recorded atmospheric pressure at two levels of 850 and 300 millibars. The results show the proper efficiency of this estimator in forecasting meteorological drought with appropriate spatial accuracy, and finally, the efficiency of this approach has been quantified by using appropriate statistical indicators.

Key words:

Drought prediction, standardized precipitation profile, fuzzy inference system, meteorological drought

Chapter One:

General research

1-1 Introduction

Changes in the pattern of precipitation and temporal changes and seasonal distribution of precipitation have many economic and social effects on our country, which mainly has an arid and semi-arid climate. Studying and investigating climate changes and understanding the behavior of various meteorological variables such as precipitation, air temperature and pressure is very important, especially in areas that face different climate variations and the occurrence of extreme dry and wet periods. The discussion of predicting various meteorological variables is very important and important, especially in countries that are partially facing drought or are on the verge of drought. On the other hand, these predictions will be useful and have a special value in areas that have multiple earthquakes and flood conditions.  One of the most important information needed for planning and managing water resources is to know the behavior of weather variables in order to predict short-term or long-term hydrological variables. In some cases, these forecasts are made in short-term time frames, which in turn are used for short-term decisions. But sometimes these forecasts are made in long-term periods such as monthly or seasonally, which is very important for seasonal and annual planning of water resources management in many watersheds of the country that rely on surface water resources. 1-2 statement of the basic research problem: One of the most serious challenges in the last decade is the availability of water resources and the effects of climate on the process of precipitation and drought. One of the major obstacles in the allocation and prioritization of water resources is the lack of reliable information and forecasting at the right time. One of the valid parameters for determining the precipitation regime is the standardized index (SPI[1]) of precipitation. McKay [2] and colleagues (1995 and 1993) presented this parameter in order to define and monitor meteorological precipitation. Today, the Colorado Climate Center, the US Western Regional Climate Center, and the US National Drought Center use this index to monitor current drought conditions in the United States. This index allows the analyst to determine the unprecedentedness of a drought or a wet year in a specific time scale for any region of the earth that has a record of historical statistics. In this research, using fuzzy inference systems, the SPI prediction model will be developed. 1-3 The importance and necessity of research In the field of SPI prediction using different statistical methods and artificial intelligence, very little research has been done at the international level. In this context, there has been no research for the use of fuzzy inference system (FIS[3]) and it is expected that the results of this research will lead to the development of a model that uses meteorological information.In this context, there has been no research on the use of the fuzzy inference system (FIS [3]) and it is expected that the results of this research will lead to the development of a model that can predict low and heavy rainfall periods using the fuzzy inference system using satellite meteorological information. In order to predict low and high rainfall periods on a seasonal scale? 1-5 research objectives: One of the most important challenges that the country's water resources management system has faced in the last decade has been the considerable frequency and intensity of continuous droughts The increase in the frequency and severity of droughts due to the change in the pattern of precipitation, evaporation and transpiration can be one of the consequences of climate change on the annual hydrological cycle, so that the increase in temperature in winter causes the phenomenon of snow melting to advance and, as a result, changes in the time of maximum runoff in spring and decrease in runoff in summer. According to the conducted research, the warming that has occurred has an increasing trend and it seems to be accompanied by changes in atmospheric extreme conditions. On the other hand, the occurrence of severe droughts in many provinces of the country has been increasing in recent years, so that in recent years, the occurrence of drought has caused more damage to infrastructure facilities compared to the past. An increase or decrease in precipitation or a significant change in the occurrence of extreme cases that lead to floods or droughts can certainly have a significant impact on the planning of large countries. This research is a step to predict the quantitative status of precipitation in different seasons of the water year using meteorological information in a possible spatial and temporal domain. For this purpose, the possibility of using meteorological information in the areas affecting the rain-producing systems of the study area (catchment basins of dams in Tehran province) is investigated in order to predict periods of low rainfall and high rainfall on a seasonal scale. Meteorology in the study area has a statistical relationship with SPI changes.

Static precipitation values ??are assumed

Mutual information index (MI [4]) is considered a suitable tool for selecting forecast model inputs.

1-7 Thesis structure:

-Chapter one: (Objective and structure of the thesis)

In this chapter, mention the introduction One of the topics investigated in this thesis has been discussed and the general purpose of the topic plan has been determined. The second chapter: (research history) In this chapter of the research, an attempt is made to present the history of studies conducted in the field of drought and the use of the SPI index and the use of the fuzzy inference system in the sciences related to water engineering. Drought is provided. Quantitative measures of drought are mainly prepared based on the processing of a large amount of information on precipitation, snow, surface flows, etc., and lead to providing a general picture of the water dynamics process in a specific area. This information is usually presented in discrete and numerical form and in different scales. In addition, the topic of each of these profiles includes a specific range of information and the water cycle. In the following, the statistical relationship between the changes of atmospheric variables and SPI in the study area and how to choose the appropriate atmospheric components in order to predict the amount of the standardized precipitation index in each basin will be investigated. In the following, after reviewing the research history and presenting the basics of this method, the results of selecting important atmospheric parameters for each scenario will be mentioned. At the end of this chapter, a brief description of the basics of the fuzzy method, including membership functions, operators, if-then rules, fuzzification and non-fuzziness, and its methods were presented.

-Chapter 4: (case study)

This chapter is related to the studied area and the information used includes the basin area of ??Karaj, Mamlo, Letian, Taleghan and Lar dams. The statistical information used includes time series of meteorological parameters that are collected by satellite from the NCEP/NCAR bank and are available to the public.