پهنه بندي خطر خشكسالي مناطق خشك با استفاده از روشهاي دانش مبنا در محيط GIS (مطالعه موردي: حوضه شيطور، يزد)

Knowledge based drought risk zonation in arid regions using GIS (Case study: Sheitoor, Yazd)

خشكسالي تاثيرات منفي بسياري روي اقتصاد، محيط زيست و كشاورزي مي گذارد و خسارات سنگيني را براي قسمت هاي مختلف جهان به بار مي آورد، لذا تخمين و پيش بيني خشكسالي همواره يك مسئله مهم براي تصميم گيرندگان و برنامه ريزان بوده است. هدف از اين تحقيق پهنه بندي خطر خشكسالي در حوضه شيطور واقع در استان يزد با تلفيق داده هاي ماهواره اي، محيطي و هواشناسي مي باشد. بدين منظور از تصاوير ماهواره اي ALOS (تير 1388)، نقشه هاي توپوگرافي مقياس 1/25000 و آمار بارندگي، دما و تبخير ايستگاههاي هواشناسي استفاده شده است. در ابتدا لايه هاي اطلاعاتي عوامل موثر بر خشكسالي (شيب، جهت، ارتفاع، دما، بارندگي، تبخير، كاربري اراضي، تراكم شبكه آبراهه ها و درصد پوشش گياهي) تهيه و سپس با استفاده از منطق فازي و براساس حساسيت به خشكسالي استاندارد گرديد. از روش سلسله مراتبي جهت تعيين وزن هر پارامتر استفاده شد. به منظور تلفيق لايه هاي مذكور از دو روش شاخص وزني و اپراتورهاي مختلف منطق فازي و به منظور ارزيابي نتايج حاصله از شاخص عمودي خشكسالي اصلاح شده(MPDI) استفاده شده است. نتايج نشان داد كه از بين روشهاي مورد استفاده، روش شاخص وزني با بالاترين دقت (0/81=R2 ) مي تواند به منظور پهنه بندي خطر خشكسالي مورد استفاده قرار بگيرد.

Drought is a severe dilemma which influences different aspects of mankind’s life. Drought has a negative impact on economy, environment and agricultural sector and cause heavy damage and losses in many parts of the world. Therefore the quantitative estimation and prediction of drought phenomena has become an important issue for policy makers and the scientific community. In the last three decades, remote sensing has provided a useful tool for drought monitoring and a variety of remotely sensed drought indices based on vegetation indices, land surface temperature (LST) and albedo, have been developed. The main objective of this study was drought riskzonation in Sheitoor basin located in Yazd province by using satellite, climatology and environmental data. The data used in this research consist of ALOS (AVNIR) image collected on 18th July 2009, topographical maps (scale: 1/25000), rainfall, temperature and evaporation data which were obtained from meteorological stations. The Sheitoor basin is located in the central part of Iran. It covers a total area of 416 km2. The altitude varies in the region between 1844 and 2989 meters. Average annual rainfall in the study area is 171 mm and average annual temperature is 14 °C. Based on the Dumartens climate classification method, the climate of study area is cold arid. At first, the ALOS image was processed to obtain the TOA[1] radiance using gains provided in header file. Next the FLAASH algorithm was used to remove the influence of atmosphere and also for conversion of the TOA spectral radiance into ground reflectance. The image was registered to UTM Zone 40 (WGS 84) coordinates using 1:25000 scale digital maps, 17 control points, a polynomial (degree 2) equation and the nearest neighbor resling method. In the next step, effective parameters on drought including environmental factors (slope, aspect, height, land cover/use, stream density and vegetation fraction) and also climatic data (temperature, rainfall and evaporation) were mapped in GIS environment. The land cover/use map was extracted from satellite data using supervised classification algorithm. Vegetation fraction was also extracted from image using MSAVI1 index. The other parameters such as height, slope and aspect were produced using topographical maps (scale: 1/25000). Data standardization is a basic task in data analysis when several incomparable criteria are involved. To make comparable various data layers, the data layers which effect on drought were standardized using linear fuzzy. For exle,  drought severity decrease with an increase in altitude and areas having more height are less sensitive to drought, so maximum and minimum altitude were converted to 0 and 1. The AHP method was used to identify the weight of each parameter. Results of weighted layers showed maximum weight for land cover/use parameter due to the human intervention in natural ecosystems. Next, Index overlay and various fuzzy logic operators (Fuzzy Sum, Fuzzy product, Fuzzy OR, Fuzzy and) were used to model the drought risk. Drought change land cover, soil moisture and surface roughness, it also influences the exchange of energy and water between the vegetation, soil and the air. Thus, it may affect surface radiation, heat and water balance by changing surface biophysical factors such as the VI, albedo and LST. In general, with the development of a drought, the NDVI decreases, the albedo and surface temperature increase and the soil moisture decrease, provided that other factors are stable. Combinations of these parameters may provide a useful tool for better understanding of the spatiotemporal patterns of drought. Most of the drought indices presented in the last decades are based on the abovementioned parameters (especially NDVI, LST and Albedo). The retrieval of the surface albedo and the LST contains uncertainties rooted in the atmospheric correction of satellite data, decomposition of mixed pixel information, bidirectional reflectance distribution function (BRDF) modeling and the spectral remedy by a narrowband to broadband conversion. As a consequence, the final error associated with the extraction and quantifying of drought information would be magnified. On the other hand, calculating these indices need time series of satellite data which increase the time and the cost of processing. In Ghulam et al., 2007, the MPDI[2] as a real time index for drought monitoring based on vegetation fraction and soil moisture is presented. This index only needs one image to be calculated. In the present study, results assessed using MPDI. Final results indicated that the index overlay method can signify highrisk areas more accurately (R2=0.81) than the fuzzy operators.       [1] . Top Of Atmosphere [2] . Modified Perpendicular Drought Index