عنوان مقاله [English]
نویسنده [English]چکیده [English]
In some countries such as Iran with large deserts and sparse rain gauge network satellite-based precipitation estimates have great potential for a wide range of applications. However, satellite-based precipitation estimates are not operational for decision making applications because of a lack of information regarding the associated uncertainties and reliability of these products. Obviously these data sets like others have error. To reduce the error, precipitation data users must have enough information about error characteristics over different parts of the world. In this analysis, satellite-based precipitation estimate data derived from the “Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks” (PERSIANN) is compared with ground-based data over Iran. The PERSIANN gridded precipitation data with 0.25°×0.25° latitude/longitude spatial and 3-hourly temporal resolution are used for evaluation of the satellite product. In this study, satellite-based precipitation data were accumulated to daily and then monthly totals for comparison with accumulated monthly gauge precipitation for the period 2003-2007 over Iran (25°-40° N, 44°-63°E). This rain gauge network included more than 2000 rain gauge stations over Iran. The arithmetic monthly mean of gauge precipitations calculated for every pixel. A comparison of the mean annual precipitation maps between these two networks over Iran shows that for the study period the spatial variation pattern of annual precipitation is reasonably accurate in PERSIANN, but it underestimates the precipitation amount over most parts of Iran. In the next step, the correlation coefficient and the scatter plot of the monthly precipitation for PERSIANN and gauge data were calculated for grid cells which included at least one gauge. Then in an attempt to better evaluate precipitation and reduce the effect of gauge uncertainties, the study was limited to pixels, each of which contained at least five rain gauges. The results for both of them (pixels which include one rain gauge and pixels which include five rain gauges) show that this satellite product underestimates the monthly precipitation. The correlation coefficient between satellite and gauge monthly precipitation for all pixels which include at least one gauge is equal to 0.3035 (99% significant). The correlation coefficient for pixels which include at least five gauges is 0.2598 (99% significant). For a comparison between these two data sets in different topography and climates, the time series of satellite and ground monthly precipitation are plotted for five cells which included at least five rain gauges. A pixel located in mountainous area of Zagros which most of annual precipitation falls in winter and summer is almost a dry season. Another pixel is in the most humid region in the coast of Caspian Sea which precipitation falls in almost all months of the year. One of the selected pixel located in the desert area of the eastern part of country. The two others located at North West and North East of country which have wet cold climate. The results show that PERSIANN underestimates the precipitation in Zagros, and it also has a very poor performance over the Caspian coast. On the other hand, this satellite product overestimates the precipitation over dry eastern area.
بارانیزاده، ا.، م. ب.، بهیار، جوانمرد، س.، و عابدینی، ی.، 1390، صحتسنجی برآورد بارندگی الگوریتم ماهوارهای PERSIANN با دادههای بارش زمینی شبکهبندی شده APHRODITE در ایران: کنفرانس فیزیک ایران، فیزیک میانرشتهای، 2615-2618.
غضنفری مقدم، م. ص.، علیزاده، ا.، موسوی بایگی، س. م.، فرید حسینی، ع.، و بنایان اول، م.، 1390، مقایسه مدل PERSIANN با روشهای درونیابی بهمنظور کاربرد در تخمین مقادیر بارندگی روزانه (مطالعه موردی: خراسان شمالی): نشریه آب و خاک دانشگاه فردوسی مشهد (علوم و صنایع کشاورزی)، 25(1)، 207-215.