Hadi Ghafourian; Seyed Hossein Sanaei Nejad; Mahdi Jabbari Nowghabi
Abstract
Extended Abstract Introduction Due to the importance of precipitation in various aspects of human life, precipitation data are largely applicable in different fields of study. Therefore, accurate measurement of precipitation is considered to be crucialin various fields such as agriculture, water resources, ...
Read More
Extended Abstract Introduction Due to the importance of precipitation in various aspects of human life, precipitation data are largely applicable in different fields of study. Therefore, accurate measurement of precipitation is considered to be crucialin various fields such as agriculture, water resources, and industrymanagement. Due to the problems related to generalization of point precipitation to regional precipitation, alternative methods have been proposed forthe measurement of this variable. In many cases, short reference period, inadequate density of stations and poor quality of data collected from precipitation measurement networks have challenged the analysis of this climate variable. In order to overcome these problems, it is necessary to identify alternative sources, evaluate and use them to estimate the amount of precipitation. The present study primarily seeks to evaluate precipitation data from the TMPA and provide calibration data for arid, semi-arid, Mediterranean, humid, and very humid regions of Iran on a monthly scale. Materials and Methods In the present study, monthly precipitation data of 15 synoptic stations in 5 regions of Iran (arid, semi-arid, Mediterranean, humid and very humid) were selected as reference data and monthly precipitation data from the TMPA (3B43-v7) were corrected based on them. To ensure reliability of results and reduce errors,stations were selectedrandomly from 15 separate provinces with different topographic conditions. A 20-year reference period (1998-2017) was selected for the study. Collected satellite data have a monthly temporal resolution and a spatial resolution of 0.25 degrees covering 50th parallel south to 50th parallel north. Table 1 shows features of the selected stations and their corresponding pixels. Pre-processing included quality control, homogeneity test, and data accuracy test. Usinga long-term reference period of 20 years, different statistical criteria to evaluate satellite data and a correction relationindependent from ground data are among the advantages of this research. In this study, a more efficient method is used to determine errors and one of the most modern methods of calibration is also used. Followingthe application of log transformation and multiplicative model, monthly C parameter was calculated to rectify satellite data collected from different climates. Results were evaluated using R2 (Coefficient of Determination), MBE, MAE and RMSE. Results and Discussion Findings indicated that the distribution of initial data obtained from TMPA satellite in a monthly scale is similar to the distribution of pattern obtained from ground data (due to a correlation of above 75% (R2>0.6)). Satellite data collected from arid areas are usually overestimated, while data collected from humid areas are generally underestimated. However, determination coefficients (R2) of different climates show a strong correlation between these two sources of data. The initial TMPA data have estimated the monthly precipitation of Bam, Piranshahr and Abali stations with the least amount of error. The highest level of errors were obtained from Marivan, Bandar Anzali, and Koohrang stations. In other words, the highest level of errors have occurred in the very humid region. Calibration of TMPA data collected from the 5 different climates indicated that correction of TMPA monthly data would improve valuesestimated from satellite images. Mean bias error (MBE) was reduced by 88.7, 95.3, 68.4, 38.4 and 63.9 percentin arid, semi-arid, Mediterranean, humid and very humid climates, respectively. Values of the correction parameter (C) in the arid climate indicate that a reduction factor has been applied to rectify satellite data collected in each month of the year. In the semi-arid climate, reduction factorswere obtained for each months of the year. A reduction factor is also required to rectify data collected in the warmest months of the year (June, July, and August) in the Mediterranean climate. Due to the low precipitation of these months, overestimation seems reasonable in these areas. A reduction factor should also be applied in the humid climate for 6 months of spring and summer. Considering the precipitation rate in these areas, decreasing precipitation rate in these seasonsresults in overestimation and error. Due to the significant precipitationrate in the cold months of the year (autumn and winter), decreasing factorand underestimation are expected to occur. In the very humid climate, a reduction factor should be appliedin the warmest months of the year (June, July, and August). Due to the low precipitation rate of these months and higherfrequency of cloudy days, overestimation will be reasonablein these areas. Due to underestimationin the coldest months of the year (autumn and winter), coefficients higher than one must be corrected. Conclusion Based on the results, the model used to correct precipitation in all 5 climates have reduced errors in precipitation measurement. However, this improvement was more obvious in arid and semi-arid climates. Sincea large part of Iran havean arid and semiarid climate, this calibration model is highly recommended. In addition, the final correction model does not depend on ground data and thus, applying the calibration modelto areas other than the specified stations will also be useful.
