ارزیابی داده‌های شبکه‌بندی شده Agmerra در شبیه‌سازی عملکرد و نیاز آبی گندم دیم در استان خراسان رضوی

شبیه‌سازی با استفاده از مدل‌های گیاهی نیازمند داده‌های بلند مدت و با کیفیت آب و هوایی است. در بسیاری از مناطق این داده‌ها با کیفیت مطلوب و یا دوره آماری مناسب در دسترس نمی‌باشند. لذا این مطالعه با هدف ارزیابی داده‌های agmerra در شبیه‌سازی نیاز آبی و عملکرد گندم دیم در استان خراسان رضوی به اجرا درآمد. در این پژوهش از داده‌های روزانه ایستگاه‌های سینوپتیک تربت ‌جام، تربت‌ حیدریه، سبزوار، سرخس، قوچان، کاشمر، گناباد، نیشابور و مشهد استفاده شد. داده های روزانه agmerra از پایگاه داده‌های سازمان فضایی امریکا جمع‌آوری و سپس با داده‌های مشاهداتی ایستگاه‌های سینوپتیک مورد مقایسه قرار گرفتند. جهت شبیه‌سازی نیاز آبی و عملکرد گندم از مدل کراپ‌وات و csmcereswheat استفاده شد. نتایج نشان داد که تشعشع خورشیدی، دمای حداقل و حداکثر agmerra در تمامی مناطق همبستگی (r^2 <0/7) و توافق خوبی (nrmse < %30) با داده‌های مشاهداتی نشان داد. اما سرعت باد و بارندگی روزانه agmerra در توافق با مقادیر مشاهداتی متناظر نبود، با این وجود استفاده از مجموع بارندگی 15 روز سبب بهبود وضعیت توافق و همبستگی مشاهده شده در تمامی مناطق گردید. ضریب تغییرات نیاز آبی و عملکرد شبیه‌سازی شده با استفاده از داده‌های agmerra در تمامی مناطق به جز تربت ‌جام، تربت‌ حیدریه و گناباد برای نیاز آبی و مشهد، کاشمر و قوچان برای عملکرد نزدیک به (بین 5- تا 5+ درصد) ضریب تغییرات نیاز آبی و عملکرد شبیه‌سازی شده با استفاده از داده‌های مشاهداتی بود. با این وجود انحراف میانگین بلند مدت نیاز آبی و عملکرد شبیه‌سازی شده با استفاده از داده‌های agmerra در تمامی مناطق به جز تربت‌ حیدریه و گناباد برای نیاز آبی در بازه 10 تا 10+ درصد میانگین بلند مدت نیاز آبی و عملکرد شبیه‌سازی شده با استفاده از داده‌های مشاهداتی قرار داشت. با توجه به نتایج حاصله می‌توان از داده‌های agmerra جهت برآورد میانگین بلند مدت نیاز آبی و عملکرد گندم دیم در منطقه مورد مطالعه استفاده نمود. با این وجود این مجموعه داده جهت شبیه‌سازی دقیق نیاز آبی و عملکرد در یک سال خاص زیاد قابل اعتماد نیست.

Evaluation of Grided AgMERRA Weather Data for Simulation of Water Requirement and Yield of Rainfed Wheat in Khorasan Razavi Province

Introduction: Estimating crop water requirement, crop yield and their temporal and spatial variability using crop simulation models are essential for analysis of food security, assessing impact of current and future climates on crop yield and yield gap analysis, however it requires longterm historical daily weather data to obtain robust predictions. Depending on the degree of weather variability among years, at least 10–20 years of daily weather data are necessary for reliable estimates of crop yield and its interannual variability. In many regions where crops are grown, daily weather data of sufficient quality and duration are not available. In this way, gridded weather databases with complete terrestrial coverage are available which require comprehensive validation before any application. These weather databases typically derived from global circulation computer models, interpolated weather station data or remotely sensed surface data from satellites. The aims of this study were to evaluate differences between grided AgMERRA weather data and ground observed data and quantify the impact of such differences on simulated water requirement and yield of rainfed wheat at 9 different locations in Khorasan Razavi province. Materials and Methods: AgMERRA dataset (NASA’s ModernEra Retrospective analysis for Research and Applications) was selected as the girded weather data source for use in this study because it is publically accessible. We evaluated AgMERRA weather data against observed weather data (OWD) from 9 meteorological stations (Torbat Jam, Torbat Heydarieh, Sabzevar, Sarakhs, Ghoochan, Kashmar, Gonabad, Mashhad, and Neyshabour) in Khorasan Razavi province. For each weather variable (solar radiation, maximum temperature, minimum temperature, precipitation, and wind speed), the degree of correlation and agreement between OWD and AgMERRA data for the grid cell in which weather stations were located were evaluated. The intercept (b), slope (m), and coefficient of determination (r2) of the linear regression were calculated to determine the strength and bias of the relationship, while the root mean square error (RMSE) and normalized root mean square error (NRMSE) were computed to measure the degree of agreement between data sources. Crop water requirement or actual crop evapotranspiration (ETc) under standard condition was computed using CROPWAT 8.0. The CSMCERESWheat (Cropping System ModelCrop Environment Resource SynthesisWheat) model, included in the Decision Support System for Agrotechnology Transfer (DSSAT v4.6) software package was used to calculate rainfed wheat yield. For each location in this study, rainfed wheat grain yield and water requirement were simulated using groundobserved and AgMERRA weather data and outputs were compared with each other. Results and Discussion: The results of this study showed that AgMERRA daily maximum and minimum temperature and solar radiation showed strong correlation and good agreement with data from ground weather stations. AgMERRA daily precipitation had low correlation and good agreement (mean r2= 0.34, RMSE= 2.25 mm and NRMSE= 4.94% across the 9 locations) with OWD daily values, but correlation with 15day precipitation totals were much better (mean r2 >0.7 across the 9 locations). There was reasonable agreement between a number of observed dry and wet days with AgMERRA compared to OWD. Results indicated that coefficient of variation of simulated water requirement and yield using AgMERRA weather data was remarkably similar to the degree of variation observed in simulated water requirement and yield using OWD at all locations (distribution of CVs in simulated water requirement and yield using AgMERRA weather data were within ±5% of the CV calculated for simulated water requirement and yield using observed weather data) except Torbat Jam, Torbat Heydarieh and Gonabad for water requirement and Mashhad, Kashmar and Ghoochan for yield. There was good agreement between longterm average yield simulated with AgMERRA weather data and longterm average yield simulated using observed weather data. For example, the distribution of simulated yields using AgMERRA data was within 10% of the simulated yields using observed data at all locations. Using AgMERRA weather data resulted in simulated crop water requirement that were not in close agreement with crop water requirement simulated with ground station data at two location including Gonabad and Torbat Heydarieh. Conclusions: These results supported the use of uncorrected AgMERRA daily maximum and minimum temperature and solar radiation in areas that their weather stations only have a few years of daily weather records available or areas without weather station. Considering the advantage of continuous coverage and availability, use of AgMERRA dataset appears to be a promising option for simulation of longterm average yield and water requirement, as well as for assessing impact of climate change on crop production and also estimating the magnitude of existing gaps between yield potential and current average farm yield in Khorasan Razavi province. But they are not very reliable for accurate simulation of water requirement and yield in a specific year and estimate their interannual variation.