Prediction of climate variables by comparing the k-nearest neighbor method and MIROC5 outputs in an arid environment
The goal of this study was to compare the ability of the k-Nearest Neighbors (k-NN) approach and the downscaled output from the MIROC5 model for generating daily precipitation (mm), and daily maximum and minimum temperature (Tmax, and Tmin) (C) for an arid environment. For this study data from the easternmost province of Iran, South Khorasan was used for a period from 1986 to 2015. We also used an ensemble method to decrease the uncertainty of the k-NN approach. Although based on the initial evaluation MIROC5 had better results, we also used the output result of kNN alongside the MIROC5's data to generate future weather data for the period 2018 to 2047. NSE (Nash-Sutcliffe eficiency) between MIROC5 estimates and observed monthly Tmax ranged from 0.86 to 0.92, and 0.89-0.93 for Tmin over the evaluation period (2006-2015). K-NN performed less well, with NSE between k-NN estimates and observed Tmax ranging from 0.54 to 0.64, and from 0.75-0.78 for Tmin. The MIROC5 simulated precipitation was close to observed historical values (NSE between 0.06 and 0.07); the k-NN simulated precipitation was less accurate (NSE between -0.36 and -0.14). For these arid regions, the k-NN precipitation results compared poorly to the MIROC5 downscaling. MIROC5 predicts increases in monthly Tmin and Tmax in summer and autumn and decreases in winter and spring and decreases in winter monthly precipitation under RCP4.5 over the 2018-2047 period of this study. This study showed that the k-NN method cannot be used for generating future data in comparison to the outputs of the MIROC5 model for arid environments.
International Journal of Biometeorology
Evaluation of different gridded rainfall datasets for rainfed wheat yield prediction in an arid environment
The accuracy of daily output of satellite and reanalysis data is quite crucial for crop yield prediction. This study has evaluated the performance of APHRODITE (Asian Precipitation-Highly-Resolved Observational Data Integration Towards Evaluation), PERSIANN (Rainfall Estimation from Remotely Sensed Information using Artificial Neural Networks), TRMM (Tropical Rainfall Measuring Mission), and AgMERRA (The Modern-Era Retrospective Analysis for Research and Applications) precipitation products to apply as input data for CSM-CERES-Wheat crop growth simulation model to predict rainfed wheat yield. Daily precipitation output from various sources for 7 years (2000–2007) was obtained and compared with corresponding ground-observed precipitation data for 16 ground stations across the northeast of Iran. Comparisons of ground-observed daily precipitation with corresponding data recorded by different sources of datasets showed a root mean square error (RMSE) of less than 3.5 for all data. AgMERRA and APHRODITE showed the highest correlation (0.68 and 0.87) and index of agreement (d) values (0.79 and 0.89) with ground-observed data. When daily precipitation data were aggregated over periods of 10 days, the RMSE values, r, and d values increased (30, 0.8, and 0.7) for AgMERRA, APHRODITE, PERSIANN, and TRMM precipitation data sources. The simulations of rainfed wheat leaf area index (LAI) and dry matter using various precipitation data, coupled with solar radiation and temperature data from observed ones, illustrated typical LAI and dry matter shape across all stations. The average values of LAImax were 0.78, 0.77, 0.74, 0.70, and 0.69 using PERSIANN, AgMERRA, ground-observed precipitation data, APHRODITE, and TRMM. Rainfed wheat grain yield simulated by using AgMERRA and APHRODITE daily precipitation data was highly correlated (r2 ≥ 70) with those simulated using observed precipitation data. Therefore, gridded data have high potential to be used to supply lack of data and gaps in ground-observed precipitation data.
Ecological Indicators, Volume 90, July 2018, Pages 569–576
Prediction of effective climate change indicators using statistical downscaling approach and impact assessment on pearl millet (Pennisetum glaucum L.) yield through Genetic Algorithm in Punjab, Pakistan
Climate change involves long term changes in climate including increase in temperature, elevated CO2 and uneven distribution of rainfall quantity and periodicity. Though daily mean temperature (Tmean) is considered a widely useful index of assessment of climate change, averaging can obscure some of the variations expected in diurnal temperature range (DTR). The aim of this study was to evaluate the effectiveness of DTR relative to Tmean as a metric for predicting millet yield using a combination of a historical dataset (1980–2010) and two climate model (MIROC5 and GFDL) projections for 2017–2046 under RCP 4.5 and 8.5 across two different environments (arid, Layyah and semi-arid, Faisalabad) of Punjab, Pakistan. Provincial datasets of pearl millet yields were collected and checked for an empirical relationship between Tmean, DTR and crop yield. The mean of projections showed increasing DTR relative to baseline in both environments. Projected Tmax and Tmin were highly correlated (0.90–0.99) for both environments and climate models. MIROC5 predicted Tmax and Tmin well and GFDL performed efficiently in predicting precipitation by 2046. The data also showed more hot days in future decades and erractic rainfall pattern by 2046 in both environments. The Genetic Algorithm (GA) appeared to be a good approach to assess climate change impact on pearl millet yield in Punjab, Pakistan, predicting negative yield impacts (11–12%) due to future warming. We suggest broadening tests of this method to other cases around the world, with similar climate regimes.
Theoretical and Applied Climatology, pp 1–10
Predictive value of Keetch-Byram Drought Index for cereal yields in a semi-arid environment
Meteorological drought indices associated with soil moisture status have potential for varying applications including predictive power for crop yields estimation. The Keetch-Byram Drought Index (KBDI) was initially developed to estimate forest flammability, based on quantification of the moisture deficiency in upper soil layer as a function of daily precipitation and maximum air temperature. In this study, we characterized the utility of KBDI to accurately trace and monitor vegetation change and crop yield fluctuation in a semi-arid environment. It is tried to find any temporal association for both the 16-day MODIS-derived NDVI and KBDI from 2002 to 2012 and the correlation between KBDI and wheat and barley yield from 1984 to 2010. Correlation between KBDI and NDVI showed a general seasonal pattern with strongest correlation in mid-growing season, but this varied across study locations. Warmer locations with very sparse vegetation showed weaker association between KBDI and NDVI. Although a robust correlation between KBDI and winter cereal crop yield was not achieved based on winter (wet and cold season) data, spring cereal crop yield was correlated with KBDI.
Journal of Arid Land, December 2017, Volume 9, Issue 6, pp 797–809
Estimation of meteorological drought indices based on AgMERRA precipitation data and station-observed precipitation data
Meteorological drought is a natural hazard that can occur under all climatic regimes. Monitoring the drought is a vital and important part of predicting and analyzing drought impacts. Because no single index can represent all facets of meteorological drought, we took a multi-index approach for drought monitoring in this study. We assessed the ability of eight precipitation-based drought indices (SPI (Standardized Precipitation Index), PNI (Percent of Normal Index), DI (Deciles index), EDI (Effective drought index), CZI (China-Z index), MCZI (Modified CZI), RAI (Rainfall Anomaly Index), and ZSI (Z-score Index)) calculated from the station-observed precipitation data and the AgMERRA gridded precipitation data to assess historical drought events during the period 1987–2010 for the Kashafrood Basin of Iran. We also presented the Degree of Dryness Index (DDI) for comparing the intensities of different drought categories in each year of the study period (1987–2010). In general, the correlations among drought indices calculated from the AgMERRA precipitation data were higher than those derived from the station-observed precipitation data. All indices indicated the most severe droughts for the study period occurred in 2001 and 2008. Regardless of data input source, SPI, PNI, and DI were highly inter-correlated (R2=0.99). Furthermore, the higher correlations (R2=0.99) were also found between CZI and MCZI, and between ZSI and RAI. All indices were able to track drought intensity, but EDI and RAI showed higher DDI values compared with the other indices. Based on the strong correlation among drought indices derived from the AgMERRA precipitation data and from the station-observed precipitation data, we suggest that the AgMERRA precipitation data can be accepted to fill the gaps existed in the station-observed precipitation data in future studies in Iran. In addition, if tested by station-observed precipitation data, the AgMERRA precipitation data may be used for the data-lacking areas.
International Agrophysics, Volume 27, Issue 4 (Oct 2013)
Using satellite data for soil cation exchange capacity studies
This study was planned to examine the use of LandSat ETM+ images to develop a model for monitoring spatial variability of soil cation exchange capacity in a semi-arid area of Neyshaboor. 300 field data were collected from specific GPS registered points, 277 of which were error free, to be analysed in the soil laboratory.The statistical analysis showed that therewas a small R-Squared value, 0.17, when we used the whole data set. Visual interpretation of the graphs showed a trend among some of the data in the data set. Forty points were filtered based on the trends, and the statistical analysis was repeated for those data. It was discovered that the 40 series were more or less in the same environmental conditions; most of them were located in disturbed soils or abandoned lands with sparse vegetation cover. The soil was classified into high and medium salinity, with variable carbon (1.0 to 1.6%), heavy textured and with high silt and clay. Finally it was concluded that two different models could be fitted in the data based on their spatial dependency. The current models are able to explain spatial variability in almost 45 to 65% of the cases.