Integrating machine learning and empirical evapotranspiration modeling with DSSAT: Implications for agricultural water management.

The availability of accurate reference evapotranspiration (ETo) data is crucial for developing decision support systems for optimal water resource management. This study aimed to evaluate the accuracy of three empirical models (Hargreaves-Samani (HS), Priestly-Taylor (PT), and Turc (TU)) and three machine learning models (Multiple linear regression (LR), Random Forest (RF), and Artificial Neural Network (NN)) in estimating daily ETo compared to the Penman-Monteith FAO-56 (PM) model. Long-term data from 42 weather stations in Florida were used. Moreover, the effect of ETo model selection on sweet corn irrigation water use was investigated by integrating simulated ETo data from empirical and ML models using the Decision Support System for Agrotechnology Transfer (DSSAT) model at two locations (Citra and Homestead) in Florida. Furthermore, a linear bias correction calibration technique was employed to improve the performance of empirical models. Results were consistent in that the NN and RF models outperformed the empirical models. The empirical models tended to underestimate and overestimate small and high daily ETo values, respectively, with the HS model exhibiting the least accuracy. However, calibrated PT and TU models performed comparably to the ML models. Results also revealed that using an inappropriate ETo model could lead to over-irrigation by up to 54 mm during a single crop season. Overall, ML models have proven reliable alternatives to the PM model, especially in regions with access to long-term data due to their site-independent performance. In areas without long-term data for ML model training and testing, calibrating empirical models is viable, but site-specific calibration is needed. It is important to highlight that distinct plant species exhibit varying transpiration characteristics and, consequently, have different water requirements. These differences play a pivotal role in shaping the overall impact of ETo models on crop water use.

Towards sustainable watershed-based landscape restoration in degraded drylands: Perceived benefits and innovative pathways learnt from project-based interventions in Ethiopia.

Land degradation is one of the contemporary environmental challenges affecting regions inhabited by over one-third of the global population. In response to land degradation, restoration of degraded landscapes through area closure has been implemented through government and bilateral organizations for the last three decades in Ethiopia. Objectives of this study were to: i) explore the effects of landscape restoration on vegetation cover; ii) identify the perceived benefits to local communities; and 3) synthesize the lessons learnt on communities' willingness to sustain the restored landscapes. The study was conducted in project-supported restoration areas including the Dimitu and Kelisa watersheds representing the central rift valley dry lands and the Gola Gagura watershed representing the eastern dry land areas around Dire Dawa. The temporal changes in land use and land cover due to area closure integrated with physical and biological soil and water conservation measures were detected using GIS/Remote sensing techniques. Moreover, eighty-eight rural households were interviewed. The results of the study revealed that landscape restoration activities such as area closure integrated with physical soil and water conservation, and planting of trees and shrubs contributed to the significant changes in land covers of the watersheds in 3-5 years. Hence, barren lands were reduced by 35-100% while there were significant increases in forest lands (15%), woody grasslands (247-785%), and bushlands (78-140%). More than 90% of the respondents in the Dimitu and Gola Gagura watersheds verified that the landscape restoration activities improved vegetation cover and ecosystem services, reduced erosion, and increased incomes. A great majority of farm households (63-100%) expressed their willingness to contribute to different forms of landscape restoration interventions. Encroachment of livestock to closed area, shortage of finance, and the growing number of wild animals in closed area were the perceived challenges. Proper planning and implementation of integrated interventions, creating local watershed user associations, ensuring appropriate benefit-sharing and implementing innovative pathways to reconcile the tradeoffs could be considered to scale up interventions and address potential conflicts of interest.

Modeling adaptation of sorghum in Ethiopia with APSIM-opportunities with G×E×M.

Sorghum is an important food and feed crop in the dry lowland areas of Ethiopia. Farmers grow both early-sown long-duration landraces and late-sown short-duration improved varieties. Because timing and intensity of drought stress can vary in space and time, an understanding of major traits (G), environments (E), management (M), and their interactions (G×E×M) is needed to optimize grain and forage yield given the limited available resources. Crop simulation modeling can provide insights into these complex G×E×M interactions and be used to identify possible avenues for adaptation to prevalent drought patterns in Ethiopia. In a previous study predictive phenology models were developed for a range of Ethiopian germplasm. In this study, the aims were to (1) further parameterize and validate the APSIM-sorghum model for crop growth and yield of Ethiopian germplasm, and (2) quantify by simulation the productivity-risk trade-offs associated with early vs late sowing strategies in the dry lowlands of Ethiopia. Field experiments involving Ethiopian germplasm with contrasting phenology and height were conducted under well-watered (Melkassa) and water-limited (Miesso) conditions and crop development, growth and yield measured. Soil characterization and weather records at the experimental sites, combined with model parameterization, enabled testing of the APSIM-sorghum model, which showed good correspondence between simulated and observed data. The simulated productivity for the Ethiopian dry lowlands environments showed trade-offs between biomass and grain yield for early and late sowing strategies. The late sowing strategy tended to produce less biomass except in poor seasons, whereas it tended to produce greater grain yield except in very good seasons. This study exemplified the systems approach to identifying traits and management options needed to quantify the production-risk trade-offs associated with crop adaptation in the Ethiopian dry lowlands and further exemplifies the general robustness of the sorghum model in APSIM for this task.