Timely and accurate forecasting of crop yields is critical for food management and trade.  However, only limited research has explored the impact of integrating crop phenotypic parameters (CPPs) with unmanned aerial vehicle (UAV) data across different phenological stages on maize yield prediction.  The extent to which multi-temporal data enhances the accuracy and reliability of yield projections compared to mono-temporal data has yet to be systematically investigated.  To attain the proper balance between accuracy and cost in crop yield estimation, this study proposed a structured framework for identifying the optimal phenological periods for summer maize yield prediction using UAV-based multispectral data.  Three classical methods of custom mean decrease accuracy (C-MDA), optimal parameters-based geographical detector (OPGD), and grey relational analysis (GRA) were first used to sort and screen both the CPPs and vegetation indices (VIs) derived from UAV-based information over six growth stages.  Ridge regression models based on multi-temporal data combinations and mono-temporal data were established separately, and their performance in yield prediction were compared to identify the optimal phenological stages and the corresponding key factors.  Our results showed that C-MDA was much better at factor screening and ranking compared to OPGD and GRA.  The green normalized difference vegetation index (GNDVI), normalized difference vegetation index (NDVI), and normalized difference red edge index (NDRE) emerged as the top-performing VIs, while the leaf area index (LAI) and above ground biomass (AGB) proved to be the most effective CPPs.  When predicting yield using only mono-temporal data, the dough stage delivered the highest predictive accuracy (R²=0.871, RMSE=0.407 t ha^–1), while the tasseling stage was the earliest that achieved yield estimates with acceptable precision (R²=0.810, RMSE=0.493 t ha^–1).  In contrast, the integration of UAV data from different crop growth stages markedly enhanced the accuracy of yield estimation.  Combinations of data from the tasseling, silking, and dough stages were recommended as the best option (R²=0.942, RMSE=0.291 t ha^–1).  These findings indicate that the precise estimation of maize yields in smallholder fields may be attainable, and present both substantial theoretical insights and practical benefits for the advancement of precision agriculture.

The irrigation districts of northern China face issues such as water scarcity, inability to effectively utilize flood resources, and groundwater overexploitation.  In view of these challenges, this study proposes a new concept of deep storage irrigation through flood resources utilization.  However, whether deep storage irrigation can recharge deep soil moisture and sustain crop production still requires further study.  A two-year field experiment was conducted on summer maize in the Guanzhong Plain with five soil wetting layer depths (T1: 60 cm; T2: 90 cm; T3: 120 cm; T4: 150 cm; T5: 180 cm) and soil saturation moisture content as the irrigation upper limit.  The results presented that the ranges of deep soil moisture recharge in the 100–200 cm soil profile (SMS_100–200) was 73.34–267.42 and 0–150.03 mm in 2021 (wet season) and 2022 (normal season).  When the effective precipitation and irrigation exceeded 390 mm, the SMS_100–200 began to linearly increase.  The highest grain yield (GY) were observed at T2 and T3 treatments in 2021 (11.44 t ha⁻¹) and 2022 (11.25 t ha⁻¹), respectively.  The maize GY of T4 in 2021 and T5 in 2022 were only 3.9 and 5.7% lower than the maximize GY, respectively.  However, the SMS_100–200 for T4 and T5 were 2.4 and 5.0 times that of T2 and T3 treatments in 2021 and 2022, respectively.  Overall, the further increase in irrigation amounts induced only a slight decrease in grain yield, but it significantly increased deep soil moisture recharge.  Therefore, the deep storage irrigation breaks through the traditional idea of water-saving irrigation with limited water resources, which can be utilized as an effective alternative to address the issues of water scarcity, low flood resources utilization, and groundwater level declines in the irrigation districts of northern China.

Winter wheat is a key grain crop in China, with its leaf area index (LAI) serving as a vital indicator for growth assessment in precision agriculture. While UAV-based remote sensing and the PROSAIL model are widely used for LAI estimation, their accuracy under plastic mulch is limited due to spectral interference from mulch. Based on field measurements from winter wheat fields in Yangling, Shaanxi Province, this study developed a linear-spectral hybrid model (LSHM) by integrating measured soil and mulch reflectance into the PROSAIL model to improve LAI inversion for mulched winter wheat. UAV-collected multi-temporal images and field LAI measurements were used, and the model was coupled with random forest (RF) and LASSO variable selection to enhance accuracy. The LSHM significantly reduced reflectance differences in PROSAIL simulations by incorporating measured soil and mulch spectra, achieving R² improvements average of 16.5% across all growth stages compared to the standard PROSAIL model. Furthermore, the RF-LASSO optimized combination of band reflectance (BR) and vegetation indices (VIs) enhanced model stability, demonstrating superior performance (R²=0.83, RMSE=0.35, MSE=0.12 at heading and R²=0.71, RMSE=0.50, MSE=0.25 at filling). These results demonstrate that the hybrid PROSAIL model holds significant promise for estimating LAI inversion in mulched winter wheat, providing a robust approach for phenotyping crop traits in ridge-furrow mulched systems.