Nitrogen Diagnosis And Fertigation Demand In Pakchoi Based On Multi-source Information Fusion

The diagnosis of nitrogen?N?nutrition status based on phenotyping,combined the effects of soil,agricultural practice and environmental factors on crop N absorption,helps to realize N supplementation timely and appropriately,which is the basis of the precise management in vegetable cultivation.In the present study,the plant N model based on machine vision and machine learning method was established,and the effects of cumulative light and temperature function?LTF?,water potential,N application rate and cultivation system on yield and N uptake in pakchoi?Brassica campestris ssp.chinensis L.?were observed.The N requirement model for pakchoi based on crop phenotyping and environmental information was established as well.In addition,the N supply decision system of packchoi was designed based on the model,which could provide theoretical and technical support for the precise management of fertilizer application in pakchoi.The specific results are as follows:1.The model constrction between phenotyping and plant nitrogen concentrationA self-made phenotypic platform was used for image acquisition.The images was collected at different growth periods under four N treatments.The images were processed by OpenCV software and python language,and three phenotypic features including color,texture and morphology were extracted.Three machine learning algorithms including random forests?RF?,support vector regression?SVR?and neural network?NN?were used,and then three plant N models based on phenotyping in pakchoi was established.The results revealed that the root mean square error?RMSE?of the RF,SVR and NN models are 0.542%,0.538%and 5.465%,respectively,and the normalized mean absolute error?MAE?are 0.418%,0.398%and 5.115%,respectively,and R² are 0.924,0.922 and 0.086,respectively.It indicated that only the RF and SVR models could predict the plant N well.Among them,the RF model is more applicable and robust.The accuracy of RF model were 0.995 and 0.998 in seedling and harvesting period,respectively;the relative error?RE?was 5.52%and 4.78%,respectively.The accuracy of RF model was 0.991 and 1.037 under N surplus and N deficiency conditions,respectively.The RE was 4.01%and 9.96%,respectively.The N nutrition status of the pakchoi can be effectively diagnosed according to the phenotyping.2.The study on the threshold of nitrogen uptake in pakchoiThe critical N concentration(_?8?,the minimal concentration of total N in shoots that produce the maximum aerial dry matter?DM?,which is expected during the whole growth cycle.To clarify the N absorption characteristics in pkchoi,the critical N concentration model was established.The results showed that the critical N concentration value was 4.78%when the DW was 1.5 t/ha or less,if not,it decreased when DW increased,and the correspondence relations between them are:N₍?8?%=4.78×(2^-0.33.3.The cumulative photothermal effect on nitrogen uptake and plant growth in pakchoiTo reveal the relationship between growth,N uptake and the cumulative light and temperature function?LTF?in pakchoi,the varieties including‘Xinxiaqing No.5',‘Huawang'and‘Changzheng No.3'were used as materials,the fresh weight,dry weight and N uptake of shoots were determined under different experimental conditions.Three models of fresh weight,dry weight and N absorption based on LTF were developed,respectively,and the values of R²values were higher than 0.97.Taking the‘Huawang'as an example,the variation of model parameters with N application rate was analyzed.The results showed that the model parameters were significantly affected by the N application rates,and the R² values were 0.965,0.903 and0.859,respectively.At last,the three models of‘huawang'variety were verified by independent test data.The validation results showed that the slope?r?,RE,R² between the measured value and the simulated value were 0.946,15.94%and 0.965 in the fresh weight model,respectively.In dry weight model,these indices were 0.940,14.74%,and 0.966,respectively.In N absorption model,the r,RE and R² were 0.970,19.78%and 0.965,respectively.The models of fresh weight,dry weight and N uptake could be used to predict yield and N accumulation effectively.4.The interactive effects of water and nitrogen on nitrogen uptake in pakchoiTo assess the influence of drought stress on the growth and N nutrition status of pakchoi at different N levels,the changes in N accumulation and enzyme activities involved in N assimilation were investigated.The drought was induced by adding polyethylene glycol?PEG?under hydroponic culture conditions.Pakchoi seedlings were exposed to a modified nutrient solution with different N concentration?N1,N2,and N3 represent 2,9 and 18 mM NaNO₃,respectively?and osmotic potential?W1,W2 and W3 represent 0,60 and 120 g/L PEG 6000?in a full factorial,replicated randomized block design.A short time?seven days?of mild drought stress has no significant effect on plant water content,transpiration rate,and shoot biomass,which indicated short-term mild drought did not cause changes in N demand.However,moderate drought stress resulted in a significant decrease in plant water content,transpiration rate,shoot biomass and shoot N concentration.Moreover,increasing N availability considerably alleviated drought stress by promoting the accumulation of total free amino acids in the roots and root biomass,and improving the nitrate reductase?NR;EC 1.7.1.1?activitiy,which would reduce moisture limitations.The results suggested relatively high plant N is beneficial to enhance the adaptability of pakchoi to drought.5.The difference of nitrogen use efficiency,yield and quality in pakchoi between soil and substrate cultivationIn this study,the effects of N application rates on biomass,quality and N use efficiency for pakchoi under substrate cultivation were observed,and the results were compared with the data obtained in soil cultivation system.The results showed that the biomass,N accumulation,N absorption rate,N use efficiency and the content of soluble protein and flavonoid increased with the increase of N application rate under the appropriate N application?0-0.2g/kg?.Additionally the N absorption rate and the N use efficiency reached the highest level when the N application rate was 0.2 g/kg,and they were 33.19%and 41.74%,respectively.However,the fresh weight and dry weight of the shoot no longer increased,the N absorption rate and the N use efficiency decreased when the N application rate reached 0.2-0.375 g/kg.Under both substrate and soil cultivation systems,the relationships between yield,N use efficiency and N application rates showed a trend of a quadratic inverted parabolic line.When the N application rate was 129.2 kg/ha under the substrate cultivation system,the yield was 11096.36 kg/ha,and the N use efficiency reached the highest,37.39%.When the N application rate was 140.70kg/ha in the soil cultivation system,The yield was 18,729.46 kg/ha,and the highest use efficiency was 16.75%.6.Nitrogen demand model based on multi-source information fusion in pakchoiBased on the critical N concentration,phenotyping,temperature,photosynthetically active radiation and N application rates in growth period,a N demand model was established.The verification results showed that the predicted values have a good agreement with the measured values,the R² was 0.948,and the RMSE was 1.96 mg/plant.Meanwhile,combined with the N use efficiency,a N supply decision model and system based on multi-parameter fusion was proposed.In this study,texture-and geometry-related phenotyping combined with color properties were analyzed for their potential use in evaluating N concentration in pakchoi,and a high-precision diagnostic model of N nutrition was obtained,which was beneficial to the N nutrition monitoring during the whole growth cycle.The results also showed that the critical N concentration in shoot was constant when the biomass of pakchoi was less than 1.5 t/ha.Meanwhile,the N demand model based on the information of crop,environment and management was established as well,which was helpful to calculate the N application rate.In summary,our study provides theoretical support and technical reference for the precise management of N fertilizer,which helps to develop agricultural management system,reduce fertilizer input,and thus achieve safe production.