Precision Farming
H. Mohamadi-Monavar; S. Zibazadeh
Abstract
IntroductionRemote sensing methods for mapping farms and crops have been widely used in the last three decades. This method is applied to identify irrigated areas around the world (Alipour et al., 2014), although most of these studies are in areas with semi-arid climates and low rainfall or lack of rainfall ...
Read More
IntroductionRemote sensing methods for mapping farms and crops have been widely used in the last three decades. This method is applied to identify irrigated areas around the world (Alipour et al., 2014), although most of these studies are in areas with semi-arid climates and low rainfall or lack of rainfall which has a significant effect on the spectral characteristics of plants. In this study, Landsat 8 and MODIS satellite images were used to identify and separate two irrigated and rain-fed wheat farms in Hamadan province. Two algorithms of support vector machine (SVM) and minimum distance (MD) were used simultaneously to classify irrigated and rain-fed farms. In the next step, the area under cultivation of rain-fed and irrigated wheat was predicted in the whole cultivated area of Hamadan province. Finally, the cultivation area of rain-fed and irrigated crops was calculated in the province using Sentinel 3 satellite images based on the random forest algorithm in 2016.Materials and MethodsThe study area is Hamedan province, which is located between 59◦ 33′ and 49◦ 35′ north latitude and also from 34◦ 47′ to 34◦ 49′ east longitude of the Greenwich meridian. A 50-hectare rain-fed wheat farm in Amzajerd was used as a sample to extract the properties of rain-fed wheat. Also, irrigated indices were extracted from a 100-hectare irrigated wheat farm located in Kaboudrahang. Satellite images were applied to separate irrigated and rain-fed wheat in Hamadan province. NDVI, EVI and NDWI indices were extracted from 16-day images of Landsat, MODIS, and Sentinel 3 sensors in the five-year period (2015-2019). Google Earth Engine (GEE) system was the environment for performing image processing calculations and extracting indices and maps.Results and DiscussionThe NDVI and EVI of irrigated and rain-fed wheat farms were calculated in 2015-2019. A small peak was observed in the rain-fed and irrigated NDVI trend in November due to the early germination of wheat leaves in winter, and the larger peak in May and June showed the maximum greenness of irrigated and rain-fed wheat, respectively. The ascending or descending trend of NDVI / EVI had no constant slope. This can be due to changes in meteorological parameters, which sometimes cause a sudden increase or decrease in the values of these indices. Despite the non-linearity of the NDVI / EVI trend over time, the maximum greenness was recorded just a month before the wheat harvest, which was seen in the third decade of May to the first decade of June. One of the cases is the sharp drop of NDVI / EVI after its final peak, which was definitely due to yellowing wheat and harvesting. Since the distinction between rain-fed and irrigated crops was difficult only based on NDVI, NDWI was also used to determine the water content of wheat so that irrigated wheat could be identified. However, the difference between rain-fed and irrigated wheat in terms of NDWI spectral density was insignificant; the maximum and minimum occurrence times of NDWI and NDVI of rain-fed and irrigated wheat were chosen for their separation. In order to map the cultivation area, in addition to the MODIS sensor, Sentinel 3 was used due to its ability to detect chlorophyll accurately. Due to the fact that the imaging of the Sentinel 3 satellite started since 2016, the map of rain-fed and irrigated cultivation as well as the cultivation area and their separation was done based on the random forest algorithm in 2016.ConclusionThe results of this study showed that the appropriate method for distinguishing between rain-fed and irrigated wheat is the simultaneous use of several indices. Also, the greatest difference is in the maximum greenness, which happened almost one month before harvest. MD and SVM classification algorithms could distinguish irrigated and rain-fed wheat from other crops with 90% and 80% accuracy, respectively. Distinguished maps of irrigated and rain-fed crops based on the random forest algorithm were obtained using Sentinel 3 satellite imagery which can show the fertility of agricultural lands in the province.
N. Bagheri; H. Mohamadi-Monavar
Abstract
Fire Blight (FB) is the most destructive bacterial disease of pome fruit trees around the world. In recent years, spectrometry has been shown to be an accurate and real-time sensing technology for plant disease detection. So, the main objective of this research is early detecting FB of pear trees by ...
Read More
Fire Blight (FB) is the most destructive bacterial disease of pome fruit trees around the world. In recent years, spectrometry has been shown to be an accurate and real-time sensing technology for plant disease detection. So, the main objective of this research is early detecting FB of pear trees by using Visible-Near-infrared spectrometry. To get this goal, the reflectance spectra of healthy leaves (ND), non-symptomatic (NS), and symptomatic diseased leaves (SY) were captured in the visible–NIR spectral regions. In order to keep the important information of spectra and reduce the dimension of data, three linear and non-linear manifold-based learning techniques were applied such as, Principal Component Analysis (PCA), Sammon mapping and Multilayer auto-encoder (MAE). The output of manifold-based learning techniques was used as an input of the SIMCA (Soft independent modeling by class analogy) classification model to discriminate NS and ND leaves. Based on the results, the best classification accuracy obtained by using PCA on the 1st derivative spectra, with accuracy of 95.8%, 89.3%, and 91.6% for ND, NS, and SY samples, respectively. These results support the capability of manifold-based learning techniques for early detection of FB via spectrometry method.
B. Sepehr; H. Mohamadi-Monavar
Abstract
Introduction One of the most important factors in agricultural production is nitrogen which has a great impact on plant growing, yield performance and plant quality production. The optimum amount of nitrogen fertilizer is varied from fields to fields. There are some time consuming and costly ways to ...
Read More
Introduction One of the most important factors in agricultural production is nitrogen which has a great impact on plant growing, yield performance and plant quality production. The optimum amount of nitrogen fertilizer is varied from fields to fields. There are some time consuming and costly ways to measure the nitrogen content of plants or soil, which are inappropriate for extended field or for a long growing season. Fast and remote optical sensors calculate greenness of plant using reflectance or absorbance of light from green leaves. Measuring chlorophyll with SPAD managed the nitrogen requirement for maize, Poinsettia and Nagoya Red. Whereas SPAD was not a suitable choice for chlorophyll measurement at the end of growing period. Therefore, GreenSeeker was applied as a non-contact to record the NDVI of tomato’s and cucumber’s leaves. The purpose of this research was the evaluation of GreenSeeker ability to estimate nitrogen requirement and then the plant health. Materials and Methods The study was performed on Matin and Nahid cultivars of tomato and cucumber, respectively. The pots were 291 and filled with 3 kg sieved soil. The bottom layer of each pot was filled with stone for better drainage. Before planting, the soil was analyzed in order to define the ingredients. All pots put in the greenhouse with polycarbonate structure in two floors. Measurements were repeated every week with SPAD and GreensSeeker and fertigation was started 50 days after planting (DAP). In order to provide other nutrient elements, all pots got Humic-acid at 37DAP and the effect was measured in 43rd DAP. Fertigation was continued until 71st DAP and first, second and third treatments were supplemented with extra fertilizer to reach the amount of fertilizer to fifth treatment. To calculate Total Nitrogen (TN), the concentrations of nitrate-N and nitrite-N are determined and added to the total Kjeldahl nitrogen. Chlorophyll meter (SPAD) and GreenSeeker optical sensor have become available for site-specific and need-based N management in greenhouse. The GS was located at 60 cm above the plant and measured the average NDVI. This sensor has red and NIR diodes which reflect and absorb the spectra in 660±15nm and 770±15nm regions, respectively. The SPAD values were recorded by inserting the middle portion of the index leaf in the slit of SPAD meter. As well as, chlorophyll meter can confirm the GreenSeeker output (NDVI). GreenSeeker is a suitable optical sensor because it is not affected by light and temperature variation or wind intensity. Statistical analyses were performed on the pooled data of both tomato and cucumber using Statistical Product and Service Solutions (SPSS). Regression equations were fitted between fertilizer and the readings recorded with different gadgets at different growth stages. Results and Discussion Chlorophyll content and NDVI of tomato and cucumber increased during the growing stages except in 71st DAP for cucumber. The percentage of total nitrogen of 1st, 2nd and 3rd treatments were further than two others because of supplementary fertilizer. According to the Kjeldahl result of cucumber, the 3rd treatment had the lowest nitrogen accumulation in fruits. In addition, chlorophyll and NDVI of cucumber almost showed the increasing correlation by fertilizer enhancement while the opposite behavior was seen for tomato. That would be related to different fertilizer needs of them. The linear regression of fertilizer and reading NDVI of 2nd to 5th treatments were ascending. The number of increasing leaves was calculated in all pots every weeks as another studied element. Each pot had new grown leaves every weeks that was more or sometimes less than last weeks. However, accurate correlation coefficient was reported with NDVI in all treatments, whereas chlorophyll did not show a direct relation. Conclusion The result of the study confirmed the useful GreanSeeker as an accurate and fast technology for prediction of NDVI. Among different fertilizer treatments of cucumber, 3rd one showed the acceptable results. Since tomatoes did not reach to fertility stage, it would not possible to extract the best nitrogen fertilizer treatments. It is obvious that evaluation of pots in complete growth stages reach us to codify manual fertilization.
H. Mohamadi-Monavar
Abstract
Introduction Field management is a part of precision agriculture (PA) which has positive environmental and economic effects on quality of plant productions. Nitrogen needs of plant, depends on climate conditions and growing pattern. The optimum of nitrogen fertilizer is varied from fields to fields. ...
Read More
Introduction Field management is a part of precision agriculture (PA) which has positive environmental and economic effects on quality of plant productions. Nitrogen needs of plant, depends on climate conditions and growing pattern. The optimum of nitrogen fertilizer is varied from fields to fields. Nitrogen management causes uniform shape and size of potatoes, on the other hand decreases the inward and outward damages (Stark and Brown, 2003). Between different herbal indices, NDVI is the most common for monitoring greenness of plants. NDVI was calculated from reflectance in red and NIR bands (equation 1). Greenseeker (GS) is a suitable optical sensor because it is not affected by light and temperature variation or wind intensity. (1) In addition to GS, satellite image was used to evaluate the NDVI of studied potato field. Landsat 8 is the last satellite of this family with new sensors (operational land imager (OLI) and thermal infrared sensor (TIRs)) and additional spectral bands (deep blue invisible (430-450 nm) and shortwave infrared (1360-1390 nm). At the end, support vector regression (SVR) and principal component regression (PCR) or multi-linear regression (MLR) was applied to estimate RMSE and R2. The input of models was synoptic data, and NDVI extracted from GS or OLI. Materials and Methods The study was performed on marfona cultivar of potato field which located in Bahar city, Hamadan. The potato was planted early March and experiments were started after growing the first leaves. The soil texture in the experimented field was sandy loam soil to 75 cm depth. The territory (the southwest corner of the field) was fertigated by poultry manure with content 4.5% of N in order to put shortage of nitrogen down. Metrology station of Bahar city reported the maximum, minimum and average temperature, relative humidity, precipitation and wind velocity which were effective on NDVI variation. The GS was put at a height of 60 cm above the plant and the average of NDVI was obtained by three times measurement. This sensor has red and NIR diodes which reflect and absorb the spectra in 660±15nm and 770±15nm regions, respectively. GS and OLI were applied for measurement every 8 and 16 days, respectively. Satellite images were analyzed two times (30cm height of plant and hilling stage) during the growing. Although, climate changing were effective on NDVI then some image corrections were necessary. Geometric and atmospheric corrections were applied for removing the absorption and distribution error with dark object subtraction and FLAASH algorithm in ENVI 5.3 Software. In addition, GS is a nondestructive and contactless optic sensor which helps farmers to manage nitrogen because using laboratory method is not easy way for them. As well as, OLI provided accurate NDVI which support the accuracy of GS. Results and Discussion In order to correlate NDVI-GS and NDVI-OLI, the third parameter (INSEY) was explained. In season estimation of yield (INSEY) was estimated by dividing NDVI by days after planting (DAP). INSEY index is suitable to predict product potential performance. PCR and SVR methods in Matlab 2011b was used to calculated the relationship of INSEY and NDVI. Also, Red and NIR bands extracted from spectrometer (AvaSpec-ULS 2048- UV-VIS) in the 300-1100 nm region were used in order to support comparison of those sensors. Results showed that the reflectance spectra changed through the growing stage, which is logic because the size and number of leaves were increased and as a result the greenness was enhanced. NDVI calculated with spectra showed more accurate R2 for NDVI-GS (0.94) than NDVI-OLI (0.81). In addition, correlation coefficients of the SVR model between INSEY and NDVI were predicted 0.947 and 0.947 for the GS and OLI, respectively. Conclusion The result of the study confirmed the useful Greanseeker as an accurate and fast technology for prediction of NDVI. Among different regression methods, SVR showed the perfect results. Since the farm is a commercial one and not belong to the university, it would not possible to test different nitrogen fertilizer treatments. It is obvious that evaluation of field in different consecutive years helps us to codify manual fertilization.
H. Mohamadi-Monavar
Abstract
Introduction Precision agriculture (PA) is a technology that measures and manages within-field variability, such as physical and chemical properties of soil. The nondestructive and rapid VIS-NIR technology detected a significant correlation between reflectance spectra and the physical and chemical properties ...
Read More
Introduction Precision agriculture (PA) is a technology that measures and manages within-field variability, such as physical and chemical properties of soil. The nondestructive and rapid VIS-NIR technology detected a significant correlation between reflectance spectra and the physical and chemical properties of soil. On the other hand, quantitatively predict of soil factors such as nitrogen, carbon, cation exchange capacity and the amount of clay in precision farming is very important. The emphasis of this paper is comparing different techniques of choosing calibration samples such as randomly selected method, chemical data and also based on PCA. Since increasing the number of samples is usually time-consuming and costly, then in this study, the best sampling way -in available methods- was predicted for calibration models. In addition, the effect of sample size on the accuracy of the calibration and validation models was analyzed. Materials and Methods Two hundred and ten soil samples were collected from cultivated farm located in Avarzaman in Hamedan province, Iran. The crop rotation was mostly potato and wheat. Samples were collected from a depth of 20 cm above ground and passed through a 2 mm sieve and air dried at room temperature. Chemical analysis was performed in the soil science laboratory, faculty of agriculture engineering, Bu-ali Sina University, Hamadan, Iran. Two Spectrometer (AvaSpec-ULS 2048- UV-VIS) and (FT-NIR100N) were used to measure the spectral bands which cover the UV-Vis and NIR region (220-2200 nm). Each soil sample was uniformly tiled in a petri dish and was scanned 20 times. Then the pre-processing methods of multivariate scatter correction (MSC) and base line correction (BC) were applied on the raw signals using Unscrambler software. The samples were divided into two groups: one group for calibration 105 and the second group was used for validation. Each time, 15 samples were selected randomly and tested the accuracy of models, then other15 samples were added randomly to the previous set and it was done continuously. Finally, seven groups (15, 30... 105) were placed in each category. Results and Discussion All regression models on the whole pre-processed soil spectra were obtained in absorption mode. By increasing the number of samples in the calibration set of random group, RMSE was decreased from 0.2 to 0.13 nonlinearly. RMSE in the chemical test was also decreased almost linearly from 0.17 to 0.11. At the same time, R2 and RPD were increased from 0.46 to 0.72 and from 1.3 to 2.0 respectively. Finally, in categories based on PCA, the RMSE fell down almost linearly (0.19-0.13). Potassium prediction model with the least amount of R2 (0.48) and phosphorus with highest number of errors (RMSE = 5.28) were the weakest models in whole data. Other properties of the soil had a higher coefficient of determination (R2> 0.5). Therefore, prediction models had acceptable accuracy. At least 77, 105 and 105 samples are required for precise calibration model of OC, nitrogen and pH respectively. Due to the different conditions of farms, comparing these results with previous findings is very complex. Furthermore, model accuracy did not improve by increasing data of calibration models to the total number of samples. While in previous studies, more precise model was calibrated by considering the entire data sets. Among all factors of soil, acidity has little dependence on the other soil properties. The pH modeling is also confirmed by Moros (2009) however, the more error was reported here. There is no certain pH range in the NIR spectra, and usually it is distinguishable from the other properties of the soil (Kuang and Mouazen, 2011). Conclusion Spectroscopic methods exhibited good potential for detecting soil properties. MSC and BC can effectively remove irrelevant information to improve prediction accuracy. Using different methods to select numbers of data for the calibration models presented similar results, but in the meantime PCA technique provided the best answer. Supplementary, the ever-increasing number of data does not always improve modeling accuracy. It is better to choose numbers of data according to principal components (PCs) of PCA to obtain acceptable answer. It must be noted that every crops requires a specific soil and nutrients, so it is necessary to develop models for other soil properties.
H. Mohamadi-Monavar; R. Alimardani; M. Omid
Abstract
Agricultural sector experiences the application of automated systems since two decades ago. These systems are applied to harvest fruits in agriculture. Computer vision is one of the technologies that are most widely used in food industries and agriculture. In this paper, an automated system based on ...
Read More
Agricultural sector experiences the application of automated systems since two decades ago. These systems are applied to harvest fruits in agriculture. Computer vision is one of the technologies that are most widely used in food industries and agriculture. In this paper, an automated system based on computer vision for harvesting greenhouse tomatoes is presented. A CCD camera takes images from workspace and tomatoes with over 50 percent ripeness are detected through an image processing algorithm. In this research three color spaces including RGB, HSI and YCbCr and three algorithms including threshold recognition, curvature of the image and red/green ratio were used in order to identify the ripe tomatoes from background under natural illumination. The average error of threshold recognition, red/green ratio and curvature of the image algorithms were 11.82%, 10.03% and 7.95% in HSI, RGB and YCbCr color spaces, respectively. Therefore, the YCbCr color space and curvature of the image algorithm were identified as the most suitable for recognizing fruits under natural illumination condition.
