Sensor based nutrient management in Tomato (Solanum lycopersicum)

Abstract

The study entitled ‘Sensor based nutrient management in tomato (Solanum lycopersicum)’ has been carried out at the Department of Soil Science and Agricultural Chemistry, College of Agriculture, Vellayani, during 2023-2025. The objective of the study was to compare soil test values obtained through different sensors and laboratory analysis; Yield assessment to analyse the efficiency of the sensor using tomato as the

test crop.

The study was carried out in two parts. The first part focused on comparing nutrient analysis across different cropping systems using sensor-based measurements and conventional laboratory analysis. Surface soil samples (0-15 cm) were collected from four land-use systems, , vegetable-based cropping system, coconut-based mixed farming, homestead farming, and fallow land within AEU 8. Five samples were collected from each system using the standard soil sampling procedure, with one composite sample taken for every 20 cents of area. Two portable sensors were deployed to measure soil nutrients at each location. Sensor 1 (Modbus) provides primary nutrient estimates (N, P, K) using an electrical-conductivity-based sensing principle. Sensor 2 (NutRE) measures primary (N, P, K), secondary nutrients (Ca, Mg, S), and selected micronutrients (Fe, Zn, Cu) through an ion-sensing mechanism, with further technical

details withheld due to a pending patent.

For field measurements, six random sampling points were selected within each 20-cent area, matching the locations used for conventional soil sampling. Modbus readings were taken at two points and averaged, and this procedure was repeated across the remaining 3-6 randomly selected spots within each location. The same sampling approach was applied for the NutRE sensor. The mean nutrient values obtained from both sensors were then compared with corresponding laboratory measurements. The Modbus sensor (Sensor 1) showed no significant difference (p > 0.05) for P under all land-use systems and across all computed mean levels (M2—M6), indicating close agreement with laboratory values for these parameters. However, N and K showed

significant deviation (p = 0.00), suggesting the need for calibration for these nutrients.

In contrast, the NutRE sensor demonstrated non-significant differences (p > 0.05) for all measured nutrients (N, P, K, Ca, Mg, S, Fe, Zn, Cu) across all land-use systems and mean levels, reflecting strong alignment with laboratory analysis. Variation in the number of readings used for mean calculation (M2—M6) did not significantly influence sensor accuracy for either sensor. Therefore, adopting the minimum mean level (M2) was adequate for accurate nutrient estimation across all systems, improving operational

efficiency by reducing sampling time and field effort.

The second part involved a field experiment that assessed sensor efficiency through tomato (cv.Anagha) yield evaluation, where the performance of both sensors was compared against soil test-based nutrient management. The experiment consisted of five treatments: T: (STB) — Split application of N P K as per soil test data (laboratory analysis). Tz (MDS-Nb) — Split application of N P K as per sensor | reading at 15 days interval up to beginning of harvest (Modbus sensor without basal). Ts (NutRE-Nb) — Split application of N P K as per sensor 2 reading at 15 days interval up to beginning of harvest (NutRE sensor without basal). Ts (MDS-B) — NPK application as basal dose based on Ist sensor data and further split up at 15 days interval (Modbus sensor with basal). Ts (NutRE-B) — NPK application as basal dose based on 2nd sensor data and further split up at 15 days interval (NutRE sensor with basal).

Biometric observations revealed, at harvest, T: recorded the highest plant height (61.17 cm), while Ts produced a significantly greater number of branches (4.81). Leaf Area Index was highest under T; (2.59). With respect to yield attributes, T: produced a significantly greater number of fruits per plant (34.81), and the maximum fruit yield per plant (1.10 kg) while Ts (35.65g) produced highest fruit weight. Post-harvest soil analysis revealed significantly higher available N (210.12 kg ha‘) and K (207.20 kg ha"') under Ti, whereas Ts recorded the highest available P (85.20 kg ha‘). Plant nutrient analysis conducted at 15 and 30 DAT showed that Ts consistently recorded maximum NPK content during early crop growth, while at harvest, N and K contents were highest in T, and P content was highest in Ts. A similar trend was observed in nutrient uptake. Economic analysis showed that Ti yielded the highest gross income (Rs. 720,254.80) and B:C ratio (1.81), closely followed by Ts.

T-test and Absolute Error Percentage (AEP) results showed that NutRE sensor accurately estimated all nutrients, with non-significant differences from laboratory values. Modbus sensor was reliable for phosphorus, while nitrogen and potassium required calibration to match laboratory standards. Field experiments revealed that soil test-based nutrient management T: (STB) achieved the highest yield, which was statistically comparable to the calibrated NutRE-based treatment Ts (NutRE-B). These results confirm that sensor-based nutrient management becomes effective only when

supported by proper calibration and the inclusion of basal nutrient doses.