Soil Sensor Monitoring via Mobile App

Implementing Soil Sensor Monitoring via Mobile Applications

A soil sensor is typically Modbus RTU or SDI-12 on wired interface, or LoRaWAN/NB-IoT wireless. Popular models: Sentek Drill & Drop (SDI-12), Vegetronix VH400 (analog 0-3V), TEROS 12 (SDI-12), Decagon 5TM — all with different output formats. Mobile applications receive this data through IoT gateway or LoRaWAN Network Server — never directly.

What the Sensor Measures and How to Interpret It

Three primary soil parameters:

VWC (Volumetric Water Content) — soil moisture percentage. Range 0-100%; in practice for most soils, working range is 10-40%. Sensor measures soil dielectric permittivity, conversion to VWC uses Topp formula or manufacturer calibration data for specific soil type.

EC (Electrical Conductivity) — conductivity in mS/cm. Indicates salinity and nutrient concentration. Normal for most crops: 0.5-2.0 mS/cm. Above 4 mS/cm — plant stress.

Soil Temperature — critical for seed germination (most crops won't germinate below 8-10°C) and microbial activity.

Applications should show not raw values but agronomic interpretation. "Moisture 19% at FC=45% for sandy loam" — this is dry. Without soil type and field capacity (FC) context, the number is meaningless.

Data Retrieval: LoRaWAN via ChirpStack

ChirpStack is open-source LoRaWAN Network and Application Server. REST API and gRPC interface:

// Kotlin, Retrofit for ChirpStack API
interface ChirpStackApi {
    @GET("api/devices/{devEui}/events")
    suspend fun getDeviceEvents(
        @Header("Grpc-Metadata-Authorization") token: String,
        @Path("devEui") devEui: String,
        @Query("limit") limit: Int = 100,
    ): DeviceEventsResponse
}

data class DeviceEvent(
    val publishedAt: String,
    val data: String,  // Base64-encoded payload
    val rxInfo: List<RxInfo>,
)

fun decodePayload(base64Data: String): SoilReading {
    val bytes = Base64.decode(base64Data, Base64.DEFAULT)
    // Decoding depends on sensor manufacturer encoding
    // TEROS 12 Cayenne LPP format:
    val vwc = ((bytes[1].toInt() and 0xFF) shl 8 or (bytes[2].toInt() and 0xFF)) / 100.0
    val temp = ((bytes[4].toInt() and 0xFF) shl 8 or (bytes[5].toInt() and 0xFF)) / 100.0 - 40
    val ec = ((bytes[7].toInt() and 0xFF) shl 8 or (bytes[8].toInt() and 0xFF)) / 100.0
    return SoilReading(vwc = vwc, temperature = temp, electricalConductivity = ec)
}

For real-time via MQTT — ChirpStack publishes events to topics like application/{appId}/device/{devEui}/event/up.

Dashboard: Multiple Sensors Per Field

Standard setup — 3-5 sensors at depth horizons (10, 30, 60, 90 cm) at one measurement point. Dashboard shows moisture profile by depth — vertical bar chart is more efficient than a list:

Widget buildMoistureProfile(List<SoilLayerReading> layers) {
  return Padding(
    padding: const EdgeInsets.all(16),
    child: Row(
      children: [
        // Depth axis
        Column(
          mainAxisAlignment: MainAxisAlignment.spaceBetween,
          children: layers.map((l) => Text('${l.depthCm} cm')).toList(),
        ),
        const SizedBox(width: 8),
        Expanded(
          child: Column(
            children: layers.map((layer) {
              final isLow = layer.vwc < layer.fieldCapacity * 0.5;
              return Container(
                margin: const EdgeInsets.symmetric(vertical: 2),
                height: 32,
                child: LinearProgressIndicator(
                  value: layer.vwc / 60.0,  // normalize to 60% max
                  backgroundColor: Colors.grey.shade200,
                  color: isLow ? Colors.orange : Colors.blue,
                ),
              );
            }).toList(),
          ),
        ),
      ],
    ),
  );
}

Trends and Irrigation Threshold

Main analytical function — show when moisture dropped to irrigation threshold and when returned to target level after watering. This helps agronomist confirm irrigation system worked correctly.

Graph from fl_chart with horizontal threshold line:

LineChartData buildTrendChart(List<SoilReading> readings, double threshold) {
  return LineChartData(
    extraLinesData: ExtraLinesData(
      horizontalLines: [
        HorizontalLine(
          y: threshold,
          color: Colors.orange,
          strokeWidth: 1.5,
          dashArray: [5, 5],
          label: HorizontalLineLabel(
            show: true,
            labelResolver: (_) => 'Irrigation Threshold',
          ),
        ),
      ],
    ),
    lineBarsData: [
      LineChartBarData(
        spots: readings
            .map((r) => FlSpot(r.timestamp.toDouble(), r.vwc))
            .toList(),
        isCurved: true,
        color: Colors.blue,
        dotData: const FlDotData(show: false),
      ),
    ],
  );
}

Alerts

Two alert types for soil sensors: by moisture threshold (below X% — needs irrigation) and by EC (above Y mS/cm — salinity risk). Delivery via FCM. Important: filter moisture alerts by time of day and weekday — if it just rained, "needs irrigation" alert is redundant. Backend should consider weather data or forecast.

Developing soil sensor monitoring application with LoRaWAN integration, moisture profiles and alerts: 3-5 weeks. Cost calculated individually.