Abstract: Abstract: Soil moisture content and matric potential are two important parameters for determining soil water characteristic curve (SWCC), which has significant differences for different types of soils. The common method to determine SWCC under field condition is using two separated sensors to measure moisture content and matric potential of the tested soils. However, measurement error may occur due to soil spatial heterogeneity around two sensors. In this study, a novel composite sensor was designed to simultaneously measure soil moisture content, matric potential, and temperature. The composite sensor consists of a 32-bit MCU main control chip, relay switching module, dielectric measurement module for determining soil moisture content and matric potential, and temperature measurement module. The main MCU control chip measures soil moisture content, matric potential, and temperature through controlling the relay switching, and performs temperature correction as well. The performances of the composite sensor were tested for soil moisture content and soil matric potential measurements including monotonicity, calibration, volume of sensitivity, response time, etc. To test the monotonicity of the dielectric mode, the length of high frequency transmission line was gradually increased and the sensor output was recorded when the probes of soil moisture content and soil water potential were moistened from dry to saturation. The three modes, soil moisture content, soil matric potential and temperature were calibrated respectively using forest soil, farmland loam and sandy soil. To determine the volume of sensitivity of the soil moisture content mode, the probe of soil moisture content was installed in a soil sample. The soil sample was gradually removed layer by layer and meanwhile the output of the soil moisture content was recorded. When the output started to decrease, the radius of volume of sensitivity was determined by measuring the thickness of residual soil layer from surface to the probe. To determine the response time of soil moisture content and soil matric potential modes, the composite sensor probe was firstly dried and then put into a saturated soil sample. Meanwhile, the output of the soil moisture content and soil matric potential was recorded as time elapse. When the recorded output of each mode did not remarkedly change, the elapsed time was the response time. Experiments for sensor observation were conducted using forest soil, farmland loam and sandy soil under laboratory condition at 25 ℃. The experimental results were compared with those measured using commercial instruments. The results showed that the composite sensor had a good monotonicity when measuring soil moisture content and matric potential in the range from dry to near saturation. After calibration, the composite sensor can determine soil moisture content, matric potential and temperature with high accuracies, achieving the determination coefficients larger than 0.98. The measurement ranges of volumetric soil water content and soil matric potential were 0-40% and -1 500--15 kPa, respectively, which is feasible for most soil textures. The response time of the soil moisture content and soil matric potential was 450 ms and 150 s, respectively. The temperature coefficient of the circuit board of the composite sensor was yielded to correct the temperature induced output variation. The SWCCs of forest soil, farmland loam and sandy soil were determined using both the commercial instruments and the composite sensor during wetting and drying processes. The hysteresis between soil drying and wetting processes was also observed as expected. The Pearson correlation coefficients between the SWCCs measured by the two methods were larger than 0.96 during wetting and drying processes, i.e., 0.962 and 0.983 for the forest soil, 0.993 and 0.995 for the farmland loam, and 0.979 and 0.998 for the sandy soil. The composite sensor developed in this study can accurately determine soil water characteristic curves of different types of soils.