After long-term operation, a large-scale artificial rainfall simulation system can exhibit low automation, insufficient stability, and limited flexibility in rainfall intensity regulation. To address these deficiencies, a comprehensive optimization and performance evaluation was conducted to improve system reliability, controllability, and experimental applicability.

Three major subsystems were upgraded: (1) Power and control system: a high-precision stepper motor was installed, enabling fully continuous and automated regulation of rainfall intensity from 0 mm/h to 200 mm/h with 1 mm/h resolution. (2) Water supply system: a variable-frequency booster pump was integrated to maintain stable water pressure, and a mechanical float valve was added to automate water level control in the storage tank. (3) Hoisting system: manually operated pulleys were replaced with remote-controlled electric winches, and all power and water lines were consolidated within a flexible energy chain (cable carrier) to prevent abrasion and eliminate on-floor obstructions. An intelligent software module was also developed to enforce digital logging of user identity, experimental purpose, and funding source and to provide automated sampling reminders. Performance was evaluated through quantitative tests of rainfall intensity stability, spatial uniformity, raindrop size and velocity distributions (using a laser disdrometer), and operational efficiency.

Key improvements include the following: (1) The upgraded system yielded substantial gains in control precision, operational stability, and rainfall uniformity. By replacing the conventional electromagnetic clutch with high-precision stepper motors, rainfall intensity can be continuously and automatically regulated between 0 to 200 mm/h in 1 mm/h increments. The retrofitted water supply system, incorporating a variable-frequency booster pump and a mechanical float valve, ensures stable water pressure and steady delivery, guaranteeing reliable performance even under prolonged, high-intensity operation. Spatial uniformity coefficients measured 88.1%–89.7% at 2.5 m height and increased markedly to 93.5%–94.9% at 4.0 m, indicating excellent areal distribution. (2) Microphysical characterization revealed that raindrop mean diameters ranged from 0.86 to 1.03 mm, with volume median diameters (D₅₀) of 1.75–2.5 mm and maximum diameters of 4.5–5.5 mm across the tested intensities. In terms of number concentration, simulated raindrops were predominantly distributed between 0.125 and 5.5 mm, with most falling below 1.0 mm; only a negligible fraction exceeded this threshold. The joint frequency distribution of drop size and fall velocity showed the highest concentration in the 0.125–1.0 mm diameter and 1.0–4.0 m/s velocity ranges. Notably, the simulated drop size–velocity coupling matched observed torrential rain events particularly closely.

The optimized system reproduces the raindrop size distribution and combined diameter–velocity characteristics of natural rainfall, with the closest resemblance to storm-type precipitation. It therefore provides a reliable, controllable, and repeatable platform for experimental studies in soil erosion, hydrology, and allied disciplines. Nevertheless, given China’s vast territory and diverse climatic regimes, the microphysical properties of natural rainfall—including drop size spectra and velocity distributions—vary considerably across regions and synoptic weather types. While the present system demonstrates good correspondence with certain natural rainfall patterns, it does not fully replicate the entire spectrum of rainfall varieties encountered across the country. This limitation should be acknowledged when extrapolating experimental results to regions or rainfall types not directly represented by the current calibration database.

Particle size data for black soil, measured using a Malvern laser particle size analyzer, were compared with those obtained using a Beckman laser particle size analyzer, and both results were benchmarked against data obtained using the pipette method. The study aimed to establish formulas for interconverting laser diffraction data to those from the pipette method, and for interconverting data obtained using the two laser particle size analyzers.

A total of 147 black soil samples were tested. Among them, 33 surface soil samples collected from different regions of the black soil area were analyzed using the laser particle size analyzers and the pipette method to establish formulas for interconversion of the data from the laser diffraction and pipette methods and for interconversion of the data from the two laser particle size analyzers. Another 114 black soil samples collected from the Hebei small watershed were only tested using the two laser particle size analyzers. The effectiveness of the formulas for interconversion of the data measured using the two particle size analyzers was evaluated. The data from the two laser particle size analyzers for the 114 samples from the small watershed were converted to align with those from the pipette method using the aforementioned conversion formulas for soil texture classification.

For the 33 samples from the black soil area, the clay content (< 0.002 mm) measured by the pipette method was significantly higher than that measured by the laser diffraction method, but there was a significant correlation between the two methods (p < 0.01). The sand content measured by the pipette method was higher than that measured using the Beckman particle size analyzer but lower than that measured using the Malvern particle size analyzer. However, corresponding conversion formulas could also be derived for each method (p < 0.01). For the black soil samples from the small watershed evaluated using the two laser particle size analyzers, the difference in the clay content was relatively small, with an average absolute error of 2.5%. However, the absolute average errors in the silt and sand content were > 10%. After conversion, the difference decreased, with an average absolute error of 1.8% for the clay content, 3.9% and 6.3% for the silt content in the International System and the American System, respectively, and 5.1% and 7.4% for the sand content in the International System and the American System, respectively. After converting the data from the two laser particle size analyzers to align with those from the pipette method, the values could be consistently used for soil texture classification of the small watershed samples, with a consistency rate of 93.0% for the International System. However, for texture classification using the American System, the consistency rate was only 58.8%.

Therefore, there are significant differences in the raw values obtained with the two particle size analyzers, making the data unsuitable for direct comparison. However, there is a highly significant correlation between the two (p < 0.01). The conversion formulas established using samples from black soil regions can considerably reduce the differences, providing a better basis for soil texture classification in the international system for black soil.