Abstract The US Army is required to be a good steward of the land per US Army regulation AR 200-1. Based on this regulation, Army installations need to manage lands, to reduce potential damage and impacts to water quality and habitat that may occur from training. Maneuver training does impact the vegetation and soil and this damage is directly related to soil moisture. Soil moisture is an important factor for understanding the potential for soil surface disturbance due to vehicle impacts and predicting soil resilience to vehicle traffic, however, producing accurate estimates of the spatial and temporal variation of soil moisture has historically been elusive. GeoWATCH, which stands for Geospatial Weather-Affected Terrain Conditions and Hazards (formerly DASSP), simulates soil moisture world-wide, at relatively small spatial and temporal scales. GeoWATCH uses a physics-based downscaling approach that uses weather-scale land surface model estimates of soil moisture and land surface water and energy fluxes, with high resolution geospatial data. GeoWATCH soil moisture outputs coupled with vehicle impact models, are anticipated to be useful for near-real-time estimation of ground disturbance, but will require ground validation. To validate GeoWATCH soil moisture estimates, we utilized Soil Climate Analysis Network (SCAN) gauge network soil moisture data from 127 sites across 34 states. Statistical analysis of the raw GeoWATCH output indicated the model performs statistically better in certain soil textures. Model bias is largest for sandy soils, whereas clayey soils were least biased. As a result, bias correction models were applied to the raw GeoWATCH simulated values using linear regression to predict correction factor (CF) values based on physical site characteristics. The bias correction models significantly improved the performance of the GeoWATCH soil moisture model in terms of average performance statistics and number of statistically cally unbiased sites. This process could easily be incorporated into GeoWATCH, allowing for a capability to rapidly estimate vehicle impacts and determine rehabilitation requirements by installation land managers.

中文翻译：

---

验证 GeoWATCH 土壤水分模型和建议的偏差校正方法

摘要 根据美国陆军条例 AR 200-1，美国陆军必须成为土地的好管家。根据这一规定，陆军设施需要管理土地，以减少训练可能对水质和栖息地造成的潜在损害和影响。机动训练确实会影响植被和土壤，而这种破坏与土壤湿度直接相关。土壤水分是了解车辆撞击对土壤表面扰动的可能性和预测土壤对车辆交通的弹性的重要因素，然而，准确估计土壤水分的空间和时间变化历来难以实现。GeoWATCH 代表受地理空间天气影响的地形条件和危害（以前称为 DASSP），在相对较小的空间和时间尺度上模拟世界范围内的土壤湿度。GeoWATCH 使用基于物理的降尺度方法，该方法使用天气尺度地表模型对土壤水分、地表水和能量通量的估计，以及高分辨率地理空间数据。GeoWATCH 土壤湿度输出与车辆碰撞模型相结合，预计可用于近实时估计地面扰动，但需要地面验证。为了验证 GeoWATCH 土壤水分估计，我们利用了来自 34 个州的 127 个站点的土壤气候分析网络 (SCAN) 测量网络土壤水分数据。原始 GeoWATCH 输出的统计分析表明，该模型在某些土壤质地中的统计性能更好。沙质土壤的模型偏差最大，而粘土的偏差最小。因此，使用线性回归将偏差校正模型应用于原始 GeoWATCH 模拟值，以根据物理站点特征预测校正因子 (CF) 值。偏差校正模型在平均性能统计数据和统计上无偏差站点的数量方面显着提高了 GeoWATCH 土壤水分模型的性能。这个过程可以很容易地合并到 GeoWATCH 中，从而能够快速估计车辆影响并确定安装土地管理人员的修复要求。