Data from: Temporal variability is a major source of uncertainty in soil carbon measurements

This dataset contains five sets of monthly samples extending up to three years each that were collected from field experiments at four locations representative of dryland farming on silt-loam soils in Oregon and Washington, USA. 
Two experiments conducted at the Columbia Plateau Conservation Research Center near Adams, OR, USA (45.767°, -118.563°, 401 mm yr‑1 precipitation) involved monthly sample collection for three years. The first experiment (“Adams-tillage”) examined annual winter wheat and included three tillage treatments: 1) no-till, 2) surface residue incorporated with tillage, and 3) surface residue removed and then replaced after tillage (Wuest, 2014). The soil was a Walla Walla silt loam (coarse-silty, mixed, superactive, mesic Typic Haploxeroll) on a level, uniform field. Plots were 3.6 by 53 m replicated in four randomized, complete blocks, for a total of 12 experimental units. Each month from 2001 to 2005 three 50-mm diameter soil cores from each plot were combined, one from the crop row, one from the center between crop rows, the third between the crop row and the center sample. Depths were six 5-cm increments to 30-cm depth. Each month’s sample was 30 cm further down a crop row from the previous month’s sample. Samples were dried at 40° C starting within an hour of collection. Soil organic carbon was measured with dry combustion analysis. Surface soil pH averaged 5.3 and lab tests indicate that carbonates were not present in the surface soil. Since the soil samples in this study were taken and chemically analyzed by length increments from the soil surface, the analytical values based on equivalent dry soil mass were calculated by interpolating between the cumulative dry mass of each incremental sample. Weather data were obtained from a station located 400 m from the plots. 
The second experiment near Adams (“Adams-residue”) examined residue management in annual winter wheat from May 2016 to Aug 2019, with three treatments: 1) no wheat residue removed, 2) all chaff and straw removed, and 3) removal of chaff only. Plots were 6 by 60 m, with four replicates in randomized complete blocks for a total of 12 plots. The field was level with a slight SOC gradient from block to block. The soil type was the same as the above Adams-tillage site. Each plot was sampled 29 times, monthly except for a few winter months due to either snow-covered or wet conditions. Three 25-mm diameter intact cores were taken from the center of each plot within 30 cm of the previous month’s sample, wrapped in paper to maintain the core intact, dried, and the top 250 kg m‑2 of soil from each core removed for cleaning (removing visible plant matter) and analysis. After weighing the mass-depths, samples from the three cores were pooled for dry combustion analysis. This site was 300 m from the weather station. 
A third experiment involved monthly samples in three sets of long-term plots near the townsites of Echo, OR, Moro, OR, and Ritzville, WA (Wuest, Schillinger, et al., 2023). These were two- or three-year-rotation studies including a fallow year. 
At the Echo site (45.7303, -119.0569, 265 mm yr-1 precipitation) the soil was Ritzville silt loam (coarse-silty, mixed, superactive, mesic Calcidic Haploxerolls) with 5.7 pH and containing <15% sand, ≥75% silt and 5 to 10% clay (Soil-Survey-Staff, 2022). Treatments were a comparison of no-till versus tilled winter wheat—fallow rotations. The two tillage treatments and two rotation entry points were randomized in four complete blocks for a total of 16 plots. Plot size was 7 m by 55 m. A weather station was located within 200 m of the plots. 
At Moro (45.4833, -120.7184, 269 mm yr⁻1 precipitation) the soil was Walla Walla silt loam (coarse, silty, mixed, superactive, mesic Typic Haploxeroll) with 5.7 pH. Soil texture was 14% fine sand, 72% silt, and 14% clay (Soil-Survey-Staff, 2022). Treatments examined winter wheat--fallow comparing no-till and minimum till, with both rotation phases represented each year. These four treatment plots were replicated in each of three randomized complete blocks for a total of 12 plots sampled monthly. Two of the blocks were on a backslope and the third on a less productive shoulder slope. The weather station was within 400 m of the plots. 
At Ritzville (47.1437, -118.4693, 296 mm yr-1 precipitation) the soil was Ritzville silt loam (coarse-silty, mixed, superactive, mesic Calcidic Haploxerolls) >2 m deep with no rocks or restrictive layers and slopes were <1%. Soil texture is 30% fine sand, 59% silt, and 11% clay (Soil-Survey-Staff, 2022). Soil pH in the surface was 6.5. Treatments compared two three-year rotations: spring wheat, fallow, and winter triticale using no-till, versus spring wheat, fallow, and winter wheat using conservation tillage. Experimental design was a randomized complete block with four replications. Individual plot size was 9 by 150 m. All rotation phases were present every year (total = 24 plots). Weather data were taken from a station near the city of Ritzville, 7 km from the plots. 
The Echo site was terminated after 18 samples in 19 months. The Moro and Ritzville sites were sampled 29 and 31 times over 33 months, missing some winter months due to snow-cover or excessive wetness. At all three sites a single 25-mm core was taken from each plot within a few meters of the same location each month. The intact core was wrapped in paper, dried, and the top 250 kg m‑2 (approximately 20 cm) removed for analysis. This sample depth is deeper than the influence of current tillage and has proven to contain the major changes in SOC in these cropping systems. As above, dry combustion analysis was used to determine SOC. 
At all four research sites lab measurements have demonstrated no significant inorganic carbon in the surface 30 cm, so total C was considered equal to SOC. Samples were processed monthly by the same lab using the same standards and procedures throughout the experiment. Our objective was to examine and characterize SOC temporal variability measured at five sites where high-frequency soil sampling campaigns were carried out in a semiarid dryland agriculture environment.
<br>
References
Soil-Survey-Staff. 2022. Natural Resources Conservation Service, United States Department of Agriculture. Official Soil Series Descriptions. Accessed 3/14/2023. Available online.
Wuest, S. 2014. Seasonal variation in soil organic carbon. Soil Sci. Soc. Am. J. 78: 1442-1447. doi:10.2136/sssaj2013.10.0447.
Wuest, S.B., W.F. Schillinger and S. Machado. 2023. Variation in soil organic carbon over time in no-till versus minimum tillage dryland wheat-fallow. Soil Tillage Res. 229: 105677. doi:https://doi.org/10.1016/j.still.2023.105677.

本数据集包含5组月度采样样本，每组样本时长最长可达3年，采集自美国俄勒冈州与华盛顿州境内4个代表粉壤土旱地农业的试验站点的田间试验。

两项试验设于美国俄勒冈州亚当斯附近的哥伦比亚高原保护研究中心（坐标45.767°，-118.563°，年降水量401 mm），均开展了为期3年的月度采样。第一项试验为"亚当斯-耕作"，研究对象为一年生冬小麦，设置3种耕作处理：1）免耕；2）地表残体翻压还田；3）地表残体移除后经耕作再覆于地表（Wuest, 2014）。试验田块地势平坦均匀，土壤为沃拉沃拉粉壤土（coarse-silty, mixed, superactive, mesic Typic Haploxeroll），试验小区尺寸为3.6×53 m，采用4个随机完全区组重复，总计12个试验单元。2001年至2005年的每月采样中，从每个小区采集3个直径50 mm的土芯，分别取自作物行、作物行间距中心以及作物行与中心采样点之间的位置，采样深度为6层5 cm增量，总深度达30 cm。每月采样点相较于上月沿作物行下移30 cm。样本于采集后1小时内开始在40℃条件下烘干。采用干烧法（dry combustion analysis）测定土壤有机碳（Soil Organic Carbon, SOC）。表层土壤pH均值为5.3，实验室检测结果表明表层土壤不含碳酸盐。由于本研究的土壤样本按土层增量自地表向下采集，因此基于等效干土质量的分析值通过插值各增量样本的累积干质量计算得到。气象数据取自距试验小区400 m的气象站。

第二项试验同样设于亚当斯附近，命名为"亚当斯-残体管理"，于2016年5月至2019年8月期间研究一年生冬小麦的残体管理措施，设置3种处理：1）保留全部小麦残体；2）移除全部颖壳与秸秆；3）仅移除颖壳。试验小区尺寸为6×60 m，采用4个随机完全区组重复，总计12个试验小区。试验田块地势平坦，区组间存在微弱的土壤有机碳梯度。土壤类型与前述"亚当斯-耕作"站点一致。每月对每个小区采样29次，少数冬季月份因积雪或土壤过湿暂停采样。从每个小区中心、距上月采样点30 cm范围内采集3个直径25 mm的原状土芯，用纸包裹以保持土芯完整，烘干后移除每个土芯顶部250 kg m⁻²的土层用于清杂（去除可见植物残体）与分析。称量各土层质量后，将3个土芯的样本混合后采用干烧法测定SOC。该站点距气象站300 m。

第三项试验设于俄勒冈州埃科、莫罗以及华盛顿州里茨维尔的长期试验田，分为3组月度采样序列（Wuest, Schillinger, et al., 2023），研究内容为包含休耕季的2年或3年轮作系统。

埃科站点（坐标45.7303, -119.0569，年降水量265 mm）的土壤为里茨维尔粉壤土（coarse-silty, mixed, superactive, mesic Calcidic Haploxeroll），pH为5.7，砂粒含量<15%，粉粒含量≥75%，黏粒含量5%~10%（Soil-Survey-Staff, 2022）。试验处理为免耕与耕作冬小麦-休耕轮作的对比。2种耕作处理与2种轮作起始点采用4个随机完全区组设计，总计16个试验小区，小区尺寸为7×55 m。试验小区附近200 m范围内设有气象站。

莫罗站点（坐标45.4833, -120.7184，年降水量269 mm）的土壤为沃拉沃拉粉壤土（coarse, silty, mixed, superactive, mesic Typic Haploxeroll），pH为5.7。土壤质地为细砂14%、粉粒72%、黏粒14%（Soil-Survey-Staff, 2022）。试验处理为对比免耕与少耕的冬小麦-休耕轮作，每年覆盖全部轮作阶段。该4种处理小区采用3个随机完全区组重复，总计12个试验小区，按月采样。其中2个区组设于背坡，第3个区组设于生产力较低的肩坡。试验小区附近400 m范围内设有气象站。

里茨维尔站点（坐标47.1437, -118.4693，年降水量296 mm）的土壤为厚度>2 m的里茨维尔粉壤土（coarse-silty, mixed, superactive, mesic Calcidic Haploxeroll），无石块或阻水层，坡度<1%。土壤质地为细砂30%、粉粒59%、黏粒11%（Soil-Survey-Staff, 2022）。表层土壤pH为6.5。试验处理对比2种3年轮作系统：免耕模式下的春小麦-休耕-冬小黑麦，与保护性耕作模式下的春小麦-休耕-冬小麦。试验设计为4次重复的随机完全区组，单个小区尺寸为9×150 m，每年覆盖全部轮作阶段（总计24个小区）。气象数据取自距试验小区7 km的里茨维尔市附近气象站。

埃科站点在19个月内完成18次采样后终止试验。莫罗与里茨维尔站点分别在33个月内完成29次与31次采样，少数冬季月份因积雪或土壤过湿暂停采样。在上述3个站点中，每月均从每个小区内与上月采样点相距数米的同一位置采集1个直径25 mm的原状土芯，用纸包裹后烘干，移除顶部250 kg m⁻²（约20 cm）土层用于分析。该采样深度超出当前耕作影响层，且已被证实可反映这些种植系统中SOC的主要变化。如前所述，采用干烧法测定SOC。

在全部4个试验站点的实验室检测均表明，表层30 cm土壤内无显著无机碳，因此总碳量可视为SOC。整个试验期间，所有样本均由同一家实验室采用统一标准与流程按月处理。本数据集旨在研究并表征半干旱旱地农业环境下，通过高频土壤采样获得的SOC时间变异性。

## 参考文献
美国农业部自然资源保护局土壤调查团队. 2022. 官方土壤系列描述. 2023年3月14日访问. 在线可获取.
Wuest, S. 2014. 土壤有机碳的季节变化. 美国土壤学会会刊 78: 1442-1447. doi:10.2136/sssaj2013.10.0447.
Wuest, S.B., W.F. Schillinger 与 S. Machado. 2023. 免耕与少耕旱地小麦-休耕系统中土壤有机碳的时间变化. 土壤耕作研究 229: 105677. doi:https://doi.org/10.1016/j.still.2023.105677.