Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/12/2015
Publication Date: 4/15/2015
Citation: Endale, D.M., Schomberg, H.H., Fisher, D.S., Jenkins, M. 2015. Curve numbers from conventional and no-till cropping: A 39-yr dataset from a small Georgia Piedmont watershed. Transactions of the ASABE. 58(2):379-391.
Interpretive Summary: Estimating the amount of water that runs off agricultural and urban lands during and after rainfall events is complex because runoff is influenced by many factors that are not necessarily all known or understood. Researchers develop mathematical models, based on knowledge available at the time, to facilitate estimates of the impact of agricultural (or urban) activities on water quantity and/or quality. Models must continuously be tested and updated as new knowledge becomes available so that they can be used with confidence. Researchers at the USDA-ARS research facility near Watkinsville, GA, collected rainfall and runoff data from 1972 to 2010 from a 6.5-acre watershed managed as a crop-field to test a popular USDA hydrologic model (curve number model) that has been in use since the 1950s. For an initial 2.5 year period when tillage was performed to produce crops in a typical manner by turning the soil over with a plow or disk (conventional tillage), the model performed as expected. The model was, however, not well adjusted to periods when tillage on the field was avoided (no-till). The average expected curve number value (an important coefficient of the model) for the field, taken from the standard manual developed from tilled sites, overestimated the average runoff for the no-till period by 142%. This average value had to be reduced by 13 to match the average runoff measured for the no-till period. The results strongly suggest the need to develop new standard curve numbers for alternate tillage practices. The latest estimate by USDA put the area under no-till cropping in the USA close to 93 million acres. These results demonstrate the critical value of long-term data from field operations for improving models that are used to evaluate the potential impacts of agricultural policies.
Technical Abstract: Since its inception in the 1950s, acceptance, use and adaptation of the curve number (CN) method for estimating direct runoff from a rainfall event has increased worldwide receiving critical reviews. There have been calls for development of locally defined CNs to address concerns with regional and seasonal variations. We derived CNs from rainfall-runoff data gathered from a 2.7 ha watershed (P1) from 1972-2010. The watershed was initially managed under conventional tillage for 2.5 yr followed by double-cropped continuous no-till rotations. Summer crops included soybean (Glycine max L. Merr), sorghum (Sorghum vulgare Pers.), millet (Pennisetum glaucum), cotton (Gossypium hirsutum L.) and corn (Zea mays L.), with barley (Hordeum vulgare L.), wheat (Triticum aestivum L.), crimson clover (Trifolium incarnatum L.) and rye (Secale cereal L.) as cover crops. On average, almost 19% of the rainfall exited the watershed as runoff during the conventional tillage phase (median 13%). The mean and median derived CNs were 82 and 85, respectively. Least square fitting of CN versus rainfall produced asymptotic CN of 81 (R2 = 0.61). Curve numbers from standard tables ranged 75-86 for these conditions. In contrast, during the no-till phase, the mean and median runoff was 7% and 0.6%, respectively, of the rainfall. The mean and median derived CNs were the same (62) and the least square fitted asymptotic CN was 58 (R2 = 0.78). The area-weighted average CN for P1 for the no-till phase from standard tables was ~72. Mean runoff for the no-till phase estimated using CN = 72 was greater by 142% compared with the measured mean runoff. Variations of asymptotic CN were observed for the no-till phase by season and crop period (45-67). The derived median value of the initial abstraction ratio ' for the no-till phase was 0.03 and supports recent literature suggesting that the CN method would be improved by lowering the standard ' value from 0.2 to 0.05. Overall, the results strongly support the hypothesis that CNs for no-till cropping systems, and perhaps other conservation tillage systems, should be smaller (perhaps up to 10 or more CN units) in some regions than the CNs provided in standard handbook tables. This has serious implications for hydrologic modeling for water quantity and quality where the use of the CN method is ubiquitous. Availability of long-term data sets that take into account variability in weather and management is essential for improving accuracy of many predictive models developed from limited data. The results of this study strongly support implementation of no-till as a best management practices in crop production systems to reduce runoff and by implication improve water quality and productivity.