Skip to main content
ARS Home » Southeast Area » Oxford, Mississippi » National Sedimentation Laboratory » Watershed Physical Processes Research » Research » Publications at this Location » Publication #299206

Title: Curve number method response to historical climate variability and trends

item Bonta, James - Jim

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/3/2015
Publication Date: 3/1/2015
Publication URL:
Citation: Bonta, J.V. 2015. Curve number method response to historical climate variability and trends. Transactions of the ASABE. 58(2):319-334. doi:10.13031/trans.58.10431.

Interpretive Summary: Conservationists are often required to estimate watershed runoff amounts for a variety of conservation purposes. The technique most often used worldwide is called the “curve number” method (CN) which uses as underlying variables, runoff event flow volume (Q), the rainfall volume causing the runoff (P), and the rainfall volume that falls prior to the start of runoff (“initial abstraction”=Ia). The ratio of Ia to the total infiltrated precipitation during the runoff event (“m”), another underlying variable, can be either fixed (m=0.2) in common engineering application or “m” can be evaluated using observed data. The question arises whether the variables underlying the CN methodology may be changing due to observed and projected increases in air temperature and precipitation. The objectives of the study were to evaluate whether there are trends in CN and variables over time of observed climate change for two cases (assuming precipitation causes changes) – using the typical assumption of m=0.2 and allowing “m” to vary. Data from a small 0.66-ha meadow watershed with a 66-yr record of runoff were used to detect trends in these and other variables. Precipitation data showed that there were 7 climate periods of similar precipitation and that monthly precipitation is slightly but notably increasing in the months from Aug through Dec. Using the engineering approach (m=0.2) showed that climate change did not have an impact on CN. However, letting “m” vary resulted in a trend of increasing CN and m. The former case (m=0.2) would be more responsive to precipitation changes compared with temperature changes. Allowing “m” to vary incorporates precipitation and temperature changes (the latter through the evapotranspiration process), and can provide guidance for setting a new “m” value for standard usage. The results show that there is an effect of changing climate on the CN method, more so due to increasing temperatures than increasing precipitation but these effects are difficult to separate. They suggest that temperature affects the evapotranspiration process by assigning a greater fraction of precipitation losses to initial abstraction, caused by a greater depletion of soil-water storage prior to a runoff event. The results are important for assigning values of underlying variables for engineering design that incorporate a changing climate. Also, the method may need occasional reevaluation if precipitation and temperature continue to increase. The engineering, scientific, conservation, and regulatory communities will benefit from this study.

Technical Abstract: With the dependence on the curve number (CN) model by the engineering community, the question arises as to whether changes in climate may affect the performance of the CN algorithm which impacts estimates of runoff. A study was conducted to determine the effects of “climate period” (period of uniform monthly precipitation accumulation rates) on CN algorithm components using data available at the USDA-ARS North Appalachian Experimental Watershed (NAEW) near Coshocton, Ohio - a 0.66-ha small experimental watershed (WS130) has been in same land management practice (hay production) for ~74 yr beginning about 1937 (monitored ~ 89% of time). Changes in precipitation mass-curve slopes were used to identify seven climate periods of uniform precipitation. Trends in event-based CN component variables evaluated were Q (event runoff), P (event causal precipitation), Q/P, Ia (initial abstraction due to infiltration, interception, etc. in CN method), Pe (effective precipitation, P-Ia), Ia/P, CNe (event-based CN), and m (=Ia/S, where S=potential maximum retention). Due to the wide variability of the data, trends due to climate period were identified for different ranges of the data. These ranges were chosen under different assumptions related to the performance of the CN procedure (e.g., larger P events are typically used for determining CN from data so larger P was used to identify CN ranges). Statistical significance of the trend of medians was determined using rank correlation across periods. There was a weak but notable increase in precipitation since 1937, primarily due to increasing monthly precipitation trends in each month from Aug through Dec. There was no trend across climate periods of CN using the assumption of Ia=0.2S. Median ratios of Q, P, Q/P, Ia Pe, and Ia/P did not show statistically significant trends with climate period, but nearly all showed positive correlation. However, Pe consistently showed negative correlation, although not significant. This suggests that Ia tended to increase more than P which tends to result in smaller Pe, causing the difference between P and Ia to decrease with time, to support an increase in m and CNe. Climate change appears to increase the m parameter (0.0045 m/yr for Pe>25 mm) and event CNe (0.29 CN/yr for P>25 mm). Larger CNe may be due to increasing m and to increasing Ia (Ia is not increasing significantly). A notable increase in runoff accumulation rate during the 2005-2011 period (with corresponding larger precipitation rate) did not affect the significant trends of CN, m, and CNe found. Precipitation likely has had less impact on CN methodology than air temperature through the evapotranspiration process, however, it is difficult to separate the effects of precipitation and temperature on the CN algorithm components (the present study focused on precipitation). The results suggest that if climate change continues, an occasional reevaluation of the effects of climate change on CN model components may be necessary. While the present study focused on precipitation, the results suggest air temperature may be a more important factor affecting CN components, but both appear important.