David R. Legates

Evaluating the use of 'goodness-of-fit' measures in hydrologic and hydroclimatic model validation

Correlation and correlation-based measures (e.g., the coefficient of determination) have been widely used to evaluate the “goodness-of-fit” of hydrologic and hydroclimatic models. These measures are oversensitive to extreme values (outliers) and are insensitive to additive and proportional differences between model predictions and observations. Because of these limitations, correlation-based measures can indicate that a model is a good predictor, even when it is not. In this paper, useful alternative goodness-of-fit or relative error measures (including the coefficient of efficiency and the index of agreement) that overcome many of the limitations of correlation-based measures are discussed. Modifications to these statistics to aid in interpretation are presented. It is concluded that correlation and correlation-based measures should not be used to assess the goodness-of-fit of a hydrologic or hydroclimatic model and that additional evaluation measures (such as summary statistics and absolute error measures) should supplement model evaluation tools.

Mean seasonal and spatial variability in global surface air temperature

SummaryUsing terrestrial observations of shelter-height air temperature and shipboard measurements, a global climatology of mean monthly surface air temperature has been compiled. Data were obtained from ten sources, screened for coding errors, and redundant station records were removed. The combined data base consists of 17 986 independent terrestrial station records and 6 955 oceanic grid-point records. These data were then interpolated to a 0.5° of latitude by 0.5° of longitude lattice using a spherically-based interpolation algorithm. Spatial distributions of the annual mean and intra-annual variance are presented along with a harmonic decomposition of the intra-annual variance.

The Accuracy of United States Precipitation Data

Abstract Precipitation measurements in the United States (as well as all other countries) are adversely affected by the gauge undercatch bias of point precipitation measurements. When these measurements are used to obtain areas averages, particularly in mountainous terrain, additional biases may be introduced because most stations are at lower elevations in exposed sites. Gauge measurements tend to be underestimates of the true precipitation, largely because of wind-induced turbulence at the gauge orifice and wetting losses on the internal walls of the gauge. These are not trivial as monthly estimates of this bias often vary from 5% to 40%. Biases are larger in winter than in summer and increase to the north in the United States due largely to the deleterious effect of the wind on snowfall. Simple spatial averaging of data from existing networks does not provide an accurate evaluation of the area-mean precipitation over mountainous terrain (e.g., over much of the western United States) since most stations...

Crop identification using harmonic analysis of time-series AVHRR NDVI data

Harmonic analysis of a time series of National Oceanic and Atmospheric Administration (NOAA) advanced very high resolution radiometer normalized difference vegetation index (NDVI) data was used to develop an innovative technique for crop type identification based on temporal changes in NDVI values. Different crops (corn, soybeans, alfalfa) exhibit distinctive seasonal patterns of NDVI variation that have strong periodic characteristics. Harmonic analysis, or Fourier analysis, decomposes a time-dependent periodic phenomenon into a series of constituent sinusoidal functions, or terms, each defined by a unique amplitude and phase value. Amplitude and phase angle images were produced by analysis of the time-series NDVI data and used within a discriminant analysis to develop a methodology for crop type identification. For crops that have a single distinct growing season and period of peak greenness, such as corn, the majority of the variance was captured by the first and additive terms, while winter wheat exhibited a bimodal NDVI periodicity with the majority of the variance accounted for by the second harmonic term.

Soil moisture: A central and unifying theme in physical geography

Soil moisture is a critical component of the earth system and plays an integrative role among the various subfields of physical geography. This paper highlights not just how soil moisture affects atmospheric, geomorphic, hydrologic, and biologic processes but that it lies at the intersection of these areas of scientific inquiry. Soil moisture impacts earth surface processes in such a way that it creates an obvious synergistic relationship among the various subfields of physical geography. The dispersive and cohesive properties of soil moisture also make it an important variable in regional and microclimatic analyses, landscape denudation and change through weathering, runoff generation and partitioning, mass wasting, and sediment transport. Thus, this paper serves as a call to use research in soil moisture as an integrative and unifying theme in physical geography.

Impacts of land use/land cover change on climate and future research priorities.

Several recommendations have been proposed for detecting land use and land cover change (LULCC) on the environment from, observed climatic records and to modeling to improve its understanding and its impacts on climate. Researchers need to detect LULCCs accurately at appropriate scales within a specified time period to better understand their impacts on climate and provide improved estimates of future climate. The US Climate Reference Network (USCRN) can be helpful in monitoring impacts of LULCC on near-surface atmospheric conditions, including temperature. The USCRN measures temperature, precipitation, solar radiation, and ground or skin temperature. It is recommended that the National Climatic Data Center (NCDC) and other climate monitoring agencies develop plans and seek funds to address any monitoring biases that are identified and for which detailed analyses have not been completed.

Reconstructing Climatic and Environmental Changes of the Past 1000 Years: A Reappraisal

The 1000-year climatic and environmental history of the Earth contained in various proxy records is examined. As indicators, the proxies duly represent or record aspects of local climate. Questions on the relevance and validity of the locality paradigm for climatological research become sharper as studies of climatic changes on timescales of 50–100 years or longer are pursued. This is because thermal and dynamical constraints imposed by local geography become increasingly important as the air-sea-land interaction and coupling timescales increase. Because the nature of the various proxy climate indicators are so different, the results cannot be combined into a simple hemispheric or global quantitative composite. However, considered as an ensemble of individual observations, an assemblage of the local representations of climate establishes the reality of both the Little Ice Age and the Medieval Warm Period as climatic anomalies with world-wide imprints, extending earlier results by Bryson et al. (1963), Lamb (1965), and numerous other research efforts. Furthermore, these individual proxies are used to determine whether the 20th century is the warmest century of the 2nd Millennium at a variety of globally dispersed locations. Many records reveal that the 20th century is likely not the warmest nor a uniquely extreme climatic period of the last millennium, although it is clear that human activity has significantly impacted some local environments.

Documenting and Detecting Long-Term Precipitation Trends: Where we are and What should be Done

A brief review of problems and achievements in documenting precipitation changes during the period of instrumental measurements is presented. Concern is expressed that without appropriate studies in the coming period of a new generation of precipitation measurements, technological progress in instrumentation may adversely and inadvertently affect our capability for monitoring and detecting future changes in terrestrial precipitation. At the same time, only a new generation of instrumentation will be capable of resolving the problems of monitoring precipitation over oceans.

A Re-Evaluation of the Average Annual Global Water Balance

Legates and Mather (1992) used the Thornthwaite/Mather water balance and mean monthly air temperature and precipitation data on a 0.5° of latitude by 0.5° of longitude grid to refine spatial estimates of annual evapotranspiration and runoff on a global scale. Because Legates and Mather used the Thornthwaite potential evapotranspiration equation, they were unable to apply bias-adjustments to their precipitation estimates because, as Legates and Mather discovered, the Thornthwaite equation implicitly accounts for gage measurement biases by underestimating potential evapotranspiration. Here, we re-evaluate the results of Legates and Mather using bias-adjusted precipitation estimates and an alternative method of estimating potential evapotranspiration (i.e., the Hamon, 1963, method). These new results are compared with those from previous studies and are considered to be an improvement over the Legates and Mather values, as they are based on bias-adjusted precipitation estimates.

Real-Time Calibration of Radar Precipitation Estimates

One of the main concerns with precipitation measurements is that gage networks are almost always too sparse to provide an adequate spatial coverage of storm-scale precipitation. Gage measurements are representative only at the measurement site and are biased underestimates of the actual precipitation, mainly as a result of the effect of wind on the gage. Consequently, storm-scale, real-time assessments using only gage-measured precipitation are frequently inadequate. With the advent of the WSR-88D (formerly NEXRAD) weather radars, precipitation estimates at higher spatial resolutions (4 km by 4 km) are now available in real time. These radars use the reflectivity of S-band (10 cm) microwaves to provide an estimate of precipitation. Unfortunately, reflectivity is a function of the surface area of the raindrops and not their volume. As a result of this and other sources of error, radar precipitation estimates using fixed reflectivity-to-rainfall relationships are subject to substantial biases. To provide better high-resolution precipitation estimates, a gage-radar precipitation compositing procedure has been developed to enhance real-time precipitation assessments. Radar estimates provide the spatial ‘footprint’ of the storm while gage data are used to enhance accuracy. This procedure calibrates each radar separately (since biases usually vary by radar), provides a composite mosaic of multiple radars for regions that lie under more than one radar umbrella, and determines an estimate of the uncertainty of the calibration.