Drought Early-warning Driven By Remote Sensing Data In Henan Province, China

Drought is the most common natural disaster that heavily affects agriculture all over theworld. In order to improve the crop production, decision-makers have to develop strategies toreduce the impact of drought. The efficient strategy depends on the accuracy of droughtmonitoring and early-warning. Although there have been lots of studies on drought monitoringduring the past decades, drought early-warning is still difficult. Most of drought early-warningresearches were carried out in one or several stations, which ignored the spatial distribution ofplants and the variety of drought sensitivity in various seasons at regional scale. In order todecrease the losses that come with drought as much as possible, it is necessary to combine remotesensing data and soil water budget model to monitor and predict the trend of drought. This studydescribes a drought early-warning model with a1-km resolution which is designed to simulate thesoil water balance for the winter wheat&maize planting areas in Henan Province on a daily basis,and5typical periods are chosen for model simulation and test. Main contents and conclusions areas follows:(1) On the basis of Zhang et al, the process of true dry determination is simplified, andTVDI is retrieved from MODIS data based on energy balance equation for surface soil moisturemonitor. In addition, soil moisture at deeper depth is estimated by fitting suitable regressionequations using soil moisture data of nearly20years at different depths recorded at variousstations.(2) Emphasis is put on the estipevoatepnottiraaln esvpairpaottiroann s p i r0mation of crop evapotranspiration. First, referenceatioisn iess ctiamlcauteladt edb yb yt mheu ltFiAplOyi nPg e n m0an-Monteith equation. Then, cropby crop coefficient Kc, a coefficientthat can be split into two factors describing separately the differences in evaporation and thetranspiration between both surfaces. Finally, water stress coefficient Ksis introduced to describethe actual evapotranspiration under water stress condition.(3) Drought early-warning model, developed by the ENVI/IDL language, is a simple soilwater budget model and a daily drought early-warning index is defined by the ratio of actual topotential evapotranspiration. Through monitor of soil moisture, estimation of crop evapotranspiration, precipitation interpolation with soil physical data, drought early-warningindex can be predicted if weather forecast data are available.(4) There is little drought early-warning during winter-wheat’s initial growth stage in winterif the antecedent soil moisture is sufficient because of the low water consumption. Atwinter-wheat’s development and mid-season stage in spring, drought early-warning might appearin the poor-precipitation area. And the extent of drought early-warning expanded in central andnorthern regions, and the intensity also enhanced at maize’s mid-season stage in summer. Theintensity of drought early-warning is low at clay and loam area because of strong capacity ofmoisture conservation, while drought early-warning usually appears in sandy area.(5) The model accuracy is various in different stages, during the winter-wheat’s initialgrowth stage, the accuracy is high with a correlation coefficient above0.5while the accuracydecreases at winter-wheat’s development and mid-season stage, finally the accuracy is very lowat maize’s mid-season stage. The model accuracy is also various in different regions. Theagreement between the simulation and accuracy is relatively higher in the south central, which isprobably relevant to the soil texture. Generally speaking, the soil moisture from simulation islower than that from test data and the accuracy is not very high, which is related to precipitation,soil moisture monitor, estimation of soil moisture at deeper depth from surface layer data, spatialinterpolation, remote sensing data quality, and disregard of irrigation, capillary rise and the otherfactors except soil moisture that affect crop growth. The model still needs to be improved.As mentioned above, so many problems still occur in this study, such as soil moistureretrieved from remote sensing data, estimation of soil moisture at deeper depth from surface layerdata, interpolation of meteorological data, evapotranspiration estimation and the simplifiedmethod to simulate soil water processes, which are in need for further study.